xref: /freebsd/sys/net/if_bridge.c (revision f5b7695d2d5abd735064870ad43f4b9c723940c1)
1 /*	$NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $	*/
2 
3 /*-
4  * SPDX-License-Identifier: BSD-4-Clause
5  *
6  * Copyright 2001 Wasabi Systems, Inc.
7  * All rights reserved.
8  *
9  * Written by Jason R. Thorpe for Wasabi Systems, Inc.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
14  * 1. Redistributions of source code must retain the above copyright
15  *    notice, this list of conditions and the following disclaimer.
16  * 2. Redistributions in binary form must reproduce the above copyright
17  *    notice, this list of conditions and the following disclaimer in the
18  *    documentation and/or other materials provided with the distribution.
19  * 3. All advertising materials mentioning features or use of this software
20  *    must display the following acknowledgement:
21  *	This product includes software developed for the NetBSD Project by
22  *	Wasabi Systems, Inc.
23  * 4. The name of Wasabi Systems, Inc. may not be used to endorse
24  *    or promote products derived from this software without specific prior
25  *    written permission.
26  *
27  * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
28  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
29  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
30  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL WASABI SYSTEMS, INC
31  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37  * POSSIBILITY OF SUCH DAMAGE.
38  */
39 
40 /*
41  * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
42  * All rights reserved.
43  *
44  * Redistribution and use in source and binary forms, with or without
45  * modification, are permitted provided that the following conditions
46  * are met:
47  * 1. Redistributions of source code must retain the above copyright
48  *    notice, this list of conditions and the following disclaimer.
49  * 2. Redistributions in binary form must reproduce the above copyright
50  *    notice, this list of conditions and the following disclaimer in the
51  *    documentation and/or other materials provided with the distribution.
52  *
53  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
54  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
55  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
56  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
57  * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
58  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
59  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
61  * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
62  * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
63  * POSSIBILITY OF SUCH DAMAGE.
64  *
65  * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
66  */
67 
68 /*
69  * Network interface bridge support.
70  *
71  * TODO:
72  *
73  *	- Currently only supports Ethernet-like interfaces (Ethernet,
74  *	  802.11, VLANs on Ethernet, etc.)  Figure out a nice way
75  *	  to bridge other types of interfaces (maybe consider
76  *	  heterogeneous bridges).
77  */
78 
79 #include <sys/cdefs.h>
80 __FBSDID("$FreeBSD$");
81 
82 #include "opt_inet.h"
83 #include "opt_inet6.h"
84 
85 #include <sys/param.h>
86 #include <sys/eventhandler.h>
87 #include <sys/mbuf.h>
88 #include <sys/malloc.h>
89 #include <sys/protosw.h>
90 #include <sys/systm.h>
91 #include <sys/jail.h>
92 #include <sys/time.h>
93 #include <sys/socket.h> /* for net/if.h */
94 #include <sys/sockio.h>
95 #include <sys/ctype.h>  /* string functions */
96 #include <sys/kernel.h>
97 #include <sys/random.h>
98 #include <sys/syslog.h>
99 #include <sys/sysctl.h>
100 #include <vm/uma.h>
101 #include <sys/module.h>
102 #include <sys/priv.h>
103 #include <sys/proc.h>
104 #include <sys/lock.h>
105 #include <sys/mutex.h>
106 
107 #include <net/bpf.h>
108 #include <net/if.h>
109 #include <net/if_clone.h>
110 #include <net/if_dl.h>
111 #include <net/if_types.h>
112 #include <net/if_var.h>
113 #include <net/pfil.h>
114 #include <net/vnet.h>
115 
116 #include <netinet/in.h>
117 #include <netinet/in_systm.h>
118 #include <netinet/in_var.h>
119 #include <netinet/ip.h>
120 #include <netinet/ip_var.h>
121 #ifdef INET6
122 #include <netinet/ip6.h>
123 #include <netinet6/ip6_var.h>
124 #include <netinet6/in6_ifattach.h>
125 #endif
126 #if defined(INET) || defined(INET6)
127 #include <netinet/ip_carp.h>
128 #endif
129 #include <machine/in_cksum.h>
130 #include <netinet/if_ether.h>
131 #include <net/bridgestp.h>
132 #include <net/if_bridgevar.h>
133 #include <net/if_llc.h>
134 #include <net/if_vlan_var.h>
135 
136 #include <net/route.h>
137 
138 #ifdef INET6
139 /*
140  * XXX: declare here to avoid to include many inet6 related files..
141  * should be more generalized?
142  */
143 extern void	nd6_setmtu(struct ifnet *);
144 #endif
145 
146 /*
147  * Size of the route hash table.  Must be a power of two.
148  */
149 #ifndef BRIDGE_RTHASH_SIZE
150 #define	BRIDGE_RTHASH_SIZE		1024
151 #endif
152 
153 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
154 
155 /*
156  * Default maximum number of addresses to cache.
157  */
158 #ifndef BRIDGE_RTABLE_MAX
159 #define	BRIDGE_RTABLE_MAX		2000
160 #endif
161 
162 /*
163  * Timeout (in seconds) for entries learned dynamically.
164  */
165 #ifndef BRIDGE_RTABLE_TIMEOUT
166 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
167 #endif
168 
169 /*
170  * Number of seconds between walks of the route list.
171  */
172 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
173 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
174 #endif
175 
176 /*
177  * List of capabilities to possibly mask on the member interface.
178  */
179 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
180 					 IFCAP_TXCSUM_IPV6)
181 
182 /*
183  * List of capabilities to strip
184  */
185 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
186 
187 /*
188  * Bridge locking
189  */
190 #define BRIDGE_LOCK_INIT(_sc)		do {			\
191 	mtx_init(&(_sc)->sc_mtx, "if_bridge", NULL, MTX_DEF);	\
192 	cv_init(&(_sc)->sc_cv, "if_bridge_cv");			\
193 } while (0)
194 #define BRIDGE_LOCK_DESTROY(_sc)	do {	\
195 	mtx_destroy(&(_sc)->sc_mtx);		\
196 	cv_destroy(&(_sc)->sc_cv);		\
197 } while (0)
198 #define BRIDGE_LOCK(_sc)		mtx_lock(&(_sc)->sc_mtx)
199 #define BRIDGE_UNLOCK(_sc)		mtx_unlock(&(_sc)->sc_mtx)
200 #define BRIDGE_LOCK_ASSERT(_sc)		mtx_assert(&(_sc)->sc_mtx, MA_OWNED)
201 #define BRIDGE_UNLOCK_ASSERT(_sc)	mtx_assert(&(_sc)->sc_mtx, MA_NOTOWNED)
202 #define	BRIDGE_LOCK2REF(_sc, _err)	do {	\
203 	mtx_assert(&(_sc)->sc_mtx, MA_OWNED);	\
204 	if ((_sc)->sc_iflist_xcnt > 0)		\
205 		(_err) = EBUSY;			\
206 	else					\
207 		(_sc)->sc_iflist_ref++;		\
208 	mtx_unlock(&(_sc)->sc_mtx);		\
209 } while (0)
210 #define	BRIDGE_UNREF(_sc)		do {				\
211 	mtx_lock(&(_sc)->sc_mtx);					\
212 	(_sc)->sc_iflist_ref--;						\
213 	if (((_sc)->sc_iflist_xcnt > 0) && ((_sc)->sc_iflist_ref == 0))	\
214 		cv_broadcast(&(_sc)->sc_cv);				\
215 	mtx_unlock(&(_sc)->sc_mtx);					\
216 } while (0)
217 #define	BRIDGE_XLOCK(_sc)		do {		\
218 	mtx_assert(&(_sc)->sc_mtx, MA_OWNED);		\
219 	(_sc)->sc_iflist_xcnt++;			\
220 	while ((_sc)->sc_iflist_ref > 0)		\
221 		cv_wait(&(_sc)->sc_cv, &(_sc)->sc_mtx);	\
222 } while (0)
223 #define	BRIDGE_XDROP(_sc)		do {	\
224 	mtx_assert(&(_sc)->sc_mtx, MA_OWNED);	\
225 	(_sc)->sc_iflist_xcnt--;		\
226 } while (0)
227 
228 /*
229  * Bridge interface list entry.
230  */
231 struct bridge_iflist {
232 	CK_LIST_ENTRY(bridge_iflist) bif_next;
233 	struct ifnet		*bif_ifp;	/* member if */
234 	struct bstp_port	bif_stp;	/* STP state */
235 	uint32_t		bif_flags;	/* member if flags */
236 	int			bif_savedcaps;	/* saved capabilities */
237 	uint32_t		bif_addrmax;	/* max # of addresses */
238 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
239 	uint32_t		bif_addrexceeded;/* # of address violations */
240 };
241 
242 /*
243  * Bridge route node.
244  */
245 struct bridge_rtnode {
246 	CK_LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
247 	CK_LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
248 	struct bridge_iflist	*brt_dst;	/* destination if */
249 	unsigned long		brt_expire;	/* expiration time */
250 	uint8_t			brt_flags;	/* address flags */
251 	uint8_t			brt_addr[ETHER_ADDR_LEN];
252 	uint16_t		brt_vlan;	/* vlan id */
253 };
254 #define	brt_ifp			brt_dst->bif_ifp
255 
256 /*
257  * Software state for each bridge.
258  */
259 struct bridge_softc {
260 	struct ifnet		*sc_ifp;	/* make this an interface */
261 	LIST_ENTRY(bridge_softc) sc_list;
262 	struct mtx		sc_mtx;
263 	struct cv		sc_cv;
264 	uint32_t		sc_brtmax;	/* max # of addresses */
265 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
266 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
267 	struct callout		sc_brcallout;	/* bridge callout */
268 	uint32_t		sc_iflist_ref;	/* refcount for sc_iflist */
269 	uint32_t		sc_iflist_xcnt;	/* refcount for sc_iflist */
270 	CK_LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
271 	CK_LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
272 	CK_LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
273 	uint32_t		sc_rthash_key;	/* key for hash */
274 	CK_LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
275 	struct bstp_state	sc_stp;		/* STP state */
276 	uint32_t		sc_brtexceeded;	/* # of cache drops */
277 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
278 	struct ether_addr	sc_defaddr;	/* Default MAC address */
279 };
280 
281 VNET_DEFINE_STATIC(struct mtx, bridge_list_mtx);
282 #define	V_bridge_list_mtx	VNET(bridge_list_mtx)
283 static eventhandler_tag bridge_detach_cookie;
284 
285 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
286 
287 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
288 #define	V_bridge_rtnode_zone	VNET(bridge_rtnode_zone)
289 
290 static int	bridge_clone_create(struct if_clone *, int, caddr_t);
291 static void	bridge_clone_destroy(struct ifnet *);
292 
293 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
294 static void	bridge_mutecaps(struct bridge_softc *);
295 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
296 		    int);
297 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
298 static void	bridge_init(void *);
299 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
300 static void	bridge_stop(struct ifnet *, int);
301 static int	bridge_transmit(struct ifnet *, struct mbuf *);
302 static void	bridge_qflush(struct ifnet *);
303 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
304 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
305 		    struct rtentry *);
306 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
307 		    struct mbuf *);
308 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
309 
310 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
311 		    struct mbuf *m);
312 
313 static void	bridge_timer(void *);
314 
315 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
316 		    struct mbuf *, int);
317 static void	bridge_span(struct bridge_softc *, struct mbuf *);
318 
319 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
320 		    uint16_t, struct bridge_iflist *, int, uint8_t);
321 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
322 		    uint16_t);
323 static void	bridge_rttrim(struct bridge_softc *);
324 static void	bridge_rtage(struct bridge_softc *);
325 static void	bridge_rtflush(struct bridge_softc *, int);
326 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
327 		    uint16_t);
328 
329 static void	bridge_rtable_init(struct bridge_softc *);
330 static void	bridge_rtable_fini(struct bridge_softc *);
331 
332 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
333 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
334 		    const uint8_t *, uint16_t);
335 static int	bridge_rtnode_insert(struct bridge_softc *,
336 		    struct bridge_rtnode *);
337 static void	bridge_rtnode_destroy(struct bridge_softc *,
338 		    struct bridge_rtnode *);
339 static void	bridge_rtable_expire(struct ifnet *, int);
340 static void	bridge_state_change(struct ifnet *, int);
341 
342 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
343 		    const char *name);
344 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
345 		    struct ifnet *ifp);
346 static void	bridge_delete_member(struct bridge_softc *,
347 		    struct bridge_iflist *, int);
348 static void	bridge_delete_span(struct bridge_softc *,
349 		    struct bridge_iflist *);
350 
351 static int	bridge_ioctl_add(struct bridge_softc *, void *);
352 static int	bridge_ioctl_del(struct bridge_softc *, void *);
353 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
354 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
355 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
356 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
357 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
358 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
359 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
360 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
361 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
362 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
363 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
364 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
365 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
366 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
367 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
368 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
369 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
370 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
371 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
372 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
373 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
374 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
375 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
376 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
377 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
