xref: /freebsd/sys/net/if_bridge.c (revision 1dfcff294e44d4b45813288ef4095c36abb22f0e)
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  * The bridge relies heavily on the epoch(9) system to protect its data
191  * structures. This means we can safely use CK_LISTs while in NET_EPOCH, but we
192  * must ensure there is only one writer at a time.
193  *
194  * That is: for read accesses we only need to be in NET_EPOCH, but for write
195  * accesses we must hold:
196  *
197  *  - BRIDGE_RT_LOCK, for any change to bridge_rtnodes
198  *  - BRIDGE_LOCK, for any other change
199  *
200  * The BRIDGE_LOCK is a sleepable lock, because it is held across ioctl()
201  * calls to bridge member interfaces and these ioctl()s can sleep.
202  * The BRIDGE_RT_LOCK is a non-sleepable mutex, because it is sometimes
203  * required while we're in NET_EPOCH and then we're not allowed to sleep.
204  */
205 #define BRIDGE_LOCK_INIT(_sc)		do {			\
206 	sx_init(&(_sc)->sc_sx, "if_bridge");			\
207 	mtx_init(&(_sc)->sc_rt_mtx, "if_bridge rt", NULL, MTX_DEF);	\
208 } while (0)
209 #define BRIDGE_LOCK_DESTROY(_sc)	do {	\
210 	sx_destroy(&(_sc)->sc_sx);		\
211 	mtx_destroy(&(_sc)->sc_rt_mtx);		\
212 } while (0)
213 #define BRIDGE_LOCK(_sc)		sx_xlock(&(_sc)->sc_sx)
214 #define BRIDGE_UNLOCK(_sc)		sx_xunlock(&(_sc)->sc_sx)
215 #define BRIDGE_LOCK_ASSERT(_sc)		sx_assert(&(_sc)->sc_sx, SX_XLOCKED)
216 #define BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(_sc)	\
217 	    MPASS(in_epoch(net_epoch_preempt) || sx_xlocked(&(_sc)->sc_sx))
218 #define BRIDGE_UNLOCK_ASSERT(_sc)	sx_assert(&(_sc)->sc_sx, SX_UNLOCKED)
219 #define BRIDGE_RT_LOCK(_sc)		mtx_lock(&(_sc)->sc_rt_mtx)
220 #define BRIDGE_RT_UNLOCK(_sc)		mtx_unlock(&(_sc)->sc_rt_mtx)
221 #define BRIDGE_RT_LOCK_ASSERT(_sc)	mtx_assert(&(_sc)->sc_rt_mtx, MA_OWNED)
222 #define BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(_sc)	\
223 	    MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(_sc)->sc_rt_mtx))
224 
225 /*
226  * Bridge interface list entry.
227  */
228 struct bridge_iflist {
229 	CK_LIST_ENTRY(bridge_iflist) bif_next;
230 	struct ifnet		*bif_ifp;	/* member if */
231 	struct bstp_port	bif_stp;	/* STP state */
232 	uint32_t		bif_flags;	/* member if flags */
233 	int			bif_savedcaps;	/* saved capabilities */
234 	uint32_t		bif_addrmax;	/* max # of addresses */
235 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
236 	uint32_t		bif_addrexceeded;/* # of address violations */
237 	struct epoch_context	bif_epoch_ctx;
238 };
239 
240 /*
241  * Bridge route node.
242  */
243 struct bridge_rtnode {
244 	CK_LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
245 	CK_LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
246 	struct bridge_iflist	*brt_dst;	/* destination if */
247 	unsigned long		brt_expire;	/* expiration time */
248 	uint8_t			brt_flags;	/* address flags */
249 	uint8_t			brt_addr[ETHER_ADDR_LEN];
250 	uint16_t		brt_vlan;	/* vlan id */
251 	struct	vnet		*brt_vnet;
252 	struct	epoch_context	brt_epoch_ctx;
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 sx		sc_sx;
263 	struct mtx		sc_rt_mtx;
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 	CK_LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
269 	CK_LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
270 	CK_LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
271 	uint32_t		sc_rthash_key;	/* key for hash */
272 	CK_LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
273 	struct bstp_state	sc_stp;		/* STP state */
274 	uint32_t		sc_brtexceeded;	/* # of cache drops */
275 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
276 	struct ether_addr	sc_defaddr;	/* Default MAC address */
277 	struct epoch_context	sc_epoch_ctx;
278 };
279 
280 VNET_DEFINE_STATIC(struct sx, bridge_list_sx);
281 #define	V_bridge_list_sx	VNET(bridge_list_sx)
282 static eventhandler_tag bridge_detach_cookie;
283 
284 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
285 
286 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
287 #define	V_bridge_rtnode_zone	VNET(bridge_rtnode_zone)
288 
289 static int	bridge_clone_create(struct if_clone *, int, caddr_t);
290 static void	bridge_clone_destroy(struct ifnet *);
291 
292 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
293 static void	bridge_mutecaps(struct bridge_softc *);
294 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
295 		    int);
296 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
297 static void	bridge_init(void *);
298 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
299 static void	bridge_stop(struct ifnet *, int);
300 static int	bridge_transmit(struct ifnet *, struct mbuf *);
301 #ifdef ALTQ
302 static void	bridge_altq_start(if_t);
303 static int	bridge_altq_transmit(if_t, struct mbuf *);
304 #endif
305 static void	bridge_qflush(struct ifnet *);
306 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
307 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
308 		    struct rtentry *);
309 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
310 		    struct mbuf *);
311 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
312 
313 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
314 		    struct mbuf *m);
315 
316 static void	bridge_timer(void *);
317 
318 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
319 		    struct mbuf *, int);
320 static void	bridge_span(struct bridge_softc *, struct mbuf *);
321 
322 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
323 		    uint16_t, struct bridge_iflist *, int, uint8_t);
324 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
325 		    uint16_t);
326 static void	bridge_rttrim(struct bridge_softc *);
327 static void	bridge_rtage(struct bridge_softc *);
328 static void	bridge_rtflush(struct bridge_softc *, int);
329 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
330 		    uint16_t);
331 
332 static void	bridge_rtable_init(struct bridge_softc *);
333 static void	bridge_rtable_fini(struct bridge_softc *);
334 
335 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
336 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
337 		    const uint8_t *, uint16_t);
338 static int	bridge_rtnode_insert(struct bridge_softc *,
339 		    struct bridge_rtnode *);
340 static void	bridge_rtnode_destroy(struct bridge_softc *,
341 		    struct bridge_rtnode *);
342 static void	bridge_rtable_expire(struct ifnet *, int);
343 static void	bridge_state_change(struct ifnet *, int);
344 
345 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
346 		    const char *name);
347 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
348 		    struct ifnet *ifp);
349 static void	bridge_delete_member(struct bridge_softc *,
350 		    struct bridge_iflist *, int);
351 static void	bridge_delete_span(struct bridge_softc *,
352 		    struct bridge_iflist *);
353 
354 static int	bridge_ioctl_add(struct bridge_softc *, void *);
355 static int	bridge_ioctl_del(struct bridge_softc *, void *);
356 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
357 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
358 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
359 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
360 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
361 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
362 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
363 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
364 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
365 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
366 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
367 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
368 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
369 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
370 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
371 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
372 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
373 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
374 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
375 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
376 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
377 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
378 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
379 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
380 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
381 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
382 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
383 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
384 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
385 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
386 		    int);
387 static int	bridge_ip_checkbasic(struct mbuf **mp);
388 #ifdef INET6
389 static int	bridge_ip6_checkbasic(struct mbuf **mp);
390 #endif /* INET6 */
391 static int	bridge_fragment(struct ifnet *, struct mbuf **mp,
392 		    struct ether_header *, int, struct llc *);
393 static void	bridge_linkstate(struct ifnet *ifp);
394 static void	bridge_linkcheck(struct bridge_softc *sc);
395 
396 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
397 #define	VLANTAGOF(_m)	\
398     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
399 
400 static struct bstp_cb_ops bridge_ops = {
401 	.bcb_state = bridge_state_change,
402 	.bcb_rtage = bridge_rtable_expire
403 };
404 
405 SYSCTL_DECL(_net_link);
406 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
407     "Bridge");
408 
409 /* only pass IP[46] packets when pfil is enabled */
410 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
411 #define	V_pfil_onlyip	VNET(pfil_onlyip)
412 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
413     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
414     "Only pass IP packets when pfil is enabled");
415 
416 /* run pfil hooks on the bridge interface */
417 VNET_DEFINE_STATIC(int, pfil_bridge) = 1;
418 #define	V_pfil_bridge	VNET(pfil_bridge)
419 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
420     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
421     "Packet filter on the bridge interface");
422 
423 /* layer2 filter with ipfw */
424 VNET_DEFINE_STATIC(int, pfil_ipfw);
425 #define	V_pfil_ipfw	VNET(pfil_ipfw)
426 
427 /* layer2 ARP filter with ipfw */
428 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
429 #define	V_pfil_ipfw_arp	VNET(pfil_ipfw_arp)
430 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
431     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
432     "Filter ARP packets through IPFW layer2");
433 
434 /* run pfil hooks on the member interface */
435 VNET_DEFINE_STATIC(int, pfil_member) = 1;
436 #define	V_pfil_member	VNET(pfil_member)
437 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
438     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
439     "Packet filter on the member interface");
440 
441 /* run pfil hooks on the physical interface for locally destined packets */
442 VNET_DEFINE_STATIC(int, pfil_local_phys);
443 #define	V_pfil_local_phys	VNET(pfil_local_phys)
444 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
445     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
446     "Packet filter on the physical interface for locally destined packets");
447 
448 /* log STP state changes */
449 VNET_DEFINE_STATIC(int, log_stp);
450 #define	V_log_stp	VNET(log_stp)
451 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
452     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
453     "Log STP state changes");
454 
455 /* share MAC with first bridge member */
456 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
457 #define	V_bridge_inherit_mac	VNET(bridge_inherit_mac)
458 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
459     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
460     "Inherit MAC address from the first bridge member");
461 
462 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
463 #define	V_allow_llz_overlap	VNET(allow_llz_overlap)
464 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