378 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
379 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
380 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
381 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
382 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
383 		    int);
384 static int	bridge_ip_checkbasic(struct mbuf **mp);
385 #ifdef INET6
386 static int	bridge_ip6_checkbasic(struct mbuf **mp);
387 #endif /* INET6 */
388 static int	bridge_fragment(struct ifnet *, struct mbuf **mp,
389 		    struct ether_header *, int, struct llc *);
390 static void	bridge_linkstate(struct ifnet *ifp);
391 static void	bridge_linkcheck(struct bridge_softc *sc);
392 
393 
394 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
395 #define	VLANTAGOF(_m)	\
396     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
397 
398 static struct bstp_cb_ops bridge_ops = {
399 	.bcb_state = bridge_state_change,
400 	.bcb_rtage = bridge_rtable_expire
401 };
402 
403 SYSCTL_DECL(_net_link);
404 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
405     "Bridge");
406 
407 /* only pass IP[46] packets when pfil is enabled */
408 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
409 #define	V_pfil_onlyip	VNET(pfil_onlyip)
410 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
411     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
412     "Only pass IP packets when pfil is enabled");
413 
414 /* run pfil hooks on the bridge interface */
415 VNET_DEFINE_STATIC(int, pfil_bridge) = 1;
416 #define	V_pfil_bridge	VNET(pfil_bridge)
417 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
418     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
419     "Packet filter on the bridge interface");
420 
421 /* layer2 filter with ipfw */
422 VNET_DEFINE_STATIC(int, pfil_ipfw);
423 #define	V_pfil_ipfw	VNET(pfil_ipfw)
424 
425 /* layer2 ARP filter with ipfw */
426 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
427 #define	V_pfil_ipfw_arp	VNET(pfil_ipfw_arp)
428 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
429     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
430     "Filter ARP packets through IPFW layer2");
431 
432 /* run pfil hooks on the member interface */
433 VNET_DEFINE_STATIC(int, pfil_member) = 1;
434 #define	V_pfil_member	VNET(pfil_member)
435 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
436     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
437     "Packet filter on the member interface");
438 
439 /* run pfil hooks on the physical interface for locally destined packets */
440 VNET_DEFINE_STATIC(int, pfil_local_phys);
441 #define	V_pfil_local_phys	VNET(pfil_local_phys)
442 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
443     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
444     "Packet filter on the physical interface for locally destined packets");
445 
446 /* log STP state changes */
447 VNET_DEFINE_STATIC(int, log_stp);
448 #define	V_log_stp	VNET(log_stp)
449 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
450     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
451     "Log STP state changes");
452 
453 /* share MAC with first bridge member */
454 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
455 #define	V_bridge_inherit_mac	VNET(bridge_inherit_mac)
456 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
457     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
458     "Inherit MAC address from the first bridge member");
459 
460 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
461 #define	V_allow_llz_overlap	VNET(allow_llz_overlap)
462 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
463     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
464     "Allow overlap of link-local scope "
465     "zones of a bridge interface and the member interfaces");
466 
467 struct bridge_control {
468 	int	(*bc_func)(struct bridge_softc *, void *);
469 	int	bc_argsize;
470 	int	bc_flags;
471 };
472 
473 #define	BC_F_COPYIN		0x01	/* copy arguments in */
474 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
475 #define	BC_F_SUSER		0x04	/* do super-user check */
476 
477 const struct bridge_control bridge_control_table[] = {
478 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
479 	  BC_F_COPYIN|BC_F_SUSER },
480 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
481 	  BC_F_COPYIN|BC_F_SUSER },
482 
483 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
484 	  BC_F_COPYIN|BC_F_COPYOUT },
485 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
486 	  BC_F_COPYIN|BC_F_SUSER },
487 
488 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
489 	  BC_F_COPYIN|BC_F_SUSER },
490 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
491 	  BC_F_COPYOUT },
492 
493 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
494 	  BC_F_COPYIN|BC_F_COPYOUT },
495 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
496 	  BC_F_COPYIN|BC_F_COPYOUT },
497 
498 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
499 	  BC_F_COPYIN|BC_F_SUSER },
500 
501 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
502 	  BC_F_COPYIN|BC_F_SUSER },
503 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
504 	  BC_F_COPYOUT },
505 
506 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
507 	  BC_F_COPYIN|BC_F_SUSER },
508 
509 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
510 	  BC_F_COPYIN|BC_F_SUSER },
511 
512 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
513 	  BC_F_COPYOUT },
514 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
515 	  BC_F_COPYIN|BC_F_SUSER },
516 
517 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
518 	  BC_F_COPYOUT },
519 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
520 	  BC_F_COPYIN|BC_F_SUSER },
521 
522 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
523 	  BC_F_COPYOUT },
524 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
525 	  BC_F_COPYIN|BC_F_SUSER },
526 
527 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
528 	  BC_F_COPYOUT },
529 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
530 	  BC_F_COPYIN|BC_F_SUSER },
531 
532 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
533 	  BC_F_COPYIN|BC_F_SUSER },
534 
535 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
536 	  BC_F_COPYIN|BC_F_SUSER },
537 
538 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
539 	  BC_F_COPYIN|BC_F_SUSER },
540 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
541 	  BC_F_COPYIN|BC_F_SUSER },
542 
543 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
544 	  BC_F_COPYOUT },
545 
546 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
547 	  BC_F_COPYOUT },
548 
549 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
550 	  BC_F_COPYIN|BC_F_COPYOUT },
551 
552 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
553 	  BC_F_COPYIN|BC_F_SUSER },
554 
555 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
556 	  BC_F_COPYIN|BC_F_SUSER },
557 
558 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
559 	  BC_F_COPYIN|BC_F_SUSER },
560 
561 };
562 const int bridge_control_table_size = nitems(bridge_control_table);
563 
564 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
565 #define	V_bridge_list	VNET(bridge_list)
566 #define	BRIDGE_LIST_LOCK_INIT(x)	mtx_init(&V_bridge_list_mtx,	\
567 					    "if_bridge list", NULL, MTX_DEF)
568 #define	BRIDGE_LIST_LOCK_DESTROY(x)	mtx_destroy(&V_bridge_list_mtx)
569 #define	BRIDGE_LIST_LOCK(x)		mtx_lock(&V_bridge_list_mtx)
570 #define	BRIDGE_LIST_UNLOCK(x)		mtx_unlock(&V_bridge_list_mtx)
571 
572 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
573 #define	V_bridge_cloner	VNET(bridge_cloner)
574 
575 static const char bridge_name[] = "bridge";
576 
577 static void
578 vnet_bridge_init(const void *unused __unused)
579 {
580 
581 	V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
582 	    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
583 	    UMA_ALIGN_PTR, 0);
584 	BRIDGE_LIST_LOCK_INIT();
585 	LIST_INIT(&V_bridge_list);
586 	V_bridge_cloner = if_clone_simple(bridge_name,
587 	    bridge_clone_create, bridge_clone_destroy, 0);
588 }
589 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
590     vnet_bridge_init, NULL);
591 
592 static void
593 vnet_bridge_uninit(const void *unused __unused)
594 {
595 
596 	if_clone_detach(V_bridge_cloner);
597 	V_bridge_cloner = NULL;
598 	BRIDGE_LIST_LOCK_DESTROY();
599 	uma_zdestroy(V_bridge_rtnode_zone);
600 }
601 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
602     vnet_bridge_uninit, NULL);
603 
604 static int
605 bridge_modevent(module_t mod, int type, void *data)
606 {
607 
608 	switch (type) {
609 	case MOD_LOAD:
610 		bridge_dn_p = bridge_dummynet;
611 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
612 		    ifnet_departure_event, bridge_ifdetach, NULL,
613 		    EVENTHANDLER_PRI_ANY);
614 		break;
615 	case MOD_UNLOAD:
616 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
617 		    bridge_detach_cookie);
618 		bridge_dn_p = NULL;
619 		break;
620 	default:
621 		return (EOPNOTSUPP);
622 	}
623 	return (0);
624 }
625 
626 static moduledata_t bridge_mod = {
627 	"if_bridge",
628 	bridge_modevent,
629 	0
630 };
631 
632 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
633 MODULE_VERSION(if_bridge, 1);
634 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
635 
636 /*
637  * handler for net.link.bridge.ipfw
638  */
639 static int
640 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
641 {
642 	int enable = V_pfil_ipfw;
643 	int error;
644 
645 	error = sysctl_handle_int(oidp, &enable, 0, req);
646 	enable &= 1;
647 
648 	if (enable != V_pfil_ipfw) {
649 		V_pfil_ipfw = enable;
650 
651 		/*
652 		 * Disable pfil so that ipfw doesnt run twice, if the user
653 		 * really wants both then they can re-enable pfil_bridge and/or
654 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
655 		 * layer2 type.
656 		 */
657 		if (V_pfil_ipfw) {
658 			V_pfil_onlyip = 0;
659 			V_pfil_bridge = 0;
660 			V_pfil_member = 0;
661 		}
662 	}
663 
664 	return (error);
665 }
666 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
667     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT,
668     &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
669     "Layer2 filter with IPFW");
670 
671 /*
672  * bridge_clone_create:
673  *
674  *	Create a new bridge instance.
675  */
676 static int
677 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
678 {
679 	struct bridge_softc *sc, *sc2;
680 	struct ifnet *bifp, *ifp;
681 	int fb, retry;
682 	unsigned long hostid;
683 
684 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
685 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
686 	if (ifp == NULL) {
687 		free(sc, M_DEVBUF);
688 		return (ENOSPC);
689 	}
690 
691 	BRIDGE_LOCK_INIT(sc);
692 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
693 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
694 
695 	/* Initialize our routing table. */
696 	bridge_rtable_init(sc);
697 
698 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
699 
700 	CK_LIST_INIT(&sc->sc_iflist);
701 	CK_LIST_INIT(&sc->sc_spanlist);
702 
703 	ifp->if_softc = sc;
704 	if_initname(ifp, bridge_name, unit);
705 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
706 	ifp->if_ioctl = bridge_ioctl;
707 	ifp->if_transmit = bridge_transmit;
708 	ifp->if_qflush = bridge_qflush;
709 	ifp->if_init = bridge_init;
710 	ifp->if_type = IFT_BRIDGE;
711 
712 	/*
713 	 * Generate an ethernet address with a locally administered address.
714 	 *
715 	 * Since we are using random ethernet addresses for the bridge, it is
716 	 * possible that we might have address collisions, so make sure that
717 	 * this hardware address isn't already in use on another bridge.
718 	 * The first try uses the hostid and falls back to arc4rand().
719 	 */
720 	fb = 0;
721 	getcredhostid(curthread->td_ucred, &hostid);
722 	do {
723 		if (fb || hostid == 0) {
724 			ether_gen_addr(ifp, &sc->sc_defaddr);
725 		} else {
726 			sc->sc_defaddr.octet[0] = 0x2;
727 			sc->sc_defaddr.octet[1] = (hostid >> 24) & 0xff;
728 			sc->sc_defaddr.octet[2] = (hostid >> 16) & 0xff;
729 			sc->sc_defaddr.octet[3] = (hostid >> 8 ) & 0xff;
730 			sc->sc_defaddr.octet[4] =  hostid        & 0xff;
731 			sc->sc_defaddr.octet[5] = ifp->if_dunit & 0xff;
732 		}
733 
734 		fb = 1;
735 		retry = 0;
736 		BRIDGE_LIST_LOCK();
737 		LIST_FOREACH(sc2, &V_bridge_list, sc_list) {
738 			bifp = sc2->sc_ifp;
739 			if (memcmp(sc->sc_defaddr.octet,
740 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
741 				retry = 1;
742 				break;
743 			}
744 		}
745 		BRIDGE_LIST_UNLOCK();
746 	} while (retry == 1);
747 
748 	bstp_attach(&sc->sc_stp, &bridge_ops);
749 	ether_ifattach(ifp, sc->sc_defaddr.octet);
750 	/* Now undo some of the damage... */
751 	ifp->if_baudrate = 0;
752 	ifp->if_type = IFT_BRIDGE;
753 
754 	BRIDGE_LIST_LOCK();
755 	LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
756 	BRIDGE_LIST_UNLOCK();
757 
758 	return (0);
759 }
760 
761 /*
762  * bridge_clone_destroy:
763  *
764  *	Destroy a bridge instance.
765  */
766 static void
767 bridge_clone_destroy(struct ifnet *ifp)
768 {
769 	struct bridge_softc *sc = ifp->if_softc;
770 	struct bridge_iflist *bif;
771 
772 	BRIDGE_LOCK(sc);
773 
774 	bridge_stop(ifp, 1);
775 	ifp->if_flags &= ~IFF_UP;
776 
777 	while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
778 		bridge_delete_member(sc, bif, 0);
779 
780 	while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
781 		bridge_delete_span(sc, bif);
782 	}
783 
784 	/* Tear down the routing table. */
785 	bridge_rtable_fini(sc);
786 
787 	BRIDGE_UNLOCK(sc);
788 
789 	callout_drain(&sc->sc_brcallout);
790 
791 	BRIDGE_LIST_LOCK();
792 	LIST_REMOVE(sc, sc_list);
793 	BRIDGE_LIST_UNLOCK();
794 
795 	bstp_detach(&sc->sc_stp);
796 	ether_ifdetach(ifp);
797 	if_free(ifp);
798 
799 	BRIDGE_LOCK_DESTROY(sc);
800 	free(sc, M_DEVBUF);
801 }
802 
803 /*
804  * bridge_ioctl:
805  *
806  *	Handle a control request from the operator.
807  */
808 static int
809 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
810 {
811 	struct bridge_softc *sc = ifp->if_softc;
812 	struct ifreq *ifr = (struct ifreq *)data;
813 	struct bridge_iflist *bif;
814 	struct thread *td = curthread;
815 	union {
816 		struct ifbreq ifbreq;
817 		struct ifbifconf ifbifconf;
818 		struct ifbareq ifbareq;
819 		struct ifbaconf ifbaconf;
820 		struct ifbrparam ifbrparam;
821 		struct ifbropreq ifbropreq;
822 	} args;
823 	struct ifdrv *ifd = (struct ifdrv *) data;
824 	const struct bridge_control *bc;
825 	int error = 0, oldmtu;
826 
827 	switch (cmd) {
828 
829 	case SIOCADDMULTI:
830 	case SIOCDELMULTI:
831 		break;
832 
833 	case SIOCGDRVSPEC:
834 	case SIOCSDRVSPEC:
835 		if (ifd->ifd_cmd >= bridge_control_table_size) {
836 			error = EINVAL;
837 			break;
838 		}
839 		bc = &bridge_control_table[ifd->ifd_cmd];
840 
841 		if (cmd == SIOCGDRVSPEC &&
842 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
843 			error = EINVAL;
844 			break;
845 		}
846 		else if (cmd == SIOCSDRVSPEC &&
847 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
848 			error = EINVAL;
849 			break;
850 		}
851 
852 		if (bc->bc_flags & BC_F_SUSER) {
853 			error = priv_check(td, PRIV_NET_BRIDGE);
854 			if (error)
855 				break;
856 		}
857 
858 		if (ifd->ifd_len != bc->bc_argsize ||
859 		    ifd->ifd_len > sizeof(args)) {
860 			error = EINVAL;
861 			break;
862 		}
863 
864 		bzero(&args, sizeof(args));
865 		if (bc->bc_flags & BC_F_COPYIN) {
866 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
867 			if (error)
868 				break;
869 		}
870 
871 		oldmtu = ifp->if_mtu;
872 		BRIDGE_LOCK(sc);
873 		error = (*bc->bc_func)(sc, &args);
874 		BRIDGE_UNLOCK(sc);
875 		if (error)
876 			break;
877 
878 		/*
879 		 * Bridge MTU may change during addition of the first port.
880 		 * If it did, do network layer specific procedure.
881 		 */
882 		if (ifp->if_mtu != oldmtu) {
883 #ifdef INET6
884 			nd6_setmtu(ifp);
885 #endif
886 			rt_updatemtu(ifp);
887 		}
888 
889 		if (bc->bc_flags & BC_F_COPYOUT)
890 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
891 
892 		break;
893 
894 	case SIOCSIFFLAGS:
895 		if (!(ifp->if_flags & IFF_UP) &&
896 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
897 			/*
898 			 * If interface is marked down and it is running,
899 			 * then stop and disable it.
900 			 */
901 			BRIDGE_LOCK(sc);
902 			bridge_stop(ifp, 1);
903 			BRIDGE_UNLOCK(sc);
904 		} else if ((ifp->if_flags & IFF_UP) &&
905 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
906 			/*
907 			 * If interface is marked up and it is stopped, then
908 			 * start it.
909 			 */
910 			(*ifp->if_init)(sc);
911 		}
912 		break;
913 
914 	case SIOCSIFMTU:
915 		if (ifr->ifr_mtu < 576) {
916 			error = EINVAL;
917 			break;
918 		}
919 		if (CK_LIST_EMPTY(&sc->sc_iflist)) {
920 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
921 			break;
922 		}
923 		BRIDGE_LOCK(sc);
924 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
925 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
926 				log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
927 				    " != %d\n", sc->sc_ifp->if_xname,
928 				    bif->bif_ifp->if_mtu,
929 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
930 				error = EINVAL;
931 				break;
932 			}
933 		}
934 		if (!error)
935 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
936 		BRIDGE_UNLOCK(sc);
937 		break;
938 	default:
939 		/*
940 		 * drop the lock as ether_ioctl() will call bridge_start() and
941 		 * cause the lock to be recursed.
942 		 */
943 		error = ether_ioctl(ifp, cmd, data);
944 		break;
945 	}
946 
947 	return (error);
948 }
949 
950 /*
951  * bridge_mutecaps:
952  *
953  *	Clear or restore unwanted capabilities on the member interface
954  */
955 static void
956 bridge_mutecaps(struct bridge_softc *sc)
957 {
958 	struct bridge_iflist *bif;
959 	int enabled, mask;
960 
961 	/* Initial bitmask of capabilities to test */
962 	mask = BRIDGE_IFCAPS_MASK;
963 
964 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
965 		/* Every member must support it or its disabled */
966 		mask &= bif->bif_savedcaps;
967 	}
968 
969 	BRIDGE_XLOCK(sc);
970 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
971 		enabled = bif->bif_ifp->if_capenable;
972 		enabled &= ~BRIDGE_IFCAPS_STRIP;
973 		/* strip off mask bits and enable them again if allowed */
974 		enabled &= ~BRIDGE_IFCAPS_MASK;
975 		enabled |= mask;
976 		BRIDGE_UNLOCK(sc);
977 		bridge_set_ifcap(sc, bif, enabled);
978 		BRIDGE_LOCK(sc);
979 	}
980 	BRIDGE_XDROP(sc);
981 
982 }
983 
984 static void
985 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
986 {
987 	struct ifnet *ifp = bif->bif_ifp;
988 	struct ifreq ifr;
989 	int error, mask, stuck;
990 
991 	BRIDGE_UNLOCK_ASSERT(sc);
992 
993 	bzero(&ifr, sizeof(ifr));
994 	ifr.ifr_reqcap = set;
995 
996 	if (ifp->if_capenable != set) {
997 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
998 		if (error)
999 			if_printf(sc->sc_ifp,
1000 			    "error setting capabilities on %s: %d\n",
1001 			    ifp->if_xname, error);
1002 		mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1003 		stuck = ifp->if_capenable & mask & ~set;
1004 		if (stuck != 0)
1005 			if_printf(sc->sc_ifp,
1006 			    "can't disable some capabilities on %s: 0x%x\n",
1007 			    ifp->if_xname, stuck);
1008 	}
1009 }
1010 
1011 /*
1012  * bridge_lookup_member:
1013  *
1014  *	Lookup a bridge member interface.
1015  */
1016 static struct bridge_iflist *
1017 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1018 {
1019 	struct bridge_iflist *bif;
1020 	struct ifnet *ifp;
1021 
1022 	BRIDGE_LOCK_ASSERT(sc);
1023 
1024 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1025 		ifp = bif->bif_ifp;
1026 		if (strcmp(ifp->if_xname, name) == 0)
1027 			return (bif);
1028 	}
1029 
1030 	return (NULL);
1031 }
1032 
1033 /*
1034  * bridge_lookup_member_if:
1035  *
1036  *	Lookup a bridge member interface by ifnet*.
1037  */
1038 static struct bridge_iflist *
1039 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1040 {
1041 	struct bridge_iflist *bif;
1042 
1043 	BRIDGE_LOCK_ASSERT(sc);
1044 
1045 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1046 		if (bif->bif_ifp == member_ifp)
1047 			return (bif);
1048 	}
1049 
1050 	return (NULL);
1051 }
1052 
1053 /*
1054  * bridge_delete_member:
1055  *
1056  *	Delete the specified member interface.
1057  */
1058 static void
1059 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1060     int gone)
1061 {
1062 	struct ifnet *ifs = bif->bif_ifp;
1063 	struct ifnet *fif = NULL;
1064 	struct bridge_iflist *bifl;
1065 
1066 	BRIDGE_LOCK_ASSERT(sc);
1067 
1068 	if (bif->bif_flags & IFBIF_STP)
1069 		bstp_disable(&bif->bif_stp);
1070 
1071 	ifs->if_bridge = NULL;
1072 	BRIDGE_XLOCK(sc);
1073 	CK_LIST_REMOVE(bif, bif_next);
1074 	BRIDGE_XDROP(sc);
1075 
1076 	/*
1077 	 * If removing the interface that gave the bridge its mac address, set
1078 	 * the mac address of the bridge to the address of the next member, or
1079 	 * to its default address if no members are left.
1080 	 */
1081 	if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1082 		if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1083 			bcopy(&sc->sc_defaddr,
1084 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1085 			sc->sc_ifaddr = NULL;
1086 		} else {
1087 			bifl = CK_LIST_FIRST(&sc->sc_iflist);
1088 			fif = bifl->bif_ifp;
1089 			bcopy(IF_LLADDR(fif),
1090 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1091 			sc->sc_ifaddr = fif;
1092 		}
1093 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1094 	}
1095 
1096 	bridge_linkcheck(sc);
1097 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
1098 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1099 	KASSERT(bif->bif_addrcnt == 0,
1100 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1101 
1102 	ifs->if_bridge_output = NULL;
1103 	ifs->if_bridge_input = NULL;
1104 	ifs->if_bridge_linkstate = NULL;
1105 	BRIDGE_UNLOCK(sc);
1106 	if (!gone) {
1107 		switch (ifs->if_type) {
1108 		case IFT_ETHER:
1109 		case IFT_L2VLAN:
1110 			/*
1111 			 * Take the interface out of promiscuous mode, but only
1112 			 * if it was promiscuous in the first place. It might
1113 			 * not be if we're in the bridge_ioctl_add() error path.
1114 			 */
1115 			if (ifs->if_flags & IFF_PROMISC)
1116 				(void) ifpromisc(ifs, 0);
1117 			break;
1118 
1119 		case IFT_GIF:
1120 			break;
1121 
1122 		default:
1123 #ifdef DIAGNOSTIC
1124 			panic("bridge_delete_member: impossible");
1125 #endif
1126 			break;
1127 		}
1128 		/* reneable any interface capabilities */
1129 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1130 	}
1131 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1132 	BRIDGE_LOCK(sc);
1133 	free(bif, M_DEVBUF);
1134 }
1135 
1136 /*
1137  * bridge_delete_span:
1138  *
1139  *	Delete the specified span interface.
1140  */
1141 static void
1142 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1143 {
1144 	BRIDGE_LOCK_ASSERT(sc);
1145 
1146 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1147 	    ("%s: not a span interface", __func__));
1148 
1149 	CK_LIST_REMOVE(bif, bif_next);
1150 	free(bif, M_DEVBUF);
1151 }
1152 
1153 static int
1154 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1155 {
1156 	struct ifbreq *req = arg;
1157 	struct bridge_iflist *bif = NULL;
1158 	struct ifnet *ifs;
1159 	int error = 0;
1160 
1161 	ifs = ifunit(req->ifbr_ifsname);
1162 	if (ifs == NULL)
1163 		return (ENOENT);
1164 	if (ifs->if_ioctl == NULL)	/* must be supported */
1165 		return (EINVAL);
1166 
1167 	/* If it's in the span list, it can't be a member. */
1168 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1169 		if (ifs == bif->bif_ifp)
1170 			return (EBUSY);
1171 
1172 	if (ifs->if_bridge == sc)
1173 		return (EEXIST);
1174 
1175 	if (ifs->if_bridge != NULL)
1176 		return (EBUSY);
1177 
1178 	switch (ifs->if_type) {
1179 	case IFT_ETHER:
1180 	case IFT_L2VLAN:
1181 	case IFT_GIF:
1182 		/* permitted interface types */
1183 		break;
1184 	default:
1185 		return (EINVAL);
1186 	}
1187 
1188 #ifdef INET6
1189 	/*
1190 	 * Two valid inet6 addresses with link-local scope must not be
1191 	 * on the parent interface and the member interfaces at the
1192 	 * same time.  This restriction is needed to prevent violation
1193 	 * of link-local scope zone.  Attempts to add a member
1194 	 * interface which has inet6 addresses when the parent has
1195 	 * inet6 triggers removal of all inet6 addresses on the member
1196 	 * interface.
1197 	 */
1198 
1199 	/* Check if the parent interface has a link-local scope addr. */
1200 	if (V_allow_llz_overlap == 0 &&
1201 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1202 		/*
1203 		 * If any, remove all inet6 addresses from the member
1204 		 * interfaces.
1205 		 */
1206 		BRIDGE_XLOCK(sc);
1207 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1208  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1209 				BRIDGE_UNLOCK(sc);
1210 				in6_ifdetach(bif->bif_ifp);
1211 				BRIDGE_LOCK(sc);
1212 				if_printf(sc->sc_ifp,
1213 				    "IPv6 addresses on %s have been removed "
1214 				    "before adding it as a member to prevent "
1215 				    "IPv6 address scope violation.\n",
1216 				    bif->bif_ifp->if_xname);
1217 			}
1218 		}
1219 		BRIDGE_XDROP(sc);
1220 		if (in6ifa_llaonifp(ifs)) {
1221 			BRIDGE_UNLOCK(sc);
1222 			in6_ifdetach(ifs);
1223 			BRIDGE_LOCK(sc);
1224 			if_printf(sc->sc_ifp,
1225 			    "IPv6 addresses on %s have been removed "
1226 			    "before adding it as a member to prevent "
1227 			    "IPv6 address scope violation.\n",
1228 			    ifs->if_xname);
1229 		}
1230 	}
1231 #endif
1232 	/* Allow the first Ethernet member to define the MTU */
1233 	if (CK_LIST_EMPTY(&sc->sc_iflist))
1234 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1235 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1236 		if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1237 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1238 		return (EINVAL);
1239 	}
1240 
1241 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1242 	if (bif == NULL)
1243 		return (ENOMEM);
1244 
1245 	bif->bif_ifp = ifs;
1246 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1247 	bif->bif_savedcaps = ifs->if_capenable;
1248 
1249 	/*
1250 	 * Assign the interface's MAC address to the bridge if it's the first
1251 	 * member and the MAC address of the bridge has not been changed from
1252 	 * the default randomly generated one.