465     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
466     "Allow overlap of link-local scope "
467     "zones of a bridge interface and the member interfaces");
468 
469 struct bridge_control {
470 	int	(*bc_func)(struct bridge_softc *, void *);
471 	int	bc_argsize;
472 	int	bc_flags;
473 };
474 
475 #define	BC_F_COPYIN		0x01	/* copy arguments in */
476 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
477 #define	BC_F_SUSER		0x04	/* do super-user check */
478 
479 const struct bridge_control bridge_control_table[] = {
480 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
481 	  BC_F_COPYIN|BC_F_SUSER },
482 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
483 	  BC_F_COPYIN|BC_F_SUSER },
484 
485 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
486 	  BC_F_COPYIN|BC_F_COPYOUT },
487 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
488 	  BC_F_COPYIN|BC_F_SUSER },
489 
490 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
491 	  BC_F_COPYIN|BC_F_SUSER },
492 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
493 	  BC_F_COPYOUT },
494 
495 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
496 	  BC_F_COPYIN|BC_F_COPYOUT },
497 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
498 	  BC_F_COPYIN|BC_F_COPYOUT },
499 
500 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
501 	  BC_F_COPYIN|BC_F_SUSER },
502 
503 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
504 	  BC_F_COPYIN|BC_F_SUSER },
505 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
506 	  BC_F_COPYOUT },
507 
508 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
509 	  BC_F_COPYIN|BC_F_SUSER },
510 
511 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
512 	  BC_F_COPYIN|BC_F_SUSER },
513 
514 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
515 	  BC_F_COPYOUT },
516 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
517 	  BC_F_COPYIN|BC_F_SUSER },
518 
519 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
520 	  BC_F_COPYOUT },
521 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
522 	  BC_F_COPYIN|BC_F_SUSER },
523 
524 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
525 	  BC_F_COPYOUT },
526 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
527 	  BC_F_COPYIN|BC_F_SUSER },
528 
529 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
530 	  BC_F_COPYOUT },
531 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
532 	  BC_F_COPYIN|BC_F_SUSER },
533 
534 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
535 	  BC_F_COPYIN|BC_F_SUSER },
536 
537 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
538 	  BC_F_COPYIN|BC_F_SUSER },
539 
540 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
541 	  BC_F_COPYIN|BC_F_SUSER },
542 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
543 	  BC_F_COPYIN|BC_F_SUSER },
544 
545 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
546 	  BC_F_COPYOUT },
547 
548 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
549 	  BC_F_COPYOUT },
550 
551 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
552 	  BC_F_COPYIN|BC_F_COPYOUT },
553 
554 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
555 	  BC_F_COPYIN|BC_F_SUSER },
556 
557 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
558 	  BC_F_COPYIN|BC_F_SUSER },
559 
560 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
561 	  BC_F_COPYIN|BC_F_SUSER },
562 
563 };
564 const int bridge_control_table_size = nitems(bridge_control_table);
565 
566 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
567 #define	V_bridge_list	VNET(bridge_list)
568 #define	BRIDGE_LIST_LOCK_INIT(x)	sx_init(&V_bridge_list_sx,	\
569 					    "if_bridge list")
570 #define	BRIDGE_LIST_LOCK_DESTROY(x)	sx_destroy(&V_bridge_list_sx)
571 #define	BRIDGE_LIST_LOCK(x)		sx_xlock(&V_bridge_list_sx)
572 #define	BRIDGE_LIST_UNLOCK(x)		sx_xunlock(&V_bridge_list_sx)
573 
574 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
575 #define	V_bridge_cloner	VNET(bridge_cloner)
576 
577 static const char bridge_name[] = "bridge";
578 
579 static void
580 vnet_bridge_init(const void *unused __unused)
581 {
582 
583 	V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
584 	    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
585 	    UMA_ALIGN_PTR, 0);
586 	BRIDGE_LIST_LOCK_INIT();
587 	LIST_INIT(&V_bridge_list);
588 	V_bridge_cloner = if_clone_simple(bridge_name,
589 	    bridge_clone_create, bridge_clone_destroy, 0);
590 }
591 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
592     vnet_bridge_init, NULL);
593 
594 static void
595 vnet_bridge_uninit(const void *unused __unused)
596 {
597 
598 	if_clone_detach(V_bridge_cloner);
599 	V_bridge_cloner = NULL;
600 	BRIDGE_LIST_LOCK_DESTROY();
601 
602 	/* Callbacks may use the UMA zone. */
603 	epoch_drain_callbacks(net_epoch_preempt);
604 
605 	uma_zdestroy(V_bridge_rtnode_zone);
606 }
607 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
608     vnet_bridge_uninit, NULL);
609 
610 static int
611 bridge_modevent(module_t mod, int type, void *data)
612 {
613 
614 	switch (type) {
615 	case MOD_LOAD:
616 		bridge_dn_p = bridge_dummynet;
617 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
618 		    ifnet_departure_event, bridge_ifdetach, NULL,
619 		    EVENTHANDLER_PRI_ANY);
620 		break;
621 	case MOD_UNLOAD:
622 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
623 		    bridge_detach_cookie);
624 		bridge_dn_p = NULL;
625 		break;
626 	default:
627 		return (EOPNOTSUPP);
628 	}
629 	return (0);
630 }
631 
632 static moduledata_t bridge_mod = {
633 	"if_bridge",
634 	bridge_modevent,
635 	0
636 };
637 
638 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
639 MODULE_VERSION(if_bridge, 1);
640 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
641 
642 /*
643  * handler for net.link.bridge.ipfw
644  */
645 static int
646 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
647 {
648 	int enable = V_pfil_ipfw;
649 	int error;
650 
651 	error = sysctl_handle_int(oidp, &enable, 0, req);
652 	enable &= 1;
653 
654 	if (enable != V_pfil_ipfw) {
655 		V_pfil_ipfw = enable;
656 
657 		/*
658 		 * Disable pfil so that ipfw doesnt run twice, if the user
659 		 * really wants both then they can re-enable pfil_bridge and/or
660 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
661 		 * layer2 type.
662 		 */
663 		if (V_pfil_ipfw) {
664 			V_pfil_onlyip = 0;
665 			V_pfil_bridge = 0;
666 			V_pfil_member = 0;
667 		}
668 	}
669 
670 	return (error);
671 }
672 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
673     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT,
674     &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
675     "Layer2 filter with IPFW");
676 
677 #ifdef VIMAGE
678 static void
679 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg)
680 {
681 	struct bridge_softc *sc = ifp->if_softc;
682 	struct bridge_iflist *bif;
683 
684 	BRIDGE_LOCK(sc);
685 
686 	while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
687 		bridge_delete_member(sc, bif, 0);
688 
689 	while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
690 		bridge_delete_span(sc, bif);
691 	}
692 
693 	BRIDGE_UNLOCK(sc);
694 
695 	ether_reassign(ifp, newvnet, arg);
696 }
697 #endif
698 
699 /*
700  * bridge_clone_create:
701  *
702  *	Create a new bridge instance.
703  */
704 static int
705 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
706 {
707 	struct bridge_softc *sc;
708 	struct ifnet *ifp;
709 
710 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
711 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
712 	if (ifp == NULL) {
713 		free(sc, M_DEVBUF);
714 		return (ENOSPC);
715 	}
716 
717 	BRIDGE_LOCK_INIT(sc);
718 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
719 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
720 
721 	/* Initialize our routing table. */
722 	bridge_rtable_init(sc);
723 
724 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
725 
726 	CK_LIST_INIT(&sc->sc_iflist);
727 	CK_LIST_INIT(&sc->sc_spanlist);
728 
729 	ifp->if_softc = sc;
730 	if_initname(ifp, bridge_name, unit);
731 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
732 	ifp->if_ioctl = bridge_ioctl;
733 #ifdef ALTQ
734 	ifp->if_start = bridge_altq_start;
735 	ifp->if_transmit = bridge_altq_transmit;
736 	IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
737 	ifp->if_snd.ifq_drv_maxlen = 0;
738 	IFQ_SET_READY(&ifp->if_snd);
739 #else
740 	ifp->if_transmit = bridge_transmit;
741 #endif
742 	ifp->if_qflush = bridge_qflush;
743 	ifp->if_init = bridge_init;
744 	ifp->if_type = IFT_BRIDGE;
745 
746 	ether_gen_addr(ifp, &sc->sc_defaddr);
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 #ifdef VIMAGE
754 	ifp->if_reassign = bridge_reassign;
755 #endif
756 
757 	BRIDGE_LIST_LOCK();
758 	LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
759 	BRIDGE_LIST_UNLOCK();
760 
761 	return (0);
762 }
763 
764 static void
765 bridge_clone_destroy_cb(struct epoch_context *ctx)
766 {
767 	struct bridge_softc *sc;
768 
769 	sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
770 
771 	BRIDGE_LOCK_DESTROY(sc);
772 	free(sc, M_DEVBUF);
773 }
774 
775 /*
776  * bridge_clone_destroy:
777  *
778  *	Destroy a bridge instance.
779  */
780 static void
781 bridge_clone_destroy(struct ifnet *ifp)
782 {
783 	struct bridge_softc *sc = ifp->if_softc;
784 	struct bridge_iflist *bif;
785 	struct epoch_tracker et;
786 
787 	BRIDGE_LOCK(sc);
788 
789 	bridge_stop(ifp, 1);
790 	ifp->if_flags &= ~IFF_UP;
791 
792 	while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
793 		bridge_delete_member(sc, bif, 0);
794 
795 	while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
796 		bridge_delete_span(sc, bif);
797 	}
798 
799 	/* Tear down the routing table. */
800 	bridge_rtable_fini(sc);
801 
802 	BRIDGE_UNLOCK(sc);
803 
804 	NET_EPOCH_ENTER(et);
805 
806 	callout_drain(&sc->sc_brcallout);
807 
808 	BRIDGE_LIST_LOCK();
809 	LIST_REMOVE(sc, sc_list);
810 	BRIDGE_LIST_UNLOCK();
811 
812 	bstp_detach(&sc->sc_stp);
813 #ifdef ALTQ
814 	IFQ_PURGE(&ifp->if_snd);
815 #endif
816 	NET_EPOCH_EXIT(et);
817 
818 	ether_ifdetach(ifp);
819 	if_free(ifp);
820 
821 	NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx);
822 }
823 
824 /*
825  * bridge_ioctl:
826  *
827  *	Handle a control request from the operator.
828  */
829 static int
830 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
831 {
832 	struct bridge_softc *sc = ifp->if_softc;
833 	struct ifreq *ifr = (struct ifreq *)data;
834 	struct bridge_iflist *bif;
835 	struct thread *td = curthread;
836 	union {
837 		struct ifbreq ifbreq;
838 		struct ifbifconf ifbifconf;
839 		struct ifbareq ifbareq;
840 		struct ifbaconf ifbaconf;
841 		struct ifbrparam ifbrparam;
842 		struct ifbropreq ifbropreq;
843 	} args;
844 	struct ifdrv *ifd = (struct ifdrv *) data;
845 	const struct bridge_control *bc;
846 	int error = 0, oldmtu;
847 
848 	BRIDGE_LOCK(sc);
849 
850 	switch (cmd) {
851 	case SIOCADDMULTI:
852 	case SIOCDELMULTI:
853 		break;
854 
855 	case SIOCGDRVSPEC:
856 	case SIOCSDRVSPEC:
857 		if (ifd->ifd_cmd >= bridge_control_table_size) {
858 			error = EINVAL;
859 			break;
860 		}
861 		bc = &bridge_control_table[ifd->ifd_cmd];
862 
863 		if (cmd == SIOCGDRVSPEC &&
864 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
865 			error = EINVAL;
866 			break;
867 		}
868 		else if (cmd == SIOCSDRVSPEC &&
869 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
870 			error = EINVAL;
871 			break;
872 		}
873 
874 		if (bc->bc_flags & BC_F_SUSER) {
875 			error = priv_check(td, PRIV_NET_BRIDGE);
876 			if (error)
877 				break;
878 		}
879 
880 		if (ifd->ifd_len != bc->bc_argsize ||
881 		    ifd->ifd_len > sizeof(args)) {
882 			error = EINVAL;
883 			break;
884 		}
885 
886 		bzero(&args, sizeof(args));
887 		if (bc->bc_flags & BC_F_COPYIN) {
888 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
889 			if (error)
890 				break;
891 		}
892 
893 		oldmtu = ifp->if_mtu;
894 		error = (*bc->bc_func)(sc, &args);
895 		if (error)
896 			break;
897 
898 		/*
899 		 * Bridge MTU may change during addition of the first port.
900 		 * If it did, do network layer specific procedure.
901 		 */
902 		if (ifp->if_mtu != oldmtu) {
903 #ifdef INET6
904 			nd6_setmtu(ifp);
905 #endif
906 			rt_updatemtu(ifp);
907 		}
908 
909 		if (bc->bc_flags & BC_F_COPYOUT)
910 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
911 
912 		break;
913 
914 	case SIOCSIFFLAGS:
915 		if (!(ifp->if_flags & IFF_UP) &&
916 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
917 			/*
918 			 * If interface is marked down and it is running,
919 			 * then stop and disable it.
920 			 */
921 			bridge_stop(ifp, 1);
922 		} else if ((ifp->if_flags & IFF_UP) &&
923 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
924 			/*
925 			 * If interface is marked up and it is stopped, then
926 			 * start it.
927 			 */
928 			BRIDGE_UNLOCK(sc);
929 			(*ifp->if_init)(sc);
930 			BRIDGE_LOCK(sc);
931 		}
932 		break;
933 
934 	case SIOCSIFMTU:
935 		oldmtu = sc->sc_ifp->if_mtu;
936 
937 		if (ifr->ifr_mtu < 576) {
938 			error = EINVAL;
939 			break;
940 		}
941 		if (CK_LIST_EMPTY(&sc->sc_iflist)) {
942 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
943 			break;
944 		}
945 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
946 			error = (*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
947 			    SIOCSIFMTU, (caddr_t)ifr);
948 			if (error != 0) {
949 				log(LOG_NOTICE, "%s: invalid MTU: %u for"
950 				    " member %s\n", sc->sc_ifp->if_xname,
951 				    ifr->ifr_mtu,
952 				    bif->bif_ifp->if_xname);
953 				error = EINVAL;
954 				break;
955 			}
956 		}
957 		if (error) {
958 			/* Restore the previous MTU on all member interfaces. */
959 			ifr->ifr_mtu = oldmtu;
960 			CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
961 				(*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
962 				    SIOCSIFMTU, (caddr_t)ifr);
963 			}
964 		} else {
965 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
966 		}
967 		break;
968 	default:
969 		/*
970 		 * drop the lock as ether_ioctl() will call bridge_start() and
971 		 * cause the lock to be recursed.
972 		 */
973 		BRIDGE_UNLOCK(sc);
974 		error = ether_ioctl(ifp, cmd, data);
975 		BRIDGE_LOCK(sc);
976 		break;
977 	}
978 
979 	BRIDGE_UNLOCK(sc);
980 
981 	return (error);
982 }
983 
984 /*
985  * bridge_mutecaps:
986  *
987  *	Clear or restore unwanted capabilities on the member interface
988  */
989 static void
990 bridge_mutecaps(struct bridge_softc *sc)
991 {
992 	struct bridge_iflist *bif;
993 	int enabled, mask;
994 
995 	BRIDGE_LOCK_ASSERT(sc);
996 
997 	/* Initial bitmask of capabilities to test */
998 	mask = BRIDGE_IFCAPS_MASK;
999 
1000 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1001 		/* Every member must support it or its disabled */
1002 		mask &= bif->bif_savedcaps;
1003 	}
1004 
1005 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1006 		enabled = bif->bif_ifp->if_capenable;
1007 		enabled &= ~BRIDGE_IFCAPS_STRIP;
1008 		/* strip off mask bits and enable them again if allowed */
1009 		enabled &= ~BRIDGE_IFCAPS_MASK;
1010 		enabled |= mask;
1011 		bridge_set_ifcap(sc, bif, enabled);
1012 	}
1013 }
1014 
1015 static void
1016 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1017 {
1018 	struct ifnet *ifp = bif->bif_ifp;
1019 	struct ifreq ifr;
1020 	int error, mask, stuck;
1021 
1022 	bzero(&ifr, sizeof(ifr));
1023 	ifr.ifr_reqcap = set;
1024 
1025 	if (ifp->if_capenable != set) {
1026 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1027 		if (error)
1028 			if_printf(sc->sc_ifp,
1029 			    "error setting capabilities on %s: %d\n",
1030 			    ifp->if_xname, error);
1031 		mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1032 		stuck = ifp->if_capenable & mask & ~set;
1033 		if (stuck != 0)
1034 			if_printf(sc->sc_ifp,
1035 			    "can't disable some capabilities on %s: 0x%x\n",
1036 			    ifp->if_xname, stuck);
1037 	}
1038 }
1039 
1040 /*
1041  * bridge_lookup_member:
1042  *
1043  *	Lookup a bridge member interface.
1044  */
1045 static struct bridge_iflist *
1046 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1047 {
1048 	struct bridge_iflist *bif;
1049 	struct ifnet *ifp;
1050 
1051 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1052 
1053 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1054 		ifp = bif->bif_ifp;
1055 		if (strcmp(ifp->if_xname, name) == 0)
1056 			return (bif);
1057 	}
1058 
1059 	return (NULL);
1060 }
1061 
1062 /*
1063  * bridge_lookup_member_if:
1064  *
1065  *	Lookup a bridge member interface by ifnet*.
1066  */
1067 static struct bridge_iflist *
1068 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1069 {
1070 	struct bridge_iflist *bif;
1071 
1072 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1073 
1074 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1075 		if (bif->bif_ifp == member_ifp)
1076 			return (bif);
1077 	}
1078 
1079 	return (NULL);
1080 }
1081 
1082 static void
1083 bridge_delete_member_cb(struct epoch_context *ctx)
1084 {
1085 	struct bridge_iflist *bif;
1086 
1087 	bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1088 
1089 	free(bif, M_DEVBUF);
1090 }
1091 
1092 /*
1093  * bridge_delete_member:
1094  *
1095  *	Delete the specified member interface.
1096  */
1097 static void
1098 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1099     int gone)
1100 {
1101 	struct ifnet *ifs = bif->bif_ifp;
1102 	struct ifnet *fif = NULL;
1103 	struct bridge_iflist *bifl;
1104 
1105 	BRIDGE_LOCK_ASSERT(sc);
1106 
1107 	if (bif->bif_flags & IFBIF_STP)
1108 		bstp_disable(&bif->bif_stp);
1109 
1110 	ifs->if_bridge = NULL;
1111 	CK_LIST_REMOVE(bif, bif_next);
1112 
1113 	/*
1114 	 * If removing the interface that gave the bridge its mac address, set
1115 	 * the mac address of the bridge to the address of the next member, or
1116 	 * to its default address if no members are left.
1117 	 */
1118 	if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1119 		if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1120 			bcopy(&sc->sc_defaddr,
1121 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1122 			sc->sc_ifaddr = NULL;
1123 		} else {
1124 			bifl = CK_LIST_FIRST(&sc->sc_iflist);
1125 			fif = bifl->bif_ifp;
1126 			bcopy(IF_LLADDR(fif),
1127 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1128 			sc->sc_ifaddr = fif;
1129 		}
1130 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1131 	}
1132 
1133 	bridge_linkcheck(sc);
1134 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
1135 	BRIDGE_RT_LOCK(sc);
1136 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1137 	BRIDGE_RT_UNLOCK(sc);
1138 	KASSERT(bif->bif_addrcnt == 0,
1139 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1140 
1141 	ifs->if_bridge_output = NULL;
1142 	ifs->if_bridge_input = NULL;
1143 	ifs->if_bridge_linkstate = NULL;
1144 	if (!gone) {
1145 		switch (ifs->if_type) {
1146 		case IFT_ETHER:
1147 		case IFT_L2VLAN:
1148 			/*
1149 			 * Take the interface out of promiscuous mode, but only
1150 			 * if it was promiscuous in the first place. It might
1151 			 * not be if we're in the bridge_ioctl_add() error path.
1152 			 */
1153 			if (ifs->if_flags & IFF_PROMISC)
1154 				(void) ifpromisc(ifs, 0);
1155 			break;
1156 
1157 		case IFT_GIF:
1158 			break;
1159 
1160 		default:
1161 #ifdef DIAGNOSTIC
1162 			panic("bridge_delete_member: impossible");
1163 #endif
1164 			break;
1165 		}
1166 		/* reneable any interface capabilities */
1167 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1168 	}
1169 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1170 
1171 	NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1172 }
1173 
1174 /*
1175  * bridge_delete_span:
1176  *
1177  *	Delete the specified span interface.
1178  */
1179 static void
1180 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1181 {
1182 	BRIDGE_LOCK_ASSERT(sc);
1183 
1184 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1185 	    ("%s: not a span interface", __func__));
1186 
1187 	CK_LIST_REMOVE(bif, bif_next);
1188 
1189 	NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1190 }
1191 
1192 static int
1193 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1194 {
1195 	struct ifbreq *req = arg;
1196 	struct bridge_iflist *bif = NULL;
1197 	struct ifnet *ifs;
1198 	int error = 0;
1199 
1200 	ifs = ifunit(req->ifbr_ifsname);
1201 	if (ifs == NULL)
1202 		return (ENOENT);
1203 	if (ifs->if_ioctl == NULL)	/* must be supported */
1204 		return (EINVAL);
1205 
1206 	/* If it's in the span list, it can't be a member. */
1207 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1208 		if (ifs == bif->bif_ifp)
1209 			return (EBUSY);
1210 
1211 	if (ifs->if_bridge == sc)
1212 		return (EEXIST);
1213 
1214 	if (ifs->if_bridge != NULL)
1215 		return (EBUSY);
1216 
1217 	switch (ifs->if_type) {
1218 	case IFT_ETHER:
1219 	case IFT_L2VLAN:
1220 	case IFT_GIF:
1221 		/* permitted interface types */
1222 		break;
1223 	default:
1224 		return (EINVAL);
1225 	}
1226 
1227 #ifdef INET6
1228 	/*
1229 	 * Two valid inet6 addresses with link-local scope must not be
1230 	 * on the parent interface and the member interfaces at the
1231 	 * same time.  This restriction is needed to prevent violation
1232 	 * of link-local scope zone.  Attempts to add a member
1233 	 * interface which has inet6 addresses when the parent has
1234 	 * inet6 triggers removal of all inet6 addresses on the member
1235 	 * interface.
1236 	 */
1237 
1238 	/* Check if the parent interface has a link-local scope addr. */
1239 	if (V_allow_llz_overlap == 0 &&
1240 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1241 		/*
1242 		 * If any, remove all inet6 addresses from the member
1243 		 * interfaces.