1253 	 */
1254 	if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1255 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1256 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1257 		sc->sc_ifaddr = ifs;
1258 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1259 	}
1260 
1261 	ifs->if_bridge = sc;
1262 	ifs->if_bridge_output = bridge_output;
1263 	ifs->if_bridge_input = bridge_input;
1264 	ifs->if_bridge_linkstate = bridge_linkstate;
1265 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1266 	/*
1267 	 * XXX: XLOCK HERE!?!
1268 	 *
1269 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1270 	 */
1271 	CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1272 
1273 	/* Set interface capabilities to the intersection set of all members */
1274 	bridge_mutecaps(sc);
1275 	bridge_linkcheck(sc);
1276 
1277 	/* Place the interface into promiscuous mode */
1278 	switch (ifs->if_type) {
1279 		case IFT_ETHER:
1280 		case IFT_L2VLAN:
1281 			BRIDGE_UNLOCK(sc);
1282 			error = ifpromisc(ifs, 1);
1283 			BRIDGE_LOCK(sc);
1284 			break;
1285 	}
1286 
1287 	if (error)
1288 		bridge_delete_member(sc, bif, 0);
1289 	return (error);
1290 }
1291 
1292 static int
1293 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1294 {
1295 	struct ifbreq *req = arg;
1296 	struct bridge_iflist *bif;
1297 
1298 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1299 	if (bif == NULL)
1300 		return (ENOENT);
1301 
1302 	bridge_delete_member(sc, bif, 0);
1303 
1304 	return (0);
1305 }
1306 
1307 static int
1308 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1309 {
1310 	struct ifbreq *req = arg;
1311 	struct bridge_iflist *bif;
1312 	struct bstp_port *bp;
1313 
1314 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1315 	if (bif == NULL)
1316 		return (ENOENT);
1317 
1318 	bp = &bif->bif_stp;
1319 	req->ifbr_ifsflags = bif->bif_flags;
1320 	req->ifbr_state = bp->bp_state;
1321 	req->ifbr_priority = bp->bp_priority;
1322 	req->ifbr_path_cost = bp->bp_path_cost;
1323 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1324 	req->ifbr_proto = bp->bp_protover;
1325 	req->ifbr_role = bp->bp_role;
1326 	req->ifbr_stpflags = bp->bp_flags;
1327 	req->ifbr_addrcnt = bif->bif_addrcnt;
1328 	req->ifbr_addrmax = bif->bif_addrmax;
1329 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1330 
1331 	/* Copy STP state options as flags */
1332 	if (bp->bp_operedge)
1333 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1334 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1335 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1336 	if (bp->bp_ptp_link)
1337 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1338 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1339 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1340 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1341 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1342 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1343 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1344 	return (0);
1345 }
1346 
1347 static int
1348 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1349 {
1350 	struct ifbreq *req = arg;
1351 	struct bridge_iflist *bif;
1352 	struct bstp_port *bp;
1353 	int error;
1354 
1355 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1356 	if (bif == NULL)
1357 		return (ENOENT);
1358 	bp = &bif->bif_stp;
1359 
1360 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1361 		/* SPAN is readonly */
1362 		return (EINVAL);
1363 
1364 	if (req->ifbr_ifsflags & IFBIF_STP) {
1365 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1366 			error = bstp_enable(&bif->bif_stp);
1367 			if (error)
1368 				return (error);
1369 		}
1370 	} else {
1371 		if ((bif->bif_flags & IFBIF_STP) != 0)
1372 			bstp_disable(&bif->bif_stp);
1373 	}
1374 
1375 	/* Pass on STP flags */
1376 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1377 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1378 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1379 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1380 
1381 	/* Save the bits relating to the bridge */
1382 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1383 
1384 	return (0);
1385 }
1386 
1387 static int
1388 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1389 {
1390 	struct ifbrparam *param = arg;
1391 
1392 	sc->sc_brtmax = param->ifbrp_csize;
1393 	bridge_rttrim(sc);
1394 
1395 	return (0);
1396 }
1397 
1398 static int
1399 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1400 {
1401 	struct ifbrparam *param = arg;
1402 
1403 	param->ifbrp_csize = sc->sc_brtmax;
1404 
1405 	return (0);
1406 }
1407 
1408 static int
1409 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1410 {
1411 	struct ifbifconf *bifc = arg;
1412 	struct bridge_iflist *bif;
1413 	struct ifbreq breq;
1414 	char *buf, *outbuf;
1415 	int count, buflen, len, error = 0;
1416 
1417 	count = 0;
1418 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1419 		count++;
1420 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1421 		count++;
1422 
1423 	buflen = sizeof(breq) * count;
1424 	if (bifc->ifbic_len == 0) {
1425 		bifc->ifbic_len = buflen;
1426 		return (0);
1427 	}
1428 	BRIDGE_UNLOCK(sc);
1429 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1430 	BRIDGE_LOCK(sc);
1431 
1432 	count = 0;
1433 	buf = outbuf;
1434 	len = min(bifc->ifbic_len, buflen);
1435 	bzero(&breq, sizeof(breq));
1436 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1437 		if (len < sizeof(breq))
1438 			break;
1439 
1440 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1441 		    sizeof(breq.ifbr_ifsname));
1442 		/* Fill in the ifbreq structure */
1443 		error = bridge_ioctl_gifflags(sc, &breq);
1444 		if (error)
1445 			break;
1446 		memcpy(buf, &breq, sizeof(breq));
1447 		count++;
1448 		buf += sizeof(breq);
1449 		len -= sizeof(breq);
1450 	}
1451 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1452 		if (len < sizeof(breq))
1453 			break;
1454 
1455 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1456 		    sizeof(breq.ifbr_ifsname));
1457 		breq.ifbr_ifsflags = bif->bif_flags;
1458 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1459 		memcpy(buf, &breq, sizeof(breq));
1460 		count++;
1461 		buf += sizeof(breq);
1462 		len -= sizeof(breq);
1463 	}
1464 
1465 	BRIDGE_UNLOCK(sc);
1466 	bifc->ifbic_len = sizeof(breq) * count;
1467 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1468 	BRIDGE_LOCK(sc);
1469 	free(outbuf, M_TEMP);
1470 	return (error);
1471 }
1472 
1473 static int
1474 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1475 {
1476 	struct ifbaconf *bac = arg;
1477 	struct bridge_rtnode *brt;
1478 	struct ifbareq bareq;
1479 	char *buf, *outbuf;
1480 	int count, buflen, len, error = 0;
1481 
1482 	if (bac->ifbac_len == 0)
1483 		return (0);
1484 
1485 	count = 0;
1486 	CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1487 		count++;
1488 	buflen = sizeof(bareq) * count;
1489 
1490 	BRIDGE_UNLOCK(sc);
1491 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1492 	BRIDGE_LOCK(sc);
1493 
1494 	count = 0;
1495 	buf = outbuf;
1496 	len = min(bac->ifbac_len, buflen);
1497 	bzero(&bareq, sizeof(bareq));
1498 	CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1499 		if (len < sizeof(bareq))
1500 			goto out;
1501 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1502 		    sizeof(bareq.ifba_ifsname));
1503 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1504 		bareq.ifba_vlan = brt->brt_vlan;
1505 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1506 				time_uptime < brt->brt_expire)
1507 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1508 		else
1509 			bareq.ifba_expire = 0;
1510 		bareq.ifba_flags = brt->brt_flags;
1511 
1512 		memcpy(buf, &bareq, sizeof(bareq));
1513 		count++;
1514 		buf += sizeof(bareq);
1515 		len -= sizeof(bareq);
1516 	}
1517 out:
1518 	BRIDGE_UNLOCK(sc);
1519 	bac->ifbac_len = sizeof(bareq) * count;
1520 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1521 	BRIDGE_LOCK(sc);
1522 	free(outbuf, M_TEMP);
1523 	return (error);
1524 }
1525 
1526 static int
1527 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1528 {
1529 	struct ifbareq *req = arg;
1530 	struct bridge_iflist *bif;
1531 	int error;
1532 
1533 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1534 	if (bif == NULL)
1535 		return (ENOENT);
1536 
1537 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1538 	    req->ifba_flags);
1539 
1540 	return (error);
1541 }
1542 
1543 static int
1544 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1545 {
1546 	struct ifbrparam *param = arg;
1547 
1548 	sc->sc_brttimeout = param->ifbrp_ctime;
1549 	return (0);
1550 }
1551 
1552 static int
1553 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1554 {
1555 	struct ifbrparam *param = arg;
1556 
1557 	param->ifbrp_ctime = sc->sc_brttimeout;
1558 	return (0);
1559 }
1560 
1561 static int
1562 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1563 {
1564 	struct ifbareq *req = arg;
1565 
1566 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1567 }
1568 
1569 static int
1570 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1571 {
1572 	struct ifbreq *req = arg;
1573 
1574 	bridge_rtflush(sc, req->ifbr_ifsflags);
1575 	return (0);
1576 }
1577 
1578 static int
1579 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1580 {
1581 	struct ifbrparam *param = arg;
1582 	struct bstp_state *bs = &sc->sc_stp;
1583 
1584 	param->ifbrp_prio = bs->bs_bridge_priority;
1585 	return (0);
1586 }
1587 
1588 static int
1589 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1590 {
1591 	struct ifbrparam *param = arg;
1592 
1593 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1594 }
1595 
1596 static int
1597 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1598 {
1599 	struct ifbrparam *param = arg;
1600 	struct bstp_state *bs = &sc->sc_stp;
1601 
1602 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1603 	return (0);
1604 }
1605 
1606 static int
1607 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1608 {
1609 	struct ifbrparam *param = arg;
1610 
1611 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1612 }
1613 
1614 static int
1615 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1616 {
1617 	struct ifbrparam *param = arg;
1618 	struct bstp_state *bs = &sc->sc_stp;
1619 
1620 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1621 	return (0);
1622 }
1623 
1624 static int
1625 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1626 {
1627 	struct ifbrparam *param = arg;
1628 
1629 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1630 }
1631 
1632 static int
1633 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1634 {
1635 	struct ifbrparam *param = arg;
1636 	struct bstp_state *bs = &sc->sc_stp;
1637 
1638 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1639 	return (0);
1640 }
1641 
1642 static int
1643 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1644 {
1645 	struct ifbrparam *param = arg;
1646 
1647 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1648 }
1649 
1650 static int
1651 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1652 {
1653 	struct ifbreq *req = arg;
1654 	struct bridge_iflist *bif;
1655 
1656 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1657 	if (bif == NULL)
1658 		return (ENOENT);
1659 
1660 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1661 }
1662 
1663 static int
1664 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1665 {
1666 	struct ifbreq *req = arg;
1667 	struct bridge_iflist *bif;
1668 
1669 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1670 	if (bif == NULL)
1671 		return (ENOENT);
1672 
1673 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1674 }
1675 
1676 static int
1677 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1678 {
1679 	struct ifbreq *req = arg;
1680 	struct bridge_iflist *bif;
1681 
1682 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1683 	if (bif == NULL)
1684 		return (ENOENT);
1685 
1686 	bif->bif_addrmax = req->ifbr_addrmax;
1687 	return (0);
1688 }
1689 
1690 static int
1691 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1692 {
1693 	struct ifbreq *req = arg;
1694 	struct bridge_iflist *bif = NULL;
1695 	struct ifnet *ifs;
1696 
1697 	ifs = ifunit(req->ifbr_ifsname);
1698 	if (ifs == NULL)
1699 		return (ENOENT);
1700 
1701 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1702 		if (ifs == bif->bif_ifp)
1703 			return (EBUSY);
1704 
1705 	if (ifs->if_bridge != NULL)
1706 		return (EBUSY);
1707 
1708 	switch (ifs->if_type) {
1709 		case IFT_ETHER:
1710 		case IFT_GIF:
1711 		case IFT_L2VLAN:
1712 			break;
1713 		default:
1714 			return (EINVAL);
1715 	}
1716 
1717 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1718 	if (bif == NULL)
1719 		return (ENOMEM);
1720 
1721 	bif->bif_ifp = ifs;
1722 	bif->bif_flags = IFBIF_SPAN;
1723 
1724 	CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1725 
1726 	return (0);
1727 }
1728 
1729 static int
1730 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1731 {
1732 	struct ifbreq *req = arg;
1733 	struct bridge_iflist *bif;
1734 	struct ifnet *ifs;
1735 
1736 	ifs = ifunit(req->ifbr_ifsname);
1737 	if (ifs == NULL)
1738 		return (ENOENT);
1739 
1740 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1741 		if (ifs == bif->bif_ifp)
1742 			break;
1743 
1744 	if (bif == NULL)
1745 		return (ENOENT);
1746 
1747 	bridge_delete_span(sc, bif);
1748 
1749 	return (0);
1750 }
1751 
1752 static int
1753 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1754 {
1755 	struct ifbropreq *req = arg;
1756 	struct bstp_state *bs = &sc->sc_stp;
1757 	struct bstp_port *root_port;
1758 
1759 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1760 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1761 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1762 
1763 	root_port = bs->bs_root_port;
1764 	if (root_port == NULL)
1765 		req->ifbop_root_port = 0;
1766 	else
1767 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1768 
1769 	req->ifbop_holdcount = bs->bs_txholdcount;
1770 	req->ifbop_priority = bs->bs_bridge_priority;
1771 	req->ifbop_protocol = bs->bs_protover;
1772 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1773 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1774 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1775 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1776 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1777 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1778 
1779 	return (0);
1780 }
1781 
1782 static int
1783 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1784 {
1785 	struct ifbrparam *param = arg;
1786 
1787 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1788 	return (0);
1789 }
1790 
1791 static int
1792 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1793 {
1794 	struct ifbpstpconf *bifstp = arg;
1795 	struct bridge_iflist *bif;
1796 	struct bstp_port *bp;
1797 	struct ifbpstpreq bpreq;
1798 	char *buf, *outbuf;
1799 	int count, buflen, len, error = 0;
1800 
1801 	count = 0;
1802 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1803 		if ((bif->bif_flags & IFBIF_STP) != 0)
1804 			count++;
1805 	}
1806 
1807 	buflen = sizeof(bpreq) * count;
1808 	if (bifstp->ifbpstp_len == 0) {
1809 		bifstp->ifbpstp_len = buflen;
1810 		return (0);
1811 	}
1812 
1813 	BRIDGE_UNLOCK(sc);
1814 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1815 	BRIDGE_LOCK(sc);
1816 
1817 	count = 0;
1818 	buf = outbuf;
1819 	len = min(bifstp->ifbpstp_len, buflen);
1820 	bzero(&bpreq, sizeof(bpreq));
1821 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1822 		if (len < sizeof(bpreq))
1823 			break;
1824 
1825 		if ((bif->bif_flags & IFBIF_STP) == 0)
1826 			continue;
1827 
1828 		bp = &bif->bif_stp;
1829 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1830 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1831 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1832 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1833 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1834 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1835 
1836 		memcpy(buf, &bpreq, sizeof(bpreq));
1837 		count++;
1838 		buf += sizeof(bpreq);
1839 		len -= sizeof(bpreq);
1840 	}
1841 
1842 	BRIDGE_UNLOCK(sc);
1843 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1844 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1845 	BRIDGE_LOCK(sc);
1846 	free(outbuf, M_TEMP);
1847 	return (error);
1848 }
1849 
1850 static int
1851 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1852 {
1853 	struct ifbrparam *param = arg;
1854 
1855 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1856 }
1857 
1858 static int
1859 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1860 {
1861 	struct ifbrparam *param = arg;
1862 
1863 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1864 }
1865 
1866 /*
1867  * bridge_ifdetach:
1868  *
1869  *	Detach an interface from a bridge.  Called when a member
1870  *	interface is detaching.
1871  */
1872 static void
1873 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1874 {
1875 	struct bridge_softc *sc = ifp->if_bridge;
1876 	struct bridge_iflist *bif;
1877 
1878 	if (ifp->if_flags & IFF_RENAMING)
1879 		return;
1880 	if (V_bridge_cloner == NULL) {
1881 		/*
1882 		 * This detach handler can be called after
1883 		 * vnet_bridge_uninit().  Just return in that case.
1884 		 */
1885 		return;
1886 	}
1887 	/* Check if the interface is a bridge member */
1888 	if (sc != NULL) {
1889 		BRIDGE_LOCK(sc);
1890 
1891 		bif = bridge_lookup_member_if(sc, ifp);
1892 		if (bif != NULL)
1893 			bridge_delete_member(sc, bif, 1);
1894 
1895 		BRIDGE_UNLOCK(sc);
1896 		return;
1897 	}
1898 
1899 	/* Check if the interface is a span port */
1900 	BRIDGE_LIST_LOCK();
1901 	LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1902 		BRIDGE_LOCK(sc);
1903 		CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1904 			if (ifp == bif->bif_ifp) {
1905 				bridge_delete_span(sc, bif);
1906 				break;
1907 			}
1908 
1909 		BRIDGE_UNLOCK(sc);
1910 	}
1911 	BRIDGE_LIST_UNLOCK();
1912 }
1913 
1914 /*
1915  * bridge_init:
1916  *
1917  *	Initialize a bridge interface.
1918  */
1919 static void
1920 bridge_init(void *xsc)
1921 {
1922 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1923 	struct ifnet *ifp = sc->sc_ifp;
1924 
1925 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1926 		return;
1927 
1928 	BRIDGE_LOCK(sc);
1929 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1930 	    bridge_timer, sc);
1931 
1932 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1933 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1934 
1935 	BRIDGE_UNLOCK(sc);
1936 }
1937 
1938 /*
1939  * bridge_stop:
1940  *
1941  *	Stop the bridge interface.
1942  */
1943 static void
1944 bridge_stop(struct ifnet *ifp, int disable)
1945 {
1946 	struct bridge_softc *sc = ifp->if_softc;
1947 
1948 	BRIDGE_LOCK_ASSERT(sc);
1949 
1950 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1951 		return;
1952 
1953 	callout_stop(&sc->sc_brcallout);
1954 	bstp_stop(&sc->sc_stp);
1955 
1956 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1957 
1958 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1959 }
1960 
1961 /*
1962  * bridge_enqueue:
1963  *
1964  *	Enqueue a packet on a bridge member interface.
1965  *
1966  */
1967 static int
1968 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1969 {
1970 	int len, err = 0;
1971 	short mflags;
1972 	struct mbuf *m0;
1973 
1974 	/* We may be sending a fragment so traverse the mbuf */
1975 	for (; m; m = m0) {
1976 		m0 = m->m_nextpkt;
1977 		m->m_nextpkt = NULL;
1978 		len = m->m_pkthdr.len;
1979 		mflags = m->m_flags;
1980 
1981 		/*
1982 		 * If underlying interface can not do VLAN tag insertion itself
1983 		 * then attach a packet tag that holds it.
1984 		 */
1985 		if ((m->m_flags & M_VLANTAG) &&
1986 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1987 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1988 			if (m == NULL) {
1989 				if_printf(dst_ifp,
1990 				    "unable to prepend VLAN header\n");
1991 				if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
1992 				continue;
1993 			}
1994 			m->m_flags &= ~M_VLANTAG;
1995 		}
1996 
1997 		M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
1998 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1999 			m_freem(m0);
2000 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2001 			break;
2002 		}
2003 
2004 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2005 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2006 		if (mflags & M_MCAST)
2007 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2008 	}
2009 
2010 	return (err);
2011 }
2012 
2013 /*
2014  * bridge_dummynet:
2015  *
2016  * 	Receive a queued packet from dummynet and pass it on to the output
2017  * 	interface.
2018  *
2019  *	The mbuf has the Ethernet header already attached.
2020  */
2021 static void
2022 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2023 {
2024 	struct bridge_softc *sc;
2025 
2026 	sc = ifp->if_bridge;
2027 
2028 	/*
2029 	 * The packet didnt originate from a member interface. This should only
2030 	 * ever happen if a member interface is removed while packets are
2031 	 * queued for it.
2032 	 */
2033 	if (sc == NULL) {
2034 		m_freem(m);
2035 		return;
2036 	}
2037 
2038 	if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2039 #ifdef INET6
2040 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2041 #endif
2042 	    ) {
2043 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2044 			return;
2045 		if (m == NULL)
2046 			return;
2047 	}
2048 
2049 	bridge_enqueue(sc, ifp, m);
2050 }
2051 
2052 /*
2053  * bridge_output:
2054  *
2055  *	Send output from a bridge member interface.  This
2056  *	performs the bridging function for locally originated
2057  *	packets.
2058  *
2059  *	The mbuf has the Ethernet header already attached.  We must
2060  *	enqueue or free the mbuf before returning.
2061  */
2062 static int
2063 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2064     struct rtentry *rt)
2065 {
2066 	struct ether_header *eh;
2067 	struct ifnet *bifp, *dst_if;
2068 	struct bridge_softc *sc;
2069 	uint16_t vlan;
2070 
2071 	if (m->m_len < ETHER_HDR_LEN) {
2072 		m = m_pullup(m, ETHER_HDR_LEN);
2073 		if (m == NULL)
2074 			return (0);
2075 	}
2076 
2077 	eh = mtod(m, struct ether_header *);
2078 	sc = ifp->if_bridge;
2079 	vlan = VLANTAGOF(m);
2080 
2081 	BRIDGE_LOCK(sc);
2082 	bifp = sc->sc_ifp;
2083 
2084 	/*
2085 	 * If bridge is down, but the original output interface is up,
2086 	 * go ahead and send out that interface.  Otherwise, the packet
2087 	 * is dropped below.
2088 	 */
2089 	if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2090 		dst_if = ifp;
2091 		goto sendunicast;
2092 	}
2093 
2094 	/*
2095 	 * If the packet is a multicast, or we don't know a better way to
2096 	 * get there, send to all interfaces.
2097 	 */
2098 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
2099 		dst_if = NULL;
2100 	else
2101 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2102 	/* Tap any traffic not passing back out the originating interface */
2103 	if (dst_if != ifp)
2104 		ETHER_BPF_MTAP(bifp, m);
2105 	if (dst_if == NULL) {
2106 		struct bridge_iflist *bif;
2107 		struct mbuf *mc;
2108 		int error = 0, used = 0;
2109 
2110 		bridge_span(sc, m);
2111 
2112 		BRIDGE_LOCK2REF(sc, error);
2113 		if (error) {
2114 			m_freem(m);
2115 			return (0);
2116 		}
2117 
2118 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2119 			dst_if = bif->bif_ifp;
2120 
2121 			if (dst_if->if_type == IFT_GIF)
2122 				continue;
2123 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2124 				continue;
2125 
2126 			/*
2127 			 * If this is not the original output interface,
2128 			 * and the interface is participating in spanning
2129 			 * tree, make sure the port is in a state that
2130 			 * allows forwarding.
2131 			 */
2132 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2133 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2134 				continue;
2135 
2136 			if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2137 				used = 1;
2138 				mc = m;
2139 			} else {
2140 				mc = m_copypacket(m, M_NOWAIT);
2141 				if (mc == NULL) {
2142 					if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2143 					continue;
2144 				}
2145 			}
2146 
2147 			bridge_enqueue(sc, dst_if, mc);
2148 		}
2149 		if (used == 0)
2150 			m_freem(m);
2151 		BRIDGE_UNREF(sc);
2152 		return (0);
2153 	}
2154 
2155 sendunicast:
2156 	/*
2157 	 * XXX Spanning tree consideration here?
2158 	 */
2159 
2160 	bridge_span(sc, m);
2161 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2162 		m_freem(m);
2163 		BRIDGE_UNLOCK(sc);
2164 		return (0);
2165 	}
2166 
2167 	BRIDGE_UNLOCK(sc);
2168 	bridge_enqueue(sc, dst_if, m);
2169 	return (0);
2170 }
2171 
2172 /*
2173  * bridge_transmit:
2174  *
2175  *	Do output on a bridge.
2176  *
2177  */
2178 static int
2179 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2180 {
2181 	struct bridge_softc *sc;
2182 	struct ether_header *eh;
2183 	struct ifnet *dst_if;
2184 	int error = 0;
2185 
2186 	sc = ifp->if_softc;
2187 
2188 	ETHER_BPF_MTAP(ifp, m);
2189 
2190 	eh = mtod(m, struct ether_header *);
2191 
2192 	BRIDGE_LOCK(sc);
2193 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2194 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2195 		BRIDGE_UNLOCK(sc);
2196 		error = bridge_enqueue(sc, dst_if, m);
2197 	} else
2198 		bridge_broadcast(sc, ifp, m, 0);
2199 
2200 	return (error);
2201 }
2202 
2203 /*
2204  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2205  */
2206 static void
2207 bridge_qflush(struct ifnet *ifp __unused)
2208 {
2209 }
2210 
2211 /*
2212  * bridge_forward:
2213  *
2214  *	The forwarding function of the bridge.
2215  *
2216  *	NOTE: Releases the lock on return.
2217  */
2218 static void
2219 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2220     struct mbuf *m)
2221 {
2222 	struct bridge_iflist *dbif;
2223 	struct ifnet *src_if, *dst_if, *ifp;
2224 	struct ether_header *eh;
2225 	uint16_t vlan;
2226 	uint8_t *dst;
2227 	int error;
2228 
2229 	src_if = m->m_pkthdr.rcvif;
2230 	ifp = sc->sc_ifp;
2231 
2232 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2233 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2234 	vlan = VLANTAGOF(m);
2235 
2236 	if ((sbif->bif_flags & IFBIF_STP) &&
2237 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2238 		goto drop;
2239 
2240 	eh = mtod(m, struct ether_header *);
2241 	dst = eh->ether_dhost;
2242 
2243 	/* If the interface is learning, record the address. */
2244 	if (sbif->bif_flags & IFBIF_LEARNING) {
2245 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2246 		    sbif, 0, IFBAF_DYNAMIC);
2247 		/*
2248 		 * If the interface has addresses limits then deny any source
2249 		 * that is not in the cache.
2250 		 */
2251 		if (error && sbif->bif_addrmax)
2252 			goto drop;
2253 	}
2254 
2255 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2256 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2257 		goto drop;
2258 
2259 	/*
2260 	 * At this point, the port either doesn't participate
2261 	 * in spanning tree or it is in the forwarding state.
2262 	 */
2263 
2264 	/*
2265 	 * If the packet is unicast, destined for someone on
2266 	 * "this" side of the bridge, drop it.
2267 	 */
2268 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2269 		dst_if = bridge_rtlookup(sc, dst, vlan);
2270 		if (src_if == dst_if)
2271 			goto drop;
2272 	} else {
2273 		/*
2274 		 * Check if its a reserved multicast address, any address
2275 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2276 		 * bridge.
2277 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2278 		 */
2279 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2280 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2281 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2282 			goto drop;
2283 
2284 		/* ...forward it to all interfaces. */
2285 		if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2286 		dst_if = NULL;
2287 	}
2288 
2289 	/*
2290 	 * If we have a destination interface which is a member of our bridge,
2291 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2292 	 * For broadcast or multicast packets, don't bother because it will
2293 	 * be reinjected into ether_input. We do this before we pass the packets
2294 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2295 	 * drop the packet, or possibly modify it, making it difficult to debug
2296 	 * firewall issues on the bridge.