1244 		 */
1245 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1246  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1247 				in6_ifdetach(bif->bif_ifp);
1248 				if_printf(sc->sc_ifp,
1249 				    "IPv6 addresses on %s have been removed "
1250 				    "before adding it as a member to prevent "
1251 				    "IPv6 address scope violation.\n",
1252 				    bif->bif_ifp->if_xname);
1253 			}
1254 		}
1255 		if (in6ifa_llaonifp(ifs)) {
1256 			in6_ifdetach(ifs);
1257 			if_printf(sc->sc_ifp,
1258 			    "IPv6 addresses on %s have been removed "
1259 			    "before adding it as a member to prevent "
1260 			    "IPv6 address scope violation.\n",
1261 			    ifs->if_xname);
1262 		}
1263 	}
1264 #endif
1265 	/* Allow the first Ethernet member to define the MTU */
1266 	if (CK_LIST_EMPTY(&sc->sc_iflist))
1267 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1268 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1269 		struct ifreq ifr;
1270 
1271 		snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s",
1272 		    ifs->if_xname);
1273 		ifr.ifr_mtu = sc->sc_ifp->if_mtu;
1274 
1275 		error = (*ifs->if_ioctl)(ifs,
1276 		    SIOCSIFMTU, (caddr_t)&ifr);
1277 		if (error != 0) {
1278 			log(LOG_NOTICE, "%s: invalid MTU: %u for"
1279 			    " new member %s\n", sc->sc_ifp->if_xname,
1280 			    ifr.ifr_mtu,
1281 			    ifs->if_xname);
1282 			return (EINVAL);
1283 		}
1284 	}
1285 
1286 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1287 	if (bif == NULL)
1288 		return (ENOMEM);
1289 
1290 	bif->bif_ifp = ifs;
1291 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1292 	bif->bif_savedcaps = ifs->if_capenable;
1293 
1294 	/*
1295 	 * Assign the interface's MAC address to the bridge if it's the first
1296 	 * member and the MAC address of the bridge has not been changed from
1297 	 * the default randomly generated one.
1298 	 */
1299 	if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1300 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1301 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1302 		sc->sc_ifaddr = ifs;
1303 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1304 	}
1305 
1306 	ifs->if_bridge = sc;
1307 	ifs->if_bridge_output = bridge_output;
1308 	ifs->if_bridge_input = bridge_input;
1309 	ifs->if_bridge_linkstate = bridge_linkstate;
1310 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1311 	/*
1312 	 * XXX: XLOCK HERE!?!
1313 	 *
1314 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1315 	 */
1316 	CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1317 
1318 	/* Set interface capabilities to the intersection set of all members */
1319 	bridge_mutecaps(sc);
1320 	bridge_linkcheck(sc);
1321 
1322 	/* Place the interface into promiscuous mode */
1323 	switch (ifs->if_type) {
1324 		case IFT_ETHER:
1325 		case IFT_L2VLAN:
1326 			error = ifpromisc(ifs, 1);
1327 			break;
1328 	}
1329 
1330 	if (error)
1331 		bridge_delete_member(sc, bif, 0);
1332 	return (error);
1333 }
1334 
1335 static int
1336 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1337 {
1338 	struct ifbreq *req = arg;
1339 	struct bridge_iflist *bif;
1340 
1341 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1342 	if (bif == NULL)
1343 		return (ENOENT);
1344 
1345 	bridge_delete_member(sc, bif, 0);
1346 
1347 	return (0);
1348 }
1349 
1350 static int
1351 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1352 {
1353 	struct ifbreq *req = arg;
1354 	struct bridge_iflist *bif;
1355 	struct bstp_port *bp;
1356 
1357 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1358 	if (bif == NULL)
1359 		return (ENOENT);
1360 
1361 	bp = &bif->bif_stp;
1362 	req->ifbr_ifsflags = bif->bif_flags;
1363 	req->ifbr_state = bp->bp_state;
1364 	req->ifbr_priority = bp->bp_priority;
1365 	req->ifbr_path_cost = bp->bp_path_cost;
1366 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1367 	req->ifbr_proto = bp->bp_protover;
1368 	req->ifbr_role = bp->bp_role;
1369 	req->ifbr_stpflags = bp->bp_flags;
1370 	req->ifbr_addrcnt = bif->bif_addrcnt;
1371 	req->ifbr_addrmax = bif->bif_addrmax;
1372 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1373 
1374 	/* Copy STP state options as flags */
1375 	if (bp->bp_operedge)
1376 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1377 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1378 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1379 	if (bp->bp_ptp_link)
1380 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1381 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1382 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1383 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1384 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1385 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1386 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1387 	return (0);
1388 }
1389 
1390 static int
1391 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1392 {
1393 	struct epoch_tracker et;
1394 	struct ifbreq *req = arg;
1395 	struct bridge_iflist *bif;
1396 	struct bstp_port *bp;
1397 	int error;
1398 
1399 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1400 	if (bif == NULL)
1401 		return (ENOENT);
1402 	bp = &bif->bif_stp;
1403 
1404 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1405 		/* SPAN is readonly */
1406 		return (EINVAL);
1407 
1408 	NET_EPOCH_ENTER(et);
1409 
1410 	if (req->ifbr_ifsflags & IFBIF_STP) {
1411 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1412 			error = bstp_enable(&bif->bif_stp);
1413 			if (error) {
1414 				NET_EPOCH_EXIT(et);
1415 				return (error);
1416 			}
1417 		}
1418 	} else {
1419 		if ((bif->bif_flags & IFBIF_STP) != 0)
1420 			bstp_disable(&bif->bif_stp);
1421 	}
1422 
1423 	/* Pass on STP flags */
1424 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1425 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1426 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1427 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1428 
1429 	/* Save the bits relating to the bridge */
1430 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1431 
1432 	NET_EPOCH_EXIT(et);
1433 
1434 	return (0);
1435 }
1436 
1437 static int
1438 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1439 {
1440 	struct ifbrparam *param = arg;
1441 
1442 	sc->sc_brtmax = param->ifbrp_csize;
1443 	bridge_rttrim(sc);
1444 
1445 	return (0);
1446 }
1447 
1448 static int
1449 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1450 {
1451 	struct ifbrparam *param = arg;
1452 
1453 	param->ifbrp_csize = sc->sc_brtmax;
1454 
1455 	return (0);
1456 }
1457 
1458 static int
1459 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1460 {
1461 	struct ifbifconf *bifc = arg;
1462 	struct bridge_iflist *bif;
1463 	struct ifbreq breq;
1464 	char *buf, *outbuf;
1465 	int count, buflen, len, error = 0;
1466 
1467 	count = 0;
1468 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1469 		count++;
1470 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1471 		count++;
1472 
1473 	buflen = sizeof(breq) * count;
1474 	if (bifc->ifbic_len == 0) {
1475 		bifc->ifbic_len = buflen;
1476 		return (0);
1477 	}
1478 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1479 	if (outbuf == NULL)
1480 		return (ENOMEM);
1481 
1482 	count = 0;
1483 	buf = outbuf;
1484 	len = min(bifc->ifbic_len, buflen);
1485 	bzero(&breq, sizeof(breq));
1486 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1487 		if (len < sizeof(breq))
1488 			break;
1489 
1490 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1491 		    sizeof(breq.ifbr_ifsname));
1492 		/* Fill in the ifbreq structure */
1493 		error = bridge_ioctl_gifflags(sc, &breq);
1494 		if (error)
1495 			break;
1496 		memcpy(buf, &breq, sizeof(breq));
1497 		count++;
1498 		buf += sizeof(breq);
1499 		len -= sizeof(breq);
1500 	}
1501 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1502 		if (len < sizeof(breq))
1503 			break;
1504 
1505 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1506 		    sizeof(breq.ifbr_ifsname));
1507 		breq.ifbr_ifsflags = bif->bif_flags;
1508 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1509 		memcpy(buf, &breq, sizeof(breq));
1510 		count++;
1511 		buf += sizeof(breq);
1512 		len -= sizeof(breq);
1513 	}
1514 
1515 	bifc->ifbic_len = sizeof(breq) * count;
1516 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1517 	free(outbuf, M_TEMP);
1518 	return (error);
1519 }
1520 
1521 static int
1522 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1523 {
1524 	struct ifbaconf *bac = arg;
1525 	struct bridge_rtnode *brt;
1526 	struct ifbareq bareq;
1527 	char *buf, *outbuf;
1528 	int count, buflen, len, error = 0;
1529 
1530 	if (bac->ifbac_len == 0)
1531 		return (0);
1532 
1533 	count = 0;
1534 	CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1535 		count++;
1536 	buflen = sizeof(bareq) * count;
1537 
1538 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1539 	if (outbuf == NULL)
1540 		return (ENOMEM);
1541 
1542 	count = 0;
1543 	buf = outbuf;
1544 	len = min(bac->ifbac_len, buflen);
1545 	bzero(&bareq, sizeof(bareq));
1546 	CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1547 		if (len < sizeof(bareq))
1548 			goto out;
1549 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1550 		    sizeof(bareq.ifba_ifsname));
1551 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1552 		bareq.ifba_vlan = brt->brt_vlan;
1553 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1554 				time_uptime < brt->brt_expire)
1555 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1556 		else
1557 			bareq.ifba_expire = 0;
1558 		bareq.ifba_flags = brt->brt_flags;
1559 
1560 		memcpy(buf, &bareq, sizeof(bareq));
1561 		count++;
1562 		buf += sizeof(bareq);
1563 		len -= sizeof(bareq);
1564 	}
1565 out:
1566 	bac->ifbac_len = sizeof(bareq) * count;
1567 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1568 	free(outbuf, M_TEMP);
1569 	return (error);
1570 }
1571 
1572 static int
1573 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1574 {
1575 	struct ifbareq *req = arg;
1576 	struct bridge_iflist *bif;
1577 	struct epoch_tracker et;
1578 	int error;
1579 
1580 	NET_EPOCH_ENTER(et);
1581 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1582 	if (bif == NULL) {
1583 		NET_EPOCH_EXIT(et);
1584 		return (ENOENT);
1585 	}
1586 
1587 	/* bridge_rtupdate() may acquire the lock. */
1588 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1589 	    req->ifba_flags);
1590 	NET_EPOCH_EXIT(et);
1591 
1592 	return (error);
1593 }
1594 
1595 static int
1596 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1597 {
1598 	struct ifbrparam *param = arg;
1599 
1600 	sc->sc_brttimeout = param->ifbrp_ctime;
1601 	return (0);
1602 }
1603 
1604 static int
1605 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1606 {
1607 	struct ifbrparam *param = arg;
1608 
1609 	param->ifbrp_ctime = sc->sc_brttimeout;
1610 	return (0);
1611 }
1612 
1613 static int
1614 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1615 {
1616 	struct ifbareq *req = arg;
1617 
1618 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1619 }
1620 
1621 static int
1622 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1623 {
1624 	struct ifbreq *req = arg;
1625 
1626 	BRIDGE_RT_LOCK(sc);
1627 	bridge_rtflush(sc, req->ifbr_ifsflags);
1628 	BRIDGE_RT_UNLOCK(sc);
1629 
1630 	return (0);
1631 }
1632 
1633 static int
1634 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1635 {
1636 	struct ifbrparam *param = arg;
1637 	struct bstp_state *bs = &sc->sc_stp;
1638 
1639 	param->ifbrp_prio = bs->bs_bridge_priority;
1640 	return (0);
1641 }
1642 
1643 static int
1644 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1645 {