2297 	 */
2298 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2299 		ETHER_BPF_MTAP(ifp, m);
2300 
2301 	/* run the packet filter */
2302 	if (PFIL_HOOKED_IN(V_inet_pfil_head)
2303 #ifdef INET6
2304 	    || PFIL_HOOKED_IN(V_inet6_pfil_head)
2305 #endif
2306 	    ) {
2307 		BRIDGE_UNLOCK(sc);
2308 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2309 			return;
2310 		if (m == NULL)
2311 			return;
2312 		BRIDGE_LOCK(sc);
2313 	}
2314 
2315 	if (dst_if == NULL) {
2316 		bridge_broadcast(sc, src_if, m, 1);
2317 		return;
2318 	}
2319 
2320 	/*
2321 	 * At this point, we're dealing with a unicast frame
2322 	 * going to a different interface.
2323 	 */
2324 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2325 		goto drop;
2326 
2327 	dbif = bridge_lookup_member_if(sc, dst_if);
2328 	if (dbif == NULL)
2329 		/* Not a member of the bridge (anymore?) */
2330 		goto drop;
2331 
2332 	/* Private segments can not talk to each other */
2333 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2334 		goto drop;
2335 
2336 	if ((dbif->bif_flags & IFBIF_STP) &&
2337 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2338 		goto drop;
2339 
2340 	BRIDGE_UNLOCK(sc);
2341 
2342 	if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2343 #ifdef INET6
2344 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2345 #endif
2346 	    ) {
2347 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2348 			return;
2349 		if (m == NULL)
2350 			return;
2351 	}
2352 
2353 	bridge_enqueue(sc, dst_if, m);
2354 	return;
2355 
2356 drop:
2357 	BRIDGE_UNLOCK(sc);
2358 	m_freem(m);
2359 }
2360 
2361 /*
2362  * bridge_input:
2363  *
2364  *	Receive input from a member interface.  Queue the packet for
2365  *	bridging if it is not for us.
2366  */
2367 static struct mbuf *
2368 bridge_input(struct ifnet *ifp, struct mbuf *m)
2369 {
2370 	struct bridge_softc *sc = ifp->if_bridge;
2371 	struct bridge_iflist *bif, *bif2;
2372 	struct ifnet *bifp;
2373 	struct ether_header *eh;
2374 	struct mbuf *mc, *mc2;
2375 	uint16_t vlan;
2376 	int error;
2377 
2378 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2379 		return (m);
2380 
2381 	bifp = sc->sc_ifp;
2382 	vlan = VLANTAGOF(m);
2383 
2384 	/*
2385 	 * Implement support for bridge monitoring. If this flag has been
2386 	 * set on this interface, discard the packet once we push it through
2387 	 * the bpf(4) machinery, but before we do, increment the byte and
2388 	 * packet counters associated with this interface.
2389 	 */
2390 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2391 		m->m_pkthdr.rcvif  = bifp;
2392 		ETHER_BPF_MTAP(bifp, m);
2393 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2394 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2395 		m_freem(m);
2396 		return (NULL);
2397 	}
2398 	BRIDGE_LOCK(sc);
2399 	bif = bridge_lookup_member_if(sc, ifp);
2400 	if (bif == NULL) {
2401 		BRIDGE_UNLOCK(sc);
2402 		return (m);
2403 	}
2404 
2405 	eh = mtod(m, struct ether_header *);
2406 
2407 	bridge_span(sc, m);
2408 
2409 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2410 		/* Tap off 802.1D packets; they do not get forwarded. */
2411 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2412 		    ETHER_ADDR_LEN) == 0) {
2413 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2414 			BRIDGE_UNLOCK(sc);
2415 			return (NULL);
2416 		}
2417 
2418 		if ((bif->bif_flags & IFBIF_STP) &&
2419 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2420 			BRIDGE_UNLOCK(sc);
2421 			return (m);
2422 		}
2423 
2424 		/*
2425 		 * Make a deep copy of the packet and enqueue the copy
2426 		 * for bridge processing; return the original packet for
2427 		 * local processing.
2428 		 */
2429 		mc = m_dup(m, M_NOWAIT);
2430 		if (mc == NULL) {
2431 			BRIDGE_UNLOCK(sc);
2432 			return (m);
2433 		}
2434 
2435 		/* Perform the bridge forwarding function with the copy. */
2436 		bridge_forward(sc, bif, mc);
2437 
2438 		/*
2439 		 * Reinject the mbuf as arriving on the bridge so we have a
2440 		 * chance at claiming multicast packets. We can not loop back
2441 		 * here from ether_input as a bridge is never a member of a
2442 		 * bridge.
2443 		 */
2444 		KASSERT(bifp->if_bridge == NULL,
2445 		    ("loop created in bridge_input"));
2446 		mc2 = m_dup(m, M_NOWAIT);
2447 		if (mc2 != NULL) {
2448 			/* Keep the layer3 header aligned */
2449 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2450 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2451 		}
2452 		if (mc2 != NULL) {
2453 			mc2->m_pkthdr.rcvif = bifp;
2454 			(*bifp->if_input)(bifp, mc2);
2455 		}
2456 
2457 		/* Return the original packet for local processing. */
2458 		return (m);
2459 	}
2460 
2461 	if ((bif->bif_flags & IFBIF_STP) &&
2462 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2463 		BRIDGE_UNLOCK(sc);
2464 		return (m);
2465 	}
2466 
2467 #if (defined(INET) || defined(INET6))
2468 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2469 	|| ((iface)->if_carp \
2470 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2471 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2472 	|| ((iface)->if_carp \
2473 	    && (*carp_forus_p)((iface), eh->ether_shost))
2474 #else
2475 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2476 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2477 #endif
2478 
2479 #ifdef INET6
2480 #   define OR_PFIL_HOOKED_INET6 \
2481 	|| PFIL_HOOKED_IN(V_inet6_pfil_head)
2482 #else
2483 #   define OR_PFIL_HOOKED_INET6
2484 #endif
2485 
2486 #define GRAB_OUR_PACKETS(iface) \
2487 	if ((iface)->if_type == IFT_GIF) \
2488 		continue; \
2489 	/* It is destined for us. */ \
2490 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2491 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2492 	    ) {								\
2493 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2494 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2495 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2496 			if (error && bif->bif_addrmax) {		\
2497 				BRIDGE_UNLOCK(sc);			\
2498 				m_freem(m);				\
2499 				return (NULL);				\
2500 			}						\
2501 		}							\
2502 		m->m_pkthdr.rcvif = iface;				\
2503 		if ((iface) == ifp) {					\
2504 			/* Skip bridge processing... src == dest */	\
2505 			BRIDGE_UNLOCK(sc);				\
2506 			return (m);					\
2507 		}							\
2508 		/* It's passing over or to the bridge, locally. */	\
2509 		ETHER_BPF_MTAP(bifp, m);				\
2510 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);		\
2511 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2512 		/* Filter on the physical interface. */			\
2513 		if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \
2514 		     OR_PFIL_HOOKED_INET6)) {				\
2515 			if (bridge_pfil(&m, NULL, ifp,			\
2516 			    PFIL_IN) != 0 || m == NULL) {		\
2517 				BRIDGE_UNLOCK(sc);			\
2518 				return (NULL);				\
2519 			}						\
2520 		}							\
2521 		if ((iface) != bifp)					\
2522 			ETHER_BPF_MTAP(iface, m);			\
2523 		BRIDGE_UNLOCK(sc);					\
2524 		return (m);						\
2525 	}								\
2526 									\
2527 	/* We just received a packet that we sent out. */		\
2528 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2529 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2530 	    ) {								\
2531 		BRIDGE_UNLOCK(sc);					\
2532 		m_freem(m);						\
2533 		return (NULL);						\
2534 	}
2535 
2536 	/*
2537 	 * Unicast.  Make sure it's not for the bridge.
2538 	 */
2539 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2540 
2541 	/*
2542 	 * Give a chance for ifp at first priority. This will help when	the
2543 	 * packet comes through the interface like VLAN's with the same MACs
2544 	 * on several interfaces from the same bridge. This also will save
2545 	 * some CPU cycles in case the destination interface and the input
2546 	 * interface (eq ifp) are the same.
2547 	 */
2548 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2549 
2550 	/* Now check the all bridge members. */
2551 	CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2552 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2553 	}
2554 
2555 #undef OR_CARP_CHECK_WE_ARE_DST
2556 #undef OR_CARP_CHECK_WE_ARE_SRC
2557 #undef OR_PFIL_HOOKED_INET6
2558 #undef GRAB_OUR_PACKETS
2559 
2560 	/* Perform the bridge forwarding function. */
2561 	bridge_forward(sc, bif, m);
2562 
2563 	return (NULL);
2564 }
2565 
2566 /*
2567  * bridge_broadcast:
2568  *
2569  *	Send a frame to all interfaces that are members of
2570  *	the bridge, except for the one on which the packet
2571  *	arrived.
2572  *
2573  *	NOTE: Releases the lock on return.
2574  */
2575 static void
2576 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2577     struct mbuf *m, int runfilt)
2578 {
2579 	struct bridge_iflist *dbif, *sbif;
2580 	struct mbuf *mc;
2581 	struct ifnet *dst_if;
2582 	int error = 0, used = 0, i;
2583 
2584 	sbif = bridge_lookup_member_if(sc, src_if);
2585 
2586 	BRIDGE_LOCK2REF(sc, error);
2587 	if (error) {
2588 		m_freem(m);
2589 		return;
2590 	}
2591 
2592 	/* Filter on the bridge interface before broadcasting */
2593 	if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2594 #ifdef INET6
2595 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2596 #endif
2597 	    )) {
2598 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2599 			goto out;
2600 		if (m == NULL)
2601 			goto out;
2602 	}
2603 
2604 	CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2605 		dst_if = dbif->bif_ifp;
2606 		if (dst_if == src_if)
2607 			continue;
2608 
2609 		/* Private segments can not talk to each other */
2610 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2611 			continue;
2612 
2613 		if ((dbif->bif_flags & IFBIF_STP) &&
2614 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2615 			continue;
2616 
2617 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2618 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2619 			continue;
2620 
2621 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2622 			continue;
2623 
2624 		if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2625 			mc = m;
2626 			used = 1;
2627 		} else {
2628 			mc = m_dup(m, M_NOWAIT);
2629 			if (mc == NULL) {
2630 				if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2631 				continue;
2632 			}
2633 		}
2634 
2635 		/*
2636 		 * Filter on the output interface. Pass a NULL bridge interface
2637 		 * pointer so we do not redundantly filter on the bridge for
2638 		 * each interface we broadcast on.
2639 		 */
2640 		if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2641 #ifdef INET6
2642 		    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2643 #endif
2644 		    )) {
2645 			if (used == 0) {
2646 				/* Keep the layer3 header aligned */
2647 				i = min(mc->m_pkthdr.len, max_protohdr);
2648 				mc = m_copyup(mc, i, ETHER_ALIGN);
2649 				if (mc == NULL) {
2650 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2651 					continue;
2652 				}
2653 			}
2654 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2655 				continue;
2656 			if (mc == NULL)
2657 				continue;
2658 		}
2659 
2660 		bridge_enqueue(sc, dst_if, mc);
2661 	}
2662 	if (used == 0)
2663 		m_freem(m);
2664 
2665 out:
2666 	BRIDGE_UNREF(sc);
2667 }
2668 
2669 /*
2670  * bridge_span:
2671  *
2672  *	Duplicate a packet out one or more interfaces that are in span mode,
2673  *	the original mbuf is unmodified.
2674  */
2675 static void
2676 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2677 {
2678 	struct bridge_iflist *bif;
2679 	struct ifnet *dst_if;
2680 	struct mbuf *mc;
2681 
2682 	if (CK_LIST_EMPTY(&sc->sc_spanlist))
2683 		return;
2684 
2685 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2686 		dst_if = bif->bif_ifp;
2687 
2688 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2689 			continue;
2690 
2691 		mc = m_copypacket(m, M_NOWAIT);
2692 		if (mc == NULL) {
2693 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2694 			continue;
2695 		}
2696 
2697 		bridge_enqueue(sc, dst_if, mc);
2698 	}
2699 }
2700 
2701 /*
2702  * bridge_rtupdate:
2703  *
2704  *	Add a bridge routing entry.
2705  */
2706 static int
2707 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2708     struct bridge_iflist *bif, int setflags, uint8_t flags)
2709 {
2710 	struct bridge_rtnode *brt;
2711 	int error;
2712 
2713 	BRIDGE_LOCK_ASSERT(sc);
2714 
2715 	/* Check the source address is valid and not multicast. */
2716 	if (ETHER_IS_MULTICAST(dst) ||
2717 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2718 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2719 		return (EINVAL);
2720 
2721 	/* 802.1p frames map to vlan 1 */
2722 	if (vlan == 0)
2723 		vlan = 1;
2724 
2725 	/*
2726 	 * A route for this destination might already exist.  If so,
2727 	 * update it, otherwise create a new one.
2728 	 */
2729 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2730 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2731 			sc->sc_brtexceeded++;
2732 			return (ENOSPC);
2733 		}
2734 		/* Check per interface address limits (if enabled) */
2735 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2736 			bif->bif_addrexceeded++;
2737 			return (ENOSPC);
2738 		}
2739 
2740 		/*
2741 		 * Allocate a new bridge forwarding node, and
2742 		 * initialize the expiration time and Ethernet
2743 		 * address.
2744 		 */
2745 		brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2746 		if (brt == NULL)
2747 			return (ENOMEM);
2748 
2749 		if (bif->bif_flags & IFBIF_STICKY)
2750 			brt->brt_flags = IFBAF_STICKY;
2751 		else
2752 			brt->brt_flags = IFBAF_DYNAMIC;
2753 
2754 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2755 		brt->brt_vlan = vlan;
2756 
2757 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2758 			uma_zfree(V_bridge_rtnode_zone, brt);
2759 			return (error);
2760 		}
2761 		brt->brt_dst = bif;
2762 		bif->bif_addrcnt++;
2763 	}
2764 
2765 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2766 	    brt->brt_dst != bif) {
2767 		brt->brt_dst->bif_addrcnt--;
2768 		brt->brt_dst = bif;
2769 		brt->brt_dst->bif_addrcnt++;
2770 	}
2771 
2772 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2773 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2774 	if (setflags)
2775 		brt->brt_flags = flags;
2776 
2777 	return (0);
2778 }
2779 
2780 /*
2781  * bridge_rtlookup:
2782  *
2783  *	Lookup the destination interface for an address.
2784  */
2785 static struct ifnet *
2786 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2787 {
2788 	struct bridge_rtnode *brt;
2789 
2790 	BRIDGE_LOCK_ASSERT(sc);
2791 
2792 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2793 		return (NULL);
2794 
2795 	return (brt->brt_ifp);
2796 }
2797 
2798 /*
2799  * bridge_rttrim:
2800  *
2801  *	Trim the routine table so that we have a number
2802  *	of routing entries less than or equal to the
2803  *	maximum number.
2804  */
2805 static void
2806 bridge_rttrim(struct bridge_softc *sc)
2807 {
2808 	struct bridge_rtnode *brt, *nbrt;
2809 
2810 	BRIDGE_LOCK_ASSERT(sc);
2811 
2812 	/* Make sure we actually need to do this. */
2813 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2814 		return;
2815 
2816 	/* Force an aging cycle; this might trim enough addresses. */
2817 	bridge_rtage(sc);
2818 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2819 		return;
2820 
2821 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2822 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2823 			bridge_rtnode_destroy(sc, brt);
2824 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2825 				return;
2826 		}
2827 	}
2828 }
2829 
2830 /*
2831  * bridge_timer:
2832  *
2833  *	Aging timer for the bridge.
2834  */
2835 static void
2836 bridge_timer(void *arg)
2837 {
2838 	struct bridge_softc *sc = arg;
2839 
2840 	BRIDGE_LOCK_ASSERT(sc);
2841 
2842 	/* Destruction of rtnodes requires a proper vnet context */
2843 	CURVNET_SET(sc->sc_ifp->if_vnet);
2844 	bridge_rtage(sc);
2845 
2846 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2847 		callout_reset(&sc->sc_brcallout,
2848 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2849 	CURVNET_RESTORE();
2850 }
2851 
2852 /*
2853  * bridge_rtage:
2854  *
2855  *	Perform an aging cycle.
2856  */
2857 static void
2858 bridge_rtage(struct bridge_softc *sc)
2859 {
2860 	struct bridge_rtnode *brt, *nbrt;
2861 
2862 	BRIDGE_LOCK_ASSERT(sc);
2863 
2864 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2865 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2866 			if (time_uptime >= brt->brt_expire)
2867 				bridge_rtnode_destroy(sc, brt);
2868 		}
2869 	}
2870 }
2871 
2872 /*
2873  * bridge_rtflush:
2874  *
2875  *	Remove all dynamic addresses from the bridge.
2876  */
2877 static void
2878 bridge_rtflush(struct bridge_softc *sc, int full)
2879 {
2880 	struct bridge_rtnode *brt, *nbrt;
2881 
2882 	BRIDGE_LOCK_ASSERT(sc);
2883 
2884 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2885 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2886 			bridge_rtnode_destroy(sc, brt);
2887 	}
2888 }
2889 
2890 /*
2891  * bridge_rtdaddr:
2892  *
2893  *	Remove an address from the table.
2894  */
2895 static int
2896 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2897 {
2898 	struct bridge_rtnode *brt;
2899 	int found = 0;
2900 
2901 	BRIDGE_LOCK_ASSERT(sc);
2902 
2903 	/*
2904 	 * If vlan is zero then we want to delete for all vlans so the lookup
2905 	 * may return more than one.
2906 	 */
2907 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2908 		bridge_rtnode_destroy(sc, brt);
2909 		found = 1;
2910 	}
2911 
2912 	return (found ? 0 : ENOENT);
2913 }
2914 
2915 /*
2916  * bridge_rtdelete:
2917  *
2918  *	Delete routes to a speicifc member interface.
2919  */
2920 static void
2921 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2922 {
2923 	struct bridge_rtnode *brt, *nbrt;
2924 
2925 	BRIDGE_LOCK_ASSERT(sc);
2926 
2927 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2928 		if (brt->brt_ifp == ifp && (full ||
2929 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2930 			bridge_rtnode_destroy(sc, brt);
2931 	}
2932 }
2933 
2934 /*
2935  * bridge_rtable_init:
2936  *
2937  *	Initialize the route table for this bridge.
2938  */
2939 static void
2940 bridge_rtable_init(struct bridge_softc *sc)
2941 {
2942 	int i;
2943 
2944 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2945 	    M_DEVBUF, M_WAITOK);
2946 
2947 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2948 		CK_LIST_INIT(&sc->sc_rthash[i]);
2949 
2950 	sc->sc_rthash_key = arc4random();
2951 	CK_LIST_INIT(&sc->sc_rtlist);
2952 }
2953 
2954 /*
2955  * bridge_rtable_fini:
2956  *
2957  *	Deconstruct the route table for this bridge.
2958  */
2959 static void
2960 bridge_rtable_fini(struct bridge_softc *sc)
2961 {
2962 
2963 	KASSERT(sc->sc_brtcnt == 0,
2964 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2965 	free(sc->sc_rthash, M_DEVBUF);
2966 }
2967 
2968 /*
2969  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2970  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2971  */
2972 #define	mix(a, b, c)							\
2973 do {									\
2974 	a -= b; a -= c; a ^= (c >> 13);					\
2975 	b -= c; b -= a; b ^= (a << 8);					\
2976 	c -= a; c -= b; c ^= (b >> 13);					\
2977 	a -= b; a -= c; a ^= (c >> 12);					\
2978 	b -= c; b -= a; b ^= (a << 16);					\
2979 	c -= a; c -= b; c ^= (b >> 5);					\
2980 	a -= b; a -= c; a ^= (c >> 3);					\
2981 	b -= c; b -= a; b ^= (a << 10);					\
2982 	c -= a; c -= b; c ^= (b >> 15);					\
2983 } while (/*CONSTCOND*/0)
2984 
2985 static __inline uint32_t
2986 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2987 {
2988 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2989 
2990 	b += addr[5] << 8;
2991 	b += addr[4];
2992 	a += addr[3] << 24;
2993 	a += addr[2] << 16;
2994 	a += addr[1] << 8;
2995 	a += addr[0];
2996 
2997 	mix(a, b, c);
2998 
2999 	return (c & BRIDGE_RTHASH_MASK);
3000 }
3001 
3002 #undef mix
3003 
3004 static int
3005 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3006 {
3007 	int i, d;
3008 
3009 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3010 		d = ((int)a[i]) - ((int)b[i]);
3011 	}
3012 
3013 	return (d);
3014 }
3015 
3016 /*
3017  * bridge_rtnode_lookup:
3018  *
3019  *	Look up a bridge route node for the specified destination. Compare the
3020  *	vlan id or if zero then just return the first match.
3021  */
3022 static struct bridge_rtnode *
3023 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3024 {
3025 	struct bridge_rtnode *brt;
3026 	uint32_t hash;
3027 	int dir;
3028 
3029 	BRIDGE_LOCK_ASSERT(sc);
3030 
3031 	hash = bridge_rthash(sc, addr);
3032 	CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3033 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3034 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3035 			return (brt);
3036 		if (dir > 0)
3037 			return (NULL);
3038 	}
3039 
3040 	return (NULL);
3041 }
3042 
3043 /*
3044  * bridge_rtnode_insert:
3045  *
3046  *	Insert the specified bridge node into the route table.  We
3047  *	assume the entry is not already in the table.
3048  */
3049 static int
3050 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3051 {
3052 	struct bridge_rtnode *lbrt;
3053 	uint32_t hash;
3054 	int dir;
3055 
3056 	BRIDGE_LOCK_ASSERT(sc);
3057 
3058 	hash = bridge_rthash(sc, brt->brt_addr);
3059 
3060 	lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3061 	if (lbrt == NULL) {
3062 		CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3063 		goto out;
3064 	}
3065 
3066 	do {
3067 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3068 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3069 			return (EEXIST);
3070 		if (dir > 0) {
3071 			CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3072 			goto out;
3073 		}
3074 		if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3075 			CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3076 			goto out;
3077 		}
3078 		lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3079 	} while (lbrt != NULL);
3080 
3081 #ifdef DIAGNOSTIC
3082 	panic("bridge_rtnode_insert: impossible");
3083 #endif
3084 
3085 out:
3086 	CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3087 	sc->sc_brtcnt++;
3088 
3089 	return (0);
3090 }
3091 
3092 /*
3093  * bridge_rtnode_destroy:
3094  *
3095  *	Destroy a bridge rtnode.
3096  */
3097 static void
3098 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3099 {
3100 	BRIDGE_LOCK_ASSERT(sc);
3101 
3102 	CK_LIST_REMOVE(brt, brt_hash);
3103 
3104 	CK_LIST_REMOVE(brt, brt_list);
3105 	sc->sc_brtcnt--;
3106 	brt->brt_dst->bif_addrcnt--;
3107 	uma_zfree(V_bridge_rtnode_zone, brt);
3108 }
3109 
3110 /*
3111  * bridge_rtable_expire:
3112  *
3113  *	Set the expiry time for all routes on an interface.
3114  */
3115 static void
3116 bridge_rtable_expire(struct ifnet *ifp, int age)
3117 {
3118 	struct bridge_softc *sc = ifp->if_bridge;
3119 	struct bridge_rtnode *brt;
3120 
3121 	CURVNET_SET(ifp->if_vnet);
3122 	BRIDGE_LOCK(sc);
3123 
3124 	/*
3125 	 * If the age is zero then flush, otherwise set all the expiry times to
3126 	 * age for the interface
3127 	 */
3128 	if (age == 0)
3129 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3130 	else {
3131 		CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3132 			/* Cap the expiry time to 'age' */
3133 			if (brt->brt_ifp == ifp &&
3134 			    brt->brt_expire > time_uptime + age &&
3135 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3136 				brt->brt_expire = time_uptime + age;
3137 		}
3138 	}
3139 	BRIDGE_UNLOCK(sc);
3140 	CURVNET_RESTORE();
3141 }
3142 
3143 /*
3144  * bridge_state_change:
3145  *
3146  *	Callback from the bridgestp code when a port changes states.
3147  */
3148 static void
3149 bridge_state_change(struct ifnet *ifp, int state)
3150 {
3151 	struct bridge_softc *sc = ifp->if_bridge;
3152 	static const char *stpstates[] = {
3153 		"disabled",
3154 		"listening",
3155 		"learning",
3156 		"forwarding",
3157 		"blocking",
3158 		"discarding"
3159 	};
3160 
3161 	CURVNET_SET(ifp->if_vnet);
3162 	if (V_log_stp)
3163 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3164 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3165 	CURVNET_RESTORE();
3166 }
3167 
3168 /*
3169  * Send bridge packets through pfil if they are one of the types pfil can deal
3170  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3171  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3172  * that interface.
3173  */
3174 static int
3175 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3176 {
3177 	int snap, error, i, hlen;
3178 	struct ether_header *eh1, eh2;
3179 	struct ip *ip;
3180 	struct llc llc1;
3181 	u_int16_t ether_type;
3182 	pfil_return_t rv;
3183 
3184 	snap = 0;
3185 	error = -1;	/* Default error if not error == 0 */
3186 
3187 #if 0
3188 	/* we may return with the IP fields swapped, ensure its not shared */
3189 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3190 #endif
3191 
3192 	if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3193 		return (0); /* filtering is disabled */
3194 
3195 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3196 	if ((*mp)->m_len < i) {
3197 	    *mp = m_pullup(*mp, i);
3198 	    if (*mp == NULL) {
3199 		printf("%s: m_pullup failed\n", __func__);
3200 		return (-1);
3201 	    }
3202 	}
3203 
3204 	eh1 = mtod(*mp, struct ether_header *);
3205 	ether_type = ntohs(eh1->ether_type);
3206 
3207 	/*
3208 	 * Check for SNAP/LLC.
3209 	 */
3210 	if (ether_type < ETHERMTU) {
3211 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3212 
3213 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3214 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3215 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3216 		    llc2->llc_control == LLC_UI) {
3217 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3218 			snap = 1;
3219 		}
3220 	}
3221 
3222 	/*
3223 	 * If we're trying to filter bridge traffic, don't look at anything
3224 	 * other than IP and ARP traffic.  If the filter doesn't understand
3225 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3226 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3227 	 * but of course we don't have an AppleTalk filter to begin with.
3228 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3229 	 * ARP traffic.)