1646 	struct ifbrparam *param = arg;
1647 
1648 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1649 }
1650 
1651 static int
1652 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1653 {
1654 	struct ifbrparam *param = arg;
1655 	struct bstp_state *bs = &sc->sc_stp;
1656 
1657 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1658 	return (0);
1659 }
1660 
1661 static int
1662 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1663 {
1664 	struct ifbrparam *param = arg;
1665 
1666 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1667 }
1668 
1669 static int
1670 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1671 {
1672 	struct ifbrparam *param = arg;
1673 	struct bstp_state *bs = &sc->sc_stp;
1674 
1675 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1676 	return (0);
1677 }
1678 
1679 static int
1680 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1681 {
1682 	struct ifbrparam *param = arg;
1683 
1684 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1685 }
1686 
1687 static int
1688 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1689 {
1690 	struct ifbrparam *param = arg;
1691 	struct bstp_state *bs = &sc->sc_stp;
1692 
1693 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1694 	return (0);
1695 }
1696 
1697 static int
1698 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1699 {
1700 	struct ifbrparam *param = arg;
1701 
1702 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1703 }
1704 
1705 static int
1706 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1707 {
1708 	struct ifbreq *req = arg;
1709 	struct bridge_iflist *bif;
1710 
1711 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1712 	if (bif == NULL)
1713 		return (ENOENT);
1714 
1715 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1716 }
1717 
1718 static int
1719 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1720 {
1721 	struct ifbreq *req = arg;
1722 	struct bridge_iflist *bif;
1723 
1724 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1725 	if (bif == NULL)
1726 		return (ENOENT);
1727 
1728 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1729 }
1730 
1731 static int
1732 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1733 {
1734 	struct ifbreq *req = arg;
1735 	struct bridge_iflist *bif;
1736 
1737 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1738 	if (bif == NULL)
1739 		return (ENOENT);
1740 
1741 	bif->bif_addrmax = req->ifbr_addrmax;
1742 	return (0);
1743 }
1744 
1745 static int
1746 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1747 {
1748 	struct ifbreq *req = arg;
1749 	struct bridge_iflist *bif = NULL;
1750 	struct ifnet *ifs;
1751 
1752 	ifs = ifunit(req->ifbr_ifsname);
1753 	if (ifs == NULL)
1754 		return (ENOENT);
1755 
1756 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1757 		if (ifs == bif->bif_ifp)
1758 			return (EBUSY);
1759 
1760 	if (ifs->if_bridge != NULL)
1761 		return (EBUSY);
1762 
1763 	switch (ifs->if_type) {
1764 		case IFT_ETHER:
1765 		case IFT_GIF:
1766 		case IFT_L2VLAN:
1767 			break;
1768 		default:
1769 			return (EINVAL);
1770 	}
1771 
1772 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1773 	if (bif == NULL)
1774 		return (ENOMEM);
1775 
1776 	bif->bif_ifp = ifs;
1777 	bif->bif_flags = IFBIF_SPAN;
1778 
1779 	CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1780 
1781 	return (0);
1782 }
1783 
1784 static int
1785 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1786 {
1787 	struct ifbreq *req = arg;
1788 	struct bridge_iflist *bif;
1789 	struct ifnet *ifs;
1790 
1791 	ifs = ifunit(req->ifbr_ifsname);
1792 	if (ifs == NULL)
1793 		return (ENOENT);
1794 
1795 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1796 		if (ifs == bif->bif_ifp)
1797 			break;
1798 
1799 	if (bif == NULL)
1800 		return (ENOENT);
1801 
1802 	bridge_delete_span(sc, bif);
1803 
1804 	return (0);
1805 }
1806 
1807 static int
1808 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1809 {
1810 	struct ifbropreq *req = arg;
1811 	struct bstp_state *bs = &sc->sc_stp;
1812 	struct bstp_port *root_port;
1813 
1814 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1815 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1816 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1817 
1818 	root_port = bs->bs_root_port;
1819 	if (root_port == NULL)
1820 		req->ifbop_root_port = 0;
1821 	else
1822 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1823 
1824 	req->ifbop_holdcount = bs->bs_txholdcount;
1825 	req->ifbop_priority = bs->bs_bridge_priority;
1826 	req->ifbop_protocol = bs->bs_protover;
1827 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1828 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1829 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1830 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1831 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1832 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1833 
1834 	return (0);
1835 }
1836 
1837 static int
1838 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1839 {
1840 	struct ifbrparam *param = arg;
1841 
1842 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1843 	return (0);
1844 }
1845 
1846 static int
1847 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1848 {
1849 	struct ifbpstpconf *bifstp = arg;
1850 	struct bridge_iflist *bif;
1851 	struct bstp_port *bp;
1852 	struct ifbpstpreq bpreq;
1853 	char *buf, *outbuf;
1854 	int count, buflen, len, error = 0;
1855 
1856 	count = 0;
1857 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1858 		if ((bif->bif_flags & IFBIF_STP) != 0)
1859 			count++;
1860 	}
1861 
1862 	buflen = sizeof(bpreq) * count;
1863 	if (bifstp->ifbpstp_len == 0) {
1864 		bifstp->ifbpstp_len = buflen;
1865 		return (0);
1866 	}
1867 
1868 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1869 	if (outbuf == NULL)
1870 		return (ENOMEM);
1871 
1872 	count = 0;
1873 	buf = outbuf;
1874 	len = min(bifstp->ifbpstp_len, buflen);
1875 	bzero(&bpreq, sizeof(bpreq));
1876 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1877 		if (len < sizeof(bpreq))
1878 			break;
1879 
1880 		if ((bif->bif_flags & IFBIF_STP) == 0)
1881 			continue;
1882 
1883 		bp = &bif->bif_stp;
1884 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1885 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1886 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1887 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1888 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1889 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1890 
1891 		memcpy(buf, &bpreq, sizeof(bpreq));
1892 		count++;
1893 		buf += sizeof(bpreq);
1894 		len -= sizeof(bpreq);
1895 	}
1896 
1897 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1898 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1899 	free(outbuf, M_TEMP);
1900 	return (error);
1901 }
1902 
1903 static int
1904 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1905 {
1906 	struct ifbrparam *param = arg;
1907 
1908 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1909 }
1910 
1911 static int
1912 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1913 {
1914 	struct ifbrparam *param = arg;
1915 
1916 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1917 }
1918 
1919 /*
1920  * bridge_ifdetach:
1921  *
1922  *	Detach an interface from a bridge.  Called when a member
1923  *	interface is detaching.
1924  */
1925 static void
1926 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1927 {
1928 	struct bridge_softc *sc = ifp->if_bridge;
1929 	struct bridge_iflist *bif;
1930 
1931 	if (ifp->if_flags & IFF_RENAMING)
1932 		return;
1933 	if (V_bridge_cloner == NULL) {
1934 		/*
1935 		 * This detach handler can be called after
1936 		 * vnet_bridge_uninit().  Just return in that case.
1937 		 */
1938 		return;
1939 	}
1940 	/* Check if the interface is a bridge member */
1941 	if (sc != NULL) {
1942 		BRIDGE_LOCK(sc);
1943 
1944 		bif = bridge_lookup_member_if(sc, ifp);
1945 		if (bif != NULL)
1946 			bridge_delete_member(sc, bif, 1);
1947 
1948 		BRIDGE_UNLOCK(sc);
1949 		return;
1950 	}
1951 
1952 	/* Check if the interface is a span port */
1953 	BRIDGE_LIST_LOCK();
1954 	LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1955 		BRIDGE_LOCK(sc);
1956 		CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1957 			if (ifp == bif->bif_ifp) {
1958 				bridge_delete_span(sc, bif);
1959 				break;
1960 			}
1961 
1962 		BRIDGE_UNLOCK(sc);
1963 	}
1964 	BRIDGE_LIST_UNLOCK();
1965 }
1966 
1967 /*
1968  * bridge_init:
1969  *
1970  *	Initialize a bridge interface.
1971  */
1972 static void
1973 bridge_init(void *xsc)
1974 {
1975 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1976 	struct ifnet *ifp = sc->sc_ifp;
1977 
1978 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1979 		return;
1980 
1981 	BRIDGE_LOCK(sc);
1982 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1983 	    bridge_timer, sc);
1984 
1985 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1986 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1987 
1988 	BRIDGE_UNLOCK(sc);
1989 }
1990 
1991 /*
1992  * bridge_stop:
1993  *
1994  *	Stop the bridge interface.
1995  */
1996 static void
1997 bridge_stop(struct ifnet *ifp, int disable)
1998 {
1999 	struct bridge_softc *sc = ifp->if_softc;
2000 
2001 	BRIDGE_LOCK_ASSERT(sc);
2002 
2003 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2004 		return;
2005 
2006 	BRIDGE_RT_LOCK(sc);
2007 	callout_stop(&sc->sc_brcallout);
2008 
2009 	bstp_stop(&sc->sc_stp);
2010 
2011 	bridge_rtflush(sc, IFBF_FLUSHDYN);
2012 	BRIDGE_RT_UNLOCK(sc);
2013 
2014 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
2015 }
2016 
2017 /*
2018  * bridge_enqueue:
2019  *
2020  *	Enqueue a packet on a bridge member interface.
2021  *
2022  */
2023 static int
2024 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
2025 {
2026 	int len, err = 0;
2027 	short mflags;
2028 	struct mbuf *m0;
2029 
2030 	/* We may be sending a fragment so traverse the mbuf */
2031 	for (; m; m = m0) {
2032 		m0 = m->m_nextpkt;
2033 		m->m_nextpkt = NULL;
2034 		len = m->m_pkthdr.len;
2035 		mflags = m->m_flags;
2036 
2037 		/*
2038 		 * If underlying interface can not do VLAN tag insertion itself
2039 		 * then attach a packet tag that holds it.
2040 		 */
2041 		if ((m->m_flags & M_VLANTAG) &&
2042 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
2043 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
2044 			if (m == NULL) {
2045 				if_printf(dst_ifp,
2046 				    "unable to prepend VLAN header\n");
2047 				if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2048 				continue;
2049 			}
2050 			m->m_flags &= ~M_VLANTAG;
2051 		}
2052 
2053 		M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2054 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2055 			m_freem(m0);
2056 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2057 			break;
2058 		}
2059 
2060 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2061 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2062 		if (mflags & M_MCAST)
2063 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2064 	}
2065 
2066 	return (err);
2067 }
2068 
2069 /*
2070  * bridge_dummynet:
2071  *
2072  * 	Receive a queued packet from dummynet and pass it on to the output
2073  * 	interface.
2074  *
2075  *	The mbuf has the Ethernet header already attached.
2076  */
2077 static void
2078 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2079 {
2080 	struct bridge_softc *sc;
2081 
2082 	sc = ifp->if_bridge;
2083 
2084 	/*
2085 	 * The packet didnt originate from a member interface. This should only
2086 	 * ever happen if a member interface is removed while packets are
2087 	 * queued for it.
2088 	 */
2089 	if (sc == NULL) {
2090 		m_freem(m);
2091 		return;
2092 	}
2093 
2094 	if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2095 #ifdef INET6
2096 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2097 #endif
2098 	    ) {
2099 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2100 			return;
2101 		if (m == NULL)
2102 			return;
2103 	}
2104 
2105 	bridge_enqueue(sc, ifp, m);
2106 }
2107 
2108 /*
2109  * bridge_output:
2110  *
2111  *	Send output from a bridge member interface.  This
2112  *	performs the bridging function for locally originated
2113  *	packets.
2114  *
2115  *	The mbuf has the Ethernet header already attached.  We must
2116  *	enqueue or free the mbuf before returning.
2117  */
2118 static int
2119 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2120     struct rtentry *rt)
2121 {
2122 	struct ether_header *eh;
2123 	struct ifnet *bifp, *dst_if;
2124 	struct bridge_softc *sc;
2125 	uint16_t vlan;
2126 
2127 	NET_EPOCH_ASSERT();
2128 
2129 	if (m->m_len < ETHER_HDR_LEN) {
2130 		m = m_pullup(m, ETHER_HDR_LEN);
2131 		if (m == NULL)
2132 			return (0);
2133 	}
2134 
2135 	eh = mtod(m, struct ether_header *);
2136 	sc = ifp->if_bridge;
2137 	vlan = VLANTAGOF(m);
2138 
2139 	bifp = sc->sc_ifp;
2140 
2141 	/*
2142 	 * If bridge is down, but the original output interface is up,
2143 	 * go ahead and send out that interface.  Otherwise, the packet
2144 	 * is dropped below.
2145 	 */
2146 	if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2147 		dst_if = ifp;
2148 		goto sendunicast;
2149 	}
2150 
2151 	/*
2152 	 * If the packet is a multicast, or we don't know a better way to
2153 	 * get there, send to all interfaces.
2154 	 */
2155 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
2156 		dst_if = NULL;
2157 	else
2158 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2159 	/* Tap any traffic not passing back out the originating interface */
2160 	if (dst_if != ifp)
2161 		ETHER_BPF_MTAP(bifp, m);
2162 	if (dst_if == NULL) {
2163 		struct bridge_iflist *bif;
2164 		struct mbuf *mc;
2165 		int used = 0;
2166 
2167 		bridge_span(sc, m);
2168 
2169 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2170 			dst_if = bif->bif_ifp;
2171 
2172 			if (dst_if->if_type == IFT_GIF)
2173 				continue;
2174 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2175 				continue;
2176 
2177 			/*
2178 			 * If this is not the original output interface,
2179 			 * and the interface is participating in spanning
2180 			 * tree, make sure the port is in a state that
2181 			 * allows forwarding.
2182 			 */
2183 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2184 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2185 				continue;
2186 
2187 			if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2188 				used = 1;
2189 				mc = m;
2190 			} else {
2191 				mc = m_dup(m, M_NOWAIT);
2192 				if (mc == NULL) {
2193 					if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2194 					continue;
2195 				}
2196 			}
2197 
2198 			bridge_enqueue(sc, dst_if, mc);
2199 		}
2200 		if (used == 0)
2201 			m_freem(m);
2202 		return (0);
2203 	}
2204 
2205 sendunicast:
2206 	/*
2207 	 * XXX Spanning tree consideration here?
2208 	 */
2209 
2210 	bridge_span(sc, m);
2211 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2212 		m_freem(m);
2213 		return (0);
2214 	}
2215 
2216 	bridge_enqueue(sc, dst_if, m);
2217 	return (0);
2218 }
2219 
2220 /*
2221  * bridge_transmit:
2222  *
2223  *	Do output on a bridge.
2224  *
2225  */
2226 static int
2227 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2228 {
2229 	struct bridge_softc *sc;
2230 	struct ether_header *eh;
2231 	struct ifnet *dst_if;
2232 	int error = 0;
2233 
2234 	sc = ifp->if_softc;
2235 
2236 	ETHER_BPF_MTAP(ifp, m);
2237 
2238 	eh = mtod(m, struct ether_header *);
2239 
2240 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2241 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2242 		error = bridge_enqueue(sc, dst_if, m);
2243 	} else
2244 		bridge_broadcast(sc, ifp, m, 0);
2245 
2246 	return (error);
2247 }
2248 
2249 #ifdef ALTQ
2250 static void
2251 bridge_altq_start(if_t ifp)
2252 {
2253 	struct ifaltq *ifq = &ifp->if_snd;
2254 	struct mbuf *m;
2255 
2256 	IFQ_LOCK(ifq);
2257 	IFQ_DEQUEUE_NOLOCK(ifq, m);
2258 	while (m != NULL) {
2259 		bridge_transmit(ifp, m);
2260 		IFQ_DEQUEUE_NOLOCK(ifq, m);
2261 	}
2262 	IFQ_UNLOCK(ifq);
2263 }
2264 
2265 static int
2266 bridge_altq_transmit(if_t ifp, struct mbuf *m)
2267 {
2268 	int err;
2269 
2270 	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
2271 		IFQ_ENQUEUE(&ifp->if_snd, m, err);
2272 		if (err == 0)
2273 			bridge_altq_start(ifp);
2274 	} else
2275 		err = bridge_transmit(ifp, m);
2276 
2277 	return (err);
2278 }
2279 #endif	/* ALTQ */
2280 
2281 /*
2282  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2283  */
2284 static void
2285 bridge_qflush(struct ifnet *ifp __unused)
2286 {
2287 }
2288 
2289 /*
2290  * bridge_forward:
2291  *
2292  *	The forwarding function of the bridge.
2293  *
2294  *	NOTE: Releases the lock on return.
2295  */
2296 static void
2297 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2298     struct mbuf *m)
2299 {
2300 	struct bridge_iflist *dbif;
2301 	struct ifnet *src_if, *dst_if, *ifp;
2302 	struct ether_header *eh;
2303 	uint16_t vlan;
2304 	uint8_t *dst;
2305 	int error;
2306 
2307 	NET_EPOCH_ASSERT();
2308 
2309 	src_if = m->m_pkthdr.rcvif;
2310 	ifp = sc->sc_ifp;
2311 
2312 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2313 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2314 	vlan = VLANTAGOF(m);
2315 
2316 	if ((sbif->bif_flags & IFBIF_STP) &&
2317 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2318 		goto drop;
2319 
2320 	eh = mtod(m, struct ether_header *);
2321 	dst = eh->ether_dhost;
2322 
2323 	/* If the interface is learning, record the address. */
2324 	if (sbif->bif_flags & IFBIF_LEARNING) {
2325 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2326 		    sbif, 0, IFBAF_DYNAMIC);
2327 		/*
2328 		 * If the interface has addresses limits then deny any source
2329 		 * that is not in the cache.
2330 		 */
2331 		if (error && sbif->bif_addrmax)
2332 			goto drop;
2333 	}
2334 
2335 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2336 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2337 		goto drop;
2338 
2339 	/*
2340 	 * At this point, the port either doesn't participate
2341 	 * in spanning tree or it is in the forwarding state.
2342 	 */
2343 
2344 	/*
2345 	 * If the packet is unicast, destined for someone on
2346 	 * "this" side of the bridge, drop it.
2347 	 */
2348 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2349 		dst_if = bridge_rtlookup(sc, dst, vlan);
2350 		if (src_if == dst_if)
2351 			goto drop;
2352 	} else {
2353 		/*
2354 		 * Check if its a reserved multicast address, any address
2355 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2356 		 * bridge.
2357 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2358 		 */
2359 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2360 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2361 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2362 			goto drop;
2363 
2364 		/* ...forward it to all interfaces. */
2365 		if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2366 		dst_if = NULL;
2367 	}
2368 
2369 	/*
2370 	 * If we have a destination interface which is a member of our bridge,
2371 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2372 	 * For broadcast or multicast packets, don't bother because it will
2373 	 * be reinjected into ether_input. We do this before we pass the packets
2374 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2375 	 * drop the packet, or possibly modify it, making it difficult to debug
2376 	 * firewall issues on the bridge.
2377 	 */
2378 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2379 		ETHER_BPF_MTAP(ifp, m);
2380 
2381 	/* run the packet filter */
2382 	if (PFIL_HOOKED_IN(V_inet_pfil_head)
2383 #ifdef INET6
2384 	    || PFIL_HOOKED_IN(V_inet6_pfil_head)
2385 #endif
2386 	    ) {
2387 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2388 			return;
2389 		if (m == NULL)
2390 			return;
2391 	}
2392 
2393 	if (dst_if == NULL) {
2394 		bridge_broadcast(sc, src_if, m, 1);
2395 		return;
2396 	}
2397 
2398 	/*
2399 	 * At this point, we're dealing with a unicast frame
2400 	 * going to a different interface.
2401 	 */
2402 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2403 		goto drop;
2404 
2405 	dbif = bridge_lookup_member_if(sc, dst_if);
2406 	if (dbif == NULL)
2407 		/* Not a member of the bridge (anymore?) */
2408 		goto drop;
2409 
2410 	/* Private segments can not talk to each other */
2411 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2412 		goto drop;
2413 
2414 	if ((dbif->bif_flags & IFBIF_STP) &&
2415 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2416 		goto drop;
2417 
2418 	if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2419 #ifdef INET6
2420 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2421 #endif
2422 	    ) {
2423 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2424 			return;
2425 		if (m == NULL)
2426 			return;
2427 	}
2428 
2429 	bridge_enqueue(sc, dst_if, m);
2430 	return;
2431 
2432 drop:
2433 	m_freem(m);
2434 }
2435 
2436 /*
2437  * bridge_input:
2438  *
2439  *	Receive input from a member interface.  Queue the packet for
2440  *	bridging if it is not for us.
2441  */
2442 static struct mbuf *
2443 bridge_input(struct ifnet *ifp, struct mbuf *m)
2444 {
2445 	struct bridge_softc *sc = ifp->if_bridge;
2446 	struct bridge_iflist *bif, *bif2;
2447 	struct ifnet *bifp;
2448 	struct ether_header *eh;
2449 	struct mbuf *mc, *mc2;
2450 	uint16_t vlan;
2451 	int error;
2452 
2453 	NET_EPOCH_ASSERT();
2454 
2455 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2456 		return (m);
2457 
2458 	bifp = sc->sc_ifp;
2459 	vlan = VLANTAGOF(m);
2460 
2461 	/*
2462 	 * Implement support for bridge monitoring. If this flag has been
2463 	 * set on this interface, discard the packet once we push it through
2464 	 * the bpf(4) machinery, but before we do, increment the byte and
2465 	 * packet counters associated with this interface.
2466 	 */
2467 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2468 		m->m_pkthdr.rcvif  = bifp;
2469 		ETHER_BPF_MTAP(bifp, m);
2470 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2471 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2472 		m_freem(m);
2473 		return (NULL);
2474 	}
2475 	bif = bridge_lookup_member_if(sc, ifp);
2476 	if (bif == NULL) {
2477 		return (m);
2478 	}
2479 
2480 	eh = mtod(m, struct ether_header *);
2481 
2482 	bridge_span(sc, m);
2483 
2484 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2485 		/* Tap off 802.1D packets; they do not get forwarded. */
2486 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2487 		    ETHER_ADDR_LEN) == 0) {
2488 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2489 			return (NULL);
2490 		}
2491 
2492 		if ((bif->bif_flags & IFBIF_STP) &&
2493 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2494 			return (m);
2495 		}
2496 
2497 		/*
2498 		 * Make a deep copy of the packet and enqueue the copy
2499 		 * for bridge processing; return the original packet for
2500 		 * local processing.