3230 	 */
3231 	switch (ether_type) {
3232 		case ETHERTYPE_ARP:
3233 		case ETHERTYPE_REVARP:
3234 			if (V_pfil_ipfw_arp == 0)
3235 				return (0); /* Automatically pass */
3236 			break;
3237 
3238 		case ETHERTYPE_IP:
3239 #ifdef INET6
3240 		case ETHERTYPE_IPV6:
3241 #endif /* INET6 */
3242 			break;
3243 		default:
3244 			/*
3245 			 * Check to see if the user wants to pass non-ip
3246 			 * packets, these will not be checked by pfil(9) and
3247 			 * passed unconditionally so the default is to drop.
3248 			 */
3249 			if (V_pfil_onlyip)
3250 				goto bad;
3251 	}
3252 
3253 	/* Run the packet through pfil before stripping link headers */
3254 	if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3255 	    dir == PFIL_OUT && ifp != NULL) {
3256 		switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) {
3257 		case PFIL_DROPPED:
3258 			return (EACCES);
3259 		case PFIL_CONSUMED:
3260 			return (0);
3261 		}
3262 	}
3263 
3264 	/* Strip off the Ethernet header and keep a copy. */
3265 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3266 	m_adj(*mp, ETHER_HDR_LEN);
3267 
3268 	/* Strip off snap header, if present */
3269 	if (snap) {
3270 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3271 		m_adj(*mp, sizeof(struct llc));
3272 	}
3273 
3274 	/*
3275 	 * Check the IP header for alignment and errors
3276 	 */
3277 	if (dir == PFIL_IN) {
3278 		switch (ether_type) {
3279 			case ETHERTYPE_IP:
3280 				error = bridge_ip_checkbasic(mp);
3281 				break;
3282 #ifdef INET6
3283 			case ETHERTYPE_IPV6:
3284 				error = bridge_ip6_checkbasic(mp);
3285 				break;
3286 #endif /* INET6 */
3287 			default:
3288 				error = 0;
3289 		}
3290 		if (error)
3291 			goto bad;
3292 	}
3293 
3294 	error = 0;
3295 
3296 	/*
3297 	 * Run the packet through pfil
3298 	 */
3299 	rv = PFIL_PASS;
3300 	switch (ether_type) {
3301 	case ETHERTYPE_IP:
3302 		/*
3303 		 * Run pfil on the member interface and the bridge, both can
3304 		 * be skipped by clearing pfil_member or pfil_bridge.
3305 		 *
3306 		 * Keep the order:
3307 		 *   in_if -> bridge_if -> out_if
3308 		 */
3309 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3310 		    pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3311 		    PFIL_PASS)
3312 			break;
3313 
3314 		if (V_pfil_member && ifp != NULL && (rv =
3315 		    pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) !=
3316 		    PFIL_PASS)
3317 			break;
3318 
3319 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3320 		    pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3321 		    PFIL_PASS)
3322 			break;
3323 
3324 		/* check if we need to fragment the packet */
3325 		/* bridge_fragment generates a mbuf chain of packets */
3326 		/* that already include eth headers */
3327 		if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3328 			i = (*mp)->m_pkthdr.len;
3329 			if (i > ifp->if_mtu) {
3330 				error = bridge_fragment(ifp, mp, &eh2, snap,
3331 					    &llc1);
3332 				return (error);
3333 			}
3334 		}
3335 
3336 		/* Recalculate the ip checksum. */
3337 		ip = mtod(*mp, struct ip *);
3338 		hlen = ip->ip_hl << 2;
3339 		if (hlen < sizeof(struct ip))
3340 			goto bad;
3341 		if (hlen > (*mp)->m_len) {
3342 			if ((*mp = m_pullup(*mp, hlen)) == NULL)
3343 				goto bad;
3344 			ip = mtod(*mp, struct ip *);
3345 			if (ip == NULL)
3346 				goto bad;
3347 		}
3348 		ip->ip_sum = 0;
3349 		if (hlen == sizeof(struct ip))
3350 			ip->ip_sum = in_cksum_hdr(ip);
3351 		else
3352 			ip->ip_sum = in_cksum(*mp, hlen);
3353 
3354 		break;
3355 #ifdef INET6
3356 	case ETHERTYPE_IPV6:
3357 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3358 		    pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3359 		    PFIL_PASS)
3360 			break;
3361 
3362 		if (V_pfil_member && ifp != NULL && (rv =
3363 		    pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) !=
3364 		    PFIL_PASS)
3365 			break;
3366 
3367 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3368 		    pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3369 		    PFIL_PASS)
3370 			break;
3371 		break;
3372 #endif
3373 	}
3374 
3375 	switch (rv) {
3376 	case PFIL_CONSUMED:
3377 		return (0);
3378 	case PFIL_DROPPED:
3379 		return (EACCES);
3380 	default:
3381 		break;
3382 	}
3383 
3384 	error = -1;
3385 
3386 	/*
3387 	 * Finally, put everything back the way it was and return
3388 	 */
3389 	if (snap) {
3390 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3391 		if (*mp == NULL)
3392 			return (error);
3393 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3394 	}
3395 
3396 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3397 	if (*mp == NULL)
3398 		return (error);
3399 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3400 
3401 	return (0);
3402 
3403 bad:
3404 	m_freem(*mp);
3405 	*mp = NULL;
3406 	return (error);
3407 }
3408 
3409 /*
3410  * Perform basic checks on header size since
3411  * pfil assumes ip_input has already processed
3412  * it for it.  Cut-and-pasted from ip_input.c.
3413  * Given how simple the IPv6 version is,
3414  * does the IPv4 version really need to be
3415  * this complicated?
3416  *
3417  * XXX Should we update ipstat here, or not?
3418  * XXX Right now we update ipstat but not
3419  * XXX csum_counter.
3420  */
3421 static int
3422 bridge_ip_checkbasic(struct mbuf **mp)
3423 {
3424 	struct mbuf *m = *mp;
3425 	struct ip *ip;
3426 	int len, hlen;
3427 	u_short sum;
3428 
3429 	if (*mp == NULL)
3430 		return (-1);
3431 
3432 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3433 		if ((m = m_copyup(m, sizeof(struct ip),
3434 			(max_linkhdr + 3) & ~3)) == NULL) {
3435 			/* XXXJRT new stat, please */
3436 			KMOD_IPSTAT_INC(ips_toosmall);
3437 			goto bad;
3438 		}
3439 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3440 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3441 			KMOD_IPSTAT_INC(ips_toosmall);
3442 			goto bad;
3443 		}
3444 	}
3445 	ip = mtod(m, struct ip *);
3446 	if (ip == NULL) goto bad;
3447 
3448 	if (ip->ip_v != IPVERSION) {
3449 		KMOD_IPSTAT_INC(ips_badvers);
3450 		goto bad;
3451 	}
3452 	hlen = ip->ip_hl << 2;
3453 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3454 		KMOD_IPSTAT_INC(ips_badhlen);
3455 		goto bad;
3456 	}
3457 	if (hlen > m->m_len) {
3458 		if ((m = m_pullup(m, hlen)) == NULL) {
3459 			KMOD_IPSTAT_INC(ips_badhlen);
3460 			goto bad;
3461 		}
3462 		ip = mtod(m, struct ip *);
3463 		if (ip == NULL) goto bad;
3464 	}
3465 
3466 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3467 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3468 	} else {
3469 		if (hlen == sizeof(struct ip)) {
3470 			sum = in_cksum_hdr(ip);
3471 		} else {
3472 			sum = in_cksum(m, hlen);
3473 		}
3474 	}
3475 	if (sum) {
3476 		KMOD_IPSTAT_INC(ips_badsum);
3477 		goto bad;
3478 	}
3479 
3480 	/* Retrieve the packet length. */
3481 	len = ntohs(ip->ip_len);
3482 
3483 	/*
3484 	 * Check for additional length bogosity
3485 	 */
3486 	if (len < hlen) {
3487 		KMOD_IPSTAT_INC(ips_badlen);
3488 		goto bad;
3489 	}
3490 
3491 	/*
3492 	 * Check that the amount of data in the buffers
3493 	 * is as at least much as the IP header would have us expect.
3494 	 * Drop packet if shorter than we expect.
3495 	 */
3496 	if (m->m_pkthdr.len < len) {
3497 		KMOD_IPSTAT_INC(ips_tooshort);
3498 		goto bad;
3499 	}
3500 
3501 	/* Checks out, proceed */
3502 	*mp = m;
3503 	return (0);
3504 
3505 bad:
3506 	*mp = m;
3507 	return (-1);
3508 }
3509 
3510 #ifdef INET6
3511 /*
3512  * Same as above, but for IPv6.
3513  * Cut-and-pasted from ip6_input.c.
3514  * XXX Should we update ip6stat, or not?
3515  */
3516 static int
3517 bridge_ip6_checkbasic(struct mbuf **mp)
3518 {
3519 	struct mbuf *m = *mp;
3520 	struct ip6_hdr *ip6;
3521 
3522 	/*
3523 	 * If the IPv6 header is not aligned, slurp it up into a new
3524 	 * mbuf with space for link headers, in the event we forward
3525 	 * it.  Otherwise, if it is aligned, make sure the entire base
3526 	 * IPv6 header is in the first mbuf of the chain.
3527 	 */
3528 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3529 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3530 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3531 			    (max_linkhdr + 3) & ~3)) == NULL) {
3532 			/* XXXJRT new stat, please */
3533 			IP6STAT_INC(ip6s_toosmall);
3534 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3535 			goto bad;
3536 		}
3537 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3538 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3539 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3540 			IP6STAT_INC(ip6s_toosmall);
3541 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3542 			goto bad;
3543 		}
3544 	}
3545 
3546 	ip6 = mtod(m, struct ip6_hdr *);
3547 
3548 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3549 		IP6STAT_INC(ip6s_badvers);
3550 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3551 		goto bad;
3552 	}
3553 
3554 	/* Checks out, proceed */
3555 	*mp = m;
3556 	return (0);
3557 
3558 bad:
3559 	*mp = m;
3560 	return (-1);
3561 }
3562 #endif /* INET6 */
3563 
3564 /*
3565  * bridge_fragment:
3566  *
3567  *	Fragment mbuf chain in multiple packets and prepend ethernet header.
3568  */
3569 static int
3570 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3571     int snap, struct llc *llc)
3572 {
3573 	struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3574 	struct ip *ip;
3575 	int error = -1;
3576 
3577 	if (m->m_len < sizeof(struct ip) &&
3578 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3579 		goto dropit;
3580 	ip = mtod(m, struct ip *);
3581 
3582 	m->m_pkthdr.csum_flags |= CSUM_IP;
3583 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3584 	if (error)
3585 		goto dropit;
3586 
3587 	/*
3588 	 * Walk the chain and re-add the Ethernet header for
3589 	 * each mbuf packet.
3590 	 */
3591 	for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3592 		nextpkt = mcur->m_nextpkt;
3593 		mcur->m_nextpkt = NULL;
3594 		if (snap) {
3595 			M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3596 			if (mcur == NULL) {
3597 				error = ENOBUFS;
3598 				if (mprev != NULL)
3599 					mprev->m_nextpkt = nextpkt;
3600 				goto dropit;
3601 			}
3602 			bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3603 		}
3604 
3605 		M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3606 		if (mcur == NULL) {
3607 			error = ENOBUFS;
3608 			if (mprev != NULL)
3609 				mprev->m_nextpkt = nextpkt;
3610 			goto dropit;
3611 		}
3612 		bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3613 
3614 		/*
3615 		 * The previous two M_PREPEND could have inserted one or two
3616 		 * mbufs in front so we have to update the previous packet's
3617 		 * m_nextpkt.
3618 		 */
3619 		mcur->m_nextpkt = nextpkt;
3620 		if (mprev != NULL)
3621 			mprev->m_nextpkt = mcur;
3622 		else {
3623 			/* The first mbuf in the original chain needs to be
3624 			 * updated. */
3625 			*mp = mcur;
3626 		}
3627 		mprev = mcur;
3628 	}
3629 
3630 	KMOD_IPSTAT_INC(ips_fragmented);
3631 	return (error);
3632 
3633 dropit:
3634 	for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3635 		m = mcur->m_nextpkt;
3636 		m_freem(mcur);
3637 	}
3638 	return (error);
3639 }
3640 
3641 static void
3642 bridge_linkstate(struct ifnet *ifp)
3643 {
3644 	struct bridge_softc *sc = ifp->if_bridge;
3645 	struct bridge_iflist *bif;
3646 
3647 	BRIDGE_LOCK(sc);
3648 	bif = bridge_lookup_member_if(sc, ifp);
3649 	if (bif == NULL) {
3650 		BRIDGE_UNLOCK(sc);
3651 		return;
3652 	}
3653 	bridge_linkcheck(sc);
3654 	BRIDGE_UNLOCK(sc);
3655 
3656 	bstp_linkstate(&bif->bif_stp);
3657 }
3658 
3659 static void
3660 bridge_linkcheck(struct bridge_softc *sc)
3661 {
3662 	struct bridge_iflist *bif;
3663 	int new_link, hasls;
3664 
3665 	BRIDGE_LOCK_ASSERT(sc);
3666 	new_link = LINK_STATE_DOWN;
3667 	hasls = 0;
3668 	/* Our link is considered up if at least one of our ports is active */
3669 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3670 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3671 			hasls++;
3672 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3673 			new_link = LINK_STATE_UP;
3674 			break;
3675 		}
3676 	}
3677 	if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3678 		/* If no interfaces support link-state then we default to up */
3679 		new_link = LINK_STATE_UP;
3680 	}
3681 	if_link_state_change(sc->sc_ifp, new_link);
3682 }
3683