2501 		 */
2502 		mc = m_dup(m, M_NOWAIT);
2503 		if (mc == NULL) {
2504 			return (m);
2505 		}
2506 
2507 		/* Perform the bridge forwarding function with the copy. */
2508 		bridge_forward(sc, bif, mc);
2509 
2510 		/*
2511 		 * Reinject the mbuf as arriving on the bridge so we have a
2512 		 * chance at claiming multicast packets. We can not loop back
2513 		 * here from ether_input as a bridge is never a member of a
2514 		 * bridge.
2515 		 */
2516 		KASSERT(bifp->if_bridge == NULL,
2517 		    ("loop created in bridge_input"));
2518 		mc2 = m_dup(m, M_NOWAIT);
2519 		if (mc2 != NULL) {
2520 			/* Keep the layer3 header aligned */
2521 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2522 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2523 		}
2524 		if (mc2 != NULL) {
2525 			mc2->m_pkthdr.rcvif = bifp;
2526 			(*bifp->if_input)(bifp, mc2);
2527 		}
2528 
2529 		/* Return the original packet for local processing. */
2530 		return (m);
2531 	}
2532 
2533 	if ((bif->bif_flags & IFBIF_STP) &&
2534 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2535 		return (m);
2536 	}
2537 
2538 #if (defined(INET) || defined(INET6))
2539 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2540 	|| ((iface)->if_carp \
2541 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2542 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2543 	|| ((iface)->if_carp \
2544 	    && (*carp_forus_p)((iface), eh->ether_shost))
2545 #else
2546 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2547 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2548 #endif
2549 
2550 #ifdef INET6
2551 #   define OR_PFIL_HOOKED_INET6 \
2552 	|| PFIL_HOOKED_IN(V_inet6_pfil_head)
2553 #else
2554 #   define OR_PFIL_HOOKED_INET6
2555 #endif
2556 
2557 #define GRAB_OUR_PACKETS(iface) \
2558 	if ((iface)->if_type == IFT_GIF) \
2559 		continue; \
2560 	/* It is destined for us. */ \
2561 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2562 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2563 	    ) {								\
2564 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2565 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2566 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2567 			if (error && bif->bif_addrmax) {		\
2568 				m_freem(m);				\
2569 				return (NULL);				\
2570 			}						\
2571 		}							\
2572 		m->m_pkthdr.rcvif = iface;				\
2573 		if ((iface) == ifp) {					\
2574 			/* Skip bridge processing... src == dest */	\
2575 			return (m);					\
2576 		}							\
2577 		/* It's passing over or to the bridge, locally. */	\
2578 		ETHER_BPF_MTAP(bifp, m);				\
2579 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);		\
2580 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2581 		/* Filter on the physical interface. */			\
2582 		if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \
2583 		     OR_PFIL_HOOKED_INET6)) {				\
2584 			if (bridge_pfil(&m, NULL, ifp,			\
2585 			    PFIL_IN) != 0 || m == NULL) {		\
2586 				return (NULL);				\
2587 			}						\
2588 		}							\
2589 		if ((iface) != bifp)					\
2590 			ETHER_BPF_MTAP(iface, m);			\
2591 		return (m);						\
2592 	}								\
2593 									\
2594 	/* We just received a packet that we sent out. */		\
2595 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2596 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2597 	    ) {								\
2598 		m_freem(m);						\
2599 		return (NULL);						\
2600 	}
2601 
2602 	/*
2603 	 * Unicast.  Make sure it's not for the bridge.
2604 	 */
2605 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2606 
2607 	/*
2608 	 * Give a chance for ifp at first priority. This will help when	the
2609 	 * packet comes through the interface like VLAN's with the same MACs
2610 	 * on several interfaces from the same bridge. This also will save
2611 	 * some CPU cycles in case the destination interface and the input
2612 	 * interface (eq ifp) are the same.
2613 	 */
2614 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2615 
2616 	/* Now check the all bridge members. */
2617 	CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2618 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2619 	}
2620 
2621 #undef OR_CARP_CHECK_WE_ARE_DST
2622 #undef OR_CARP_CHECK_WE_ARE_SRC
2623 #undef OR_PFIL_HOOKED_INET6
2624 #undef GRAB_OUR_PACKETS
2625 
2626 	/* Perform the bridge forwarding function. */
2627 	bridge_forward(sc, bif, m);
2628 
2629 	return (NULL);
2630 }
2631 
2632 /*
2633  * bridge_broadcast:
2634  *
2635  *	Send a frame to all interfaces that are members of
2636  *	the bridge, except for the one on which the packet
2637  *	arrived.
2638  *
2639  *	NOTE: Releases the lock on return.
2640  */
2641 static void
2642 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2643     struct mbuf *m, int runfilt)
2644 {
2645 	struct bridge_iflist *dbif, *sbif;
2646 	struct mbuf *mc;
2647 	struct ifnet *dst_if;
2648 	int used = 0, i;
2649 
2650 	NET_EPOCH_ASSERT();
2651 
2652 	sbif = bridge_lookup_member_if(sc, src_if);
2653 
2654 	/* Filter on the bridge interface before broadcasting */
2655 	if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2656 #ifdef INET6
2657 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2658 #endif
2659 	    )) {
2660 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2661 			return;
2662 		if (m == NULL)
2663 			return;
2664 	}
2665 
2666 	CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2667 		dst_if = dbif->bif_ifp;
2668 		if (dst_if == src_if)
2669 			continue;
2670 
2671 		/* Private segments can not talk to each other */
2672 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2673 			continue;
2674 
2675 		if ((dbif->bif_flags & IFBIF_STP) &&
2676 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2677 			continue;
2678 
2679 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2680 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2681 			continue;
2682 
2683 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2684 			continue;
2685 
2686 		if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2687 			mc = m;
2688 			used = 1;
2689 		} else {
2690 			mc = m_dup(m, M_NOWAIT);
2691 			if (mc == NULL) {
2692 				if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2693 				continue;
2694 			}
2695 		}
2696 
2697 		/*
2698 		 * Filter on the output interface. Pass a NULL bridge interface
2699 		 * pointer so we do not redundantly filter on the bridge for
2700 		 * each interface we broadcast on.
2701 		 */
2702 		if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2703 #ifdef INET6
2704 		    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2705 #endif
2706 		    )) {
2707 			if (used == 0) {
2708 				/* Keep the layer3 header aligned */
2709 				i = min(mc->m_pkthdr.len, max_protohdr);
2710 				mc = m_copyup(mc, i, ETHER_ALIGN);
2711 				if (mc == NULL) {
2712 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2713 					continue;
2714 				}
2715 			}
2716 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2717 				continue;
2718 			if (mc == NULL)
2719 				continue;
2720 		}
2721 
2722 		bridge_enqueue(sc, dst_if, mc);
2723 	}
2724 	if (used == 0)
2725 		m_freem(m);
2726 }
2727 
2728 /*
2729  * bridge_span:
2730  *
2731  *	Duplicate a packet out one or more interfaces that are in span mode,
2732  *	the original mbuf is unmodified.
2733  */
2734 static void
2735 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2736 {
2737 	struct bridge_iflist *bif;
2738 	struct ifnet *dst_if;
2739 	struct mbuf *mc;
2740 
2741 	NET_EPOCH_ASSERT();
2742 
2743 	if (CK_LIST_EMPTY(&sc->sc_spanlist))
2744 		return;
2745 
2746 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2747 		dst_if = bif->bif_ifp;
2748 
2749 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2750 			continue;
2751 
2752 		mc = m_dup(m, M_NOWAIT);
2753 		if (mc == NULL) {
2754 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2755 			continue;
2756 		}
2757 
2758 		bridge_enqueue(sc, dst_if, mc);
2759 	}
2760 }
2761 
2762 /*
2763  * bridge_rtupdate:
2764  *
2765  *	Add a bridge routing entry.
2766  */
2767 static int
2768 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2769     struct bridge_iflist *bif, int setflags, uint8_t flags)
2770 {
2771 	struct bridge_rtnode *brt;
2772 	int error;
2773 
2774 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
2775 
2776 	/* Check the source address is valid and not multicast. */
2777 	if (ETHER_IS_MULTICAST(dst) ||
2778 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2779 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2780 		return (EINVAL);
2781 
2782 	/* 802.1p frames map to vlan 1 */
2783 	if (vlan == 0)
2784 		vlan = 1;
2785 
2786 	/*
2787 	 * A route for this destination might already exist.  If so,
2788 	 * update it, otherwise create a new one.
2789 	 */
2790 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2791 		BRIDGE_RT_LOCK(sc);
2792 
2793 		/* Check again, now that we have the lock. There could have
2794 		 * been a race and we only want to insert this once. */
2795 		if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) {
2796 			BRIDGE_RT_UNLOCK(sc);
2797 			return (0);
2798 		}
2799 
2800 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2801 			sc->sc_brtexceeded++;
2802 			BRIDGE_RT_UNLOCK(sc);
2803 			return (ENOSPC);
2804 		}
2805 		/* Check per interface address limits (if enabled) */
2806 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2807 			bif->bif_addrexceeded++;
2808 			BRIDGE_RT_UNLOCK(sc);
2809 			return (ENOSPC);
2810 		}
2811 
2812 		/*
2813 		 * Allocate a new bridge forwarding node, and
2814 		 * initialize the expiration time and Ethernet
2815 		 * address.
2816 		 */
2817 		brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2818 		if (brt == NULL) {
2819 			BRIDGE_RT_UNLOCK(sc);
2820 			return (ENOMEM);
2821 		}
2822 		brt->brt_vnet = curvnet;
2823 
2824 		if (bif->bif_flags & IFBIF_STICKY)
2825 			brt->brt_flags = IFBAF_STICKY;
2826 		else
2827 			brt->brt_flags = IFBAF_DYNAMIC;
2828 
2829 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2830 		brt->brt_vlan = vlan;
2831 
2832 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2833 			uma_zfree(V_bridge_rtnode_zone, brt);
2834 			BRIDGE_RT_UNLOCK(sc);
2835 			return (error);
2836 		}
2837 		brt->brt_dst = bif;
2838 		bif->bif_addrcnt++;
2839 
2840 		BRIDGE_RT_UNLOCK(sc);
2841 	}
2842 
2843 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2844 	    brt->brt_dst != bif) {
2845 		BRIDGE_RT_LOCK(sc);
2846 		brt->brt_dst->bif_addrcnt--;
2847 		brt->brt_dst = bif;
2848 		brt->brt_dst->bif_addrcnt++;
2849 		BRIDGE_RT_UNLOCK(sc);
2850 	}
2851 
2852 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2853 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2854 	if (setflags)
2855 		brt->brt_flags = flags;
2856 
2857 	return (0);
2858 }
2859 
2860 /*
2861  * bridge_rtlookup:
2862  *
2863  *	Lookup the destination interface for an address.
2864  */
2865 static struct ifnet *
2866 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2867 {
2868 	struct bridge_rtnode *brt;
2869 
2870 	NET_EPOCH_ASSERT();
2871 
2872 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2873 		return (NULL);
2874 
2875 	return (brt->brt_ifp);
2876 }
2877 
2878 /*
2879  * bridge_rttrim:
2880  *
2881  *	Trim the routine table so that we have a number
2882  *	of routing entries less than or equal to the
2883  *	maximum number.
2884  */
2885 static void
2886 bridge_rttrim(struct bridge_softc *sc)
2887 {
2888 	struct bridge_rtnode *brt, *nbrt;
2889 
2890 	NET_EPOCH_ASSERT();
2891 	BRIDGE_RT_LOCK_ASSERT(sc);
2892 
2893 	/* Make sure we actually need to do this. */
2894 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2895 		return;
2896 
2897 	/* Force an aging cycle; this might trim enough addresses. */
2898 	bridge_rtage(sc);
2899 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2900 		return;
2901 
2902 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2903 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2904 			bridge_rtnode_destroy(sc, brt);
2905 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2906 				return;
2907 		}
2908 	}
2909 }
2910 
2911 /*
2912  * bridge_timer:
2913  *
2914  *	Aging timer for the bridge.
2915  */
2916 static void
2917 bridge_timer(void *arg)
2918 {
2919 	struct bridge_softc *sc = arg;
2920 
2921 	BRIDGE_RT_LOCK_ASSERT(sc);
2922 
2923 	/* Destruction of rtnodes requires a proper vnet context */
2924 	CURVNET_SET(sc->sc_ifp->if_vnet);
2925 	bridge_rtage(sc);
2926 
2927 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2928 		callout_reset(&sc->sc_brcallout,
2929 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2930 	CURVNET_RESTORE();
2931 }
2932 
2933 /*
2934  * bridge_rtage:
2935  *
2936  *	Perform an aging cycle.
2937  */
2938 static void
2939 bridge_rtage(struct bridge_softc *sc)
2940 {
2941 	struct bridge_rtnode *brt, *nbrt;
2942 
2943 	BRIDGE_RT_LOCK_ASSERT(sc);
2944 
2945 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2946 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2947 			if (time_uptime >= brt->brt_expire)
2948 				bridge_rtnode_destroy(sc, brt);
2949 		}
2950 	}
2951 }
2952 
2953 /*
2954  * bridge_rtflush:
2955  *
2956  *	Remove all dynamic addresses from the bridge.
2957  */
2958 static void
2959 bridge_rtflush(struct bridge_softc *sc, int full)
2960 {
2961 	struct bridge_rtnode *brt, *nbrt;
2962 
2963 	BRIDGE_RT_LOCK_ASSERT(sc);
2964 
2965 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2966 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2967 			bridge_rtnode_destroy(sc, brt);
2968 	}
2969 }
2970 
2971 /*
2972  * bridge_rtdaddr:
2973  *
2974  *	Remove an address from the table.
2975  */
2976 static int
2977 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2978 {
2979 	struct bridge_rtnode *brt;
2980 	int found = 0;
2981 
2982 	BRIDGE_RT_LOCK(sc);
2983 
2984 	/*
2985 	 * If vlan is zero then we want to delete for all vlans so the lookup
2986 	 * may return more than one.
2987 	 */
2988 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2989 		bridge_rtnode_destroy(sc, brt);
2990 		found = 1;
2991 	}
2992 
2993 	BRIDGE_RT_UNLOCK(sc);
2994 
2995 	return (found ? 0 : ENOENT);
2996 }
2997 
2998 /*
2999  * bridge_rtdelete:
3000  *
3001  *	Delete routes to a speicifc member interface.
3002  */
3003 static void
3004 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
3005 {
3006 	struct bridge_rtnode *brt, *nbrt;
3007 
3008 	BRIDGE_RT_LOCK_ASSERT(sc);
3009 
3010 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3011 		if (brt->brt_ifp == ifp && (full ||
3012 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
3013 			bridge_rtnode_destroy(sc, brt);
3014 	}
3015 }
3016 
3017 /*
3018  * bridge_rtable_init:
3019  *
3020  *	Initialize the route table for this bridge.
3021  */
3022 static void
3023 bridge_rtable_init(struct bridge_softc *sc)
3024 {
3025 	int i;
3026 
3027 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
3028 	    M_DEVBUF, M_WAITOK);
3029 
3030 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
3031 		CK_LIST_INIT(&sc->sc_rthash[i]);
3032 
3033 	sc->sc_rthash_key = arc4random();
3034 	CK_LIST_INIT(&sc->sc_rtlist);
3035 }
3036 
3037 /*
3038  * bridge_rtable_fini:
3039  *
3040  *	Deconstruct the route table for this bridge.
3041  */
3042 static void
3043 bridge_rtable_fini(struct bridge_softc *sc)
3044 {
3045 
3046 	KASSERT(sc->sc_brtcnt == 0,
3047 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3048 	free(sc->sc_rthash, M_DEVBUF);
3049 }
3050 
3051 /*
3052  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3053  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3054  */
3055 #define	mix(a, b, c)							\
3056 do {									\
3057 	a -= b; a -= c; a ^= (c >> 13);					\
3058 	b -= c; b -= a; b ^= (a << 8);					\
3059 	c -= a; c -= b; c ^= (b >> 13);					\
3060 	a -= b; a -= c; a ^= (c >> 12);					\
3061 	b -= c; b -= a; b ^= (a << 16);					\
3062 	c -= a; c -= b; c ^= (b >> 5);					\
3063 	a -= b; a -= c; a ^= (c >> 3);					\
3064 	b -= c; b -= a; b ^= (a << 10);					\
3065 	c -= a; c -= b; c ^= (b >> 15);					\
3066 } while (/*CONSTCOND*/0)
3067 
3068 static __inline uint32_t
3069 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3070 {
3071 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3072 
3073 	b += addr[5] << 8;
3074 	b += addr[4];
3075 	a += addr[3] << 24;
3076 	a += addr[2] << 16;
3077 	a += addr[1] << 8;
3078 	a += addr[0];
3079 
3080 	mix(a, b, c);
3081 
3082 	return (c & BRIDGE_RTHASH_MASK);
3083 }
3084 
3085 #undef mix
3086 
3087 static int
3088 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3089 {
3090 	int i, d;
3091 
3092 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3093 		d = ((int)a[i]) - ((int)b[i]);
3094 	}
3095 
3096 	return (d);
3097 }
3098 
3099 /*
3100  * bridge_rtnode_lookup:
3101  *
3102  *	Look up a bridge route node for the specified destination. Compare the
3103  *	vlan id or if zero then just return the first match.
3104  */
3105 static struct bridge_rtnode *
3106 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3107 {
3108 	struct bridge_rtnode *brt;
3109 	uint32_t hash;
3110 	int dir;
3111 
3112 	BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3113 
3114 	hash = bridge_rthash(sc, addr);
3115 	CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3116 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3117 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3118 			return (brt);
3119 		if (dir > 0)
3120 			return (NULL);
3121 	}
3122 
3123 	return (NULL);
3124 }
3125 
3126 /*
3127  * bridge_rtnode_insert:
3128  *
3129  *	Insert the specified bridge node into the route table.  We
3130  *	assume the entry is not already in the table.
3131  */
3132 static int
3133 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3134 {
3135 	struct bridge_rtnode *lbrt;
3136 	uint32_t hash;
3137 	int dir;
3138 
3139 	BRIDGE_RT_LOCK_ASSERT(sc);
3140 
3141 	hash = bridge_rthash(sc, brt->brt_addr);
3142 
3143 	lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3144 	if (lbrt == NULL) {
3145 		CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3146 		goto out;
3147 	}
3148 
3149 	do {
3150 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3151 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3152 			return (EEXIST);
3153 		if (dir > 0) {
3154 			CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3155 			goto out;
3156 		}
3157 		if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3158 			CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3159 			goto out;
3160 		}
3161 		lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3162 	} while (lbrt != NULL);
3163 
3164 #ifdef DIAGNOSTIC
3165 	panic("bridge_rtnode_insert: impossible");
3166 #endif
3167 
3168 out:
3169 	CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3170 	sc->sc_brtcnt++;
3171 
3172 	return (0);
3173 }
3174 
3175 static void
3176 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3177 {
3178 	struct bridge_rtnode *brt;
3179 
3180 	brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3181 
3182 	CURVNET_SET(brt->brt_vnet);
3183 	uma_zfree(V_bridge_rtnode_zone, brt);
3184 	CURVNET_RESTORE();
3185 }
3186 
3187 /*
3188  * bridge_rtnode_destroy:
3189  *
3190  *	Destroy a bridge rtnode.
3191  */
3192 static void
3193 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3194 {
3195 	BRIDGE_RT_LOCK_ASSERT(sc);
3196 
3197 	CK_LIST_REMOVE(brt, brt_hash);
3198 
3199 	CK_LIST_REMOVE(brt, brt_list);
3200 	sc->sc_brtcnt--;
3201 	brt->brt_dst->bif_addrcnt--;
3202 
3203 	NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx);
3204 }
3205 
3206 /*
3207  * bridge_rtable_expire:
3208  *
3209  *	Set the expiry time for all routes on an interface.
3210  */
3211 static void
3212 bridge_rtable_expire(struct ifnet *ifp, int age)
3213 {
3214 	struct bridge_softc *sc = ifp->if_bridge;
3215 	struct bridge_rtnode *brt;
3216 
3217 	CURVNET_SET(ifp->if_vnet);
3218 	BRIDGE_RT_LOCK(sc);
3219 
3220 	/*
3221 	 * If the age is zero then flush, otherwise set all the expiry times to
3222 	 * age for the interface
3223 	 */
3224 	if (age == 0)
3225 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3226 	else {
3227 		CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3228 			/* Cap the expiry time to 'age' */
3229 			if (brt->brt_ifp == ifp &&
3230 			    brt->brt_expire > time_uptime + age &&
3231 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3232 				brt->brt_expire = time_uptime + age;
3233 		}
3234 	}
3235 	BRIDGE_RT_UNLOCK(sc);
3236 	CURVNET_RESTORE();
3237 }
3238 
3239 /*
3240  * bridge_state_change:
3241  *
3242  *	Callback from the bridgestp code when a port changes states.
3243  */
3244 static void
3245 bridge_state_change(struct ifnet *ifp, int state)
3246 {
3247 	struct bridge_softc *sc = ifp->if_bridge;
3248 	static const char *stpstates[] = {
3249 		"disabled",
3250 		"listening",
3251 		"learning",
3252 		"forwarding",
3253 		"blocking",
3254 		"discarding"
3255 	};
3256 
3257 	CURVNET_SET(ifp->if_vnet);
3258 	if (V_log_stp)
3259 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3260 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3261 	CURVNET_RESTORE();
3262 }
3263 
3264 /*
3265  * Send bridge packets through pfil if they are one of the types pfil can deal
3266  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3267  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3268  * that interface.
3269  */
3270 static int
3271 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3272 {
3273 	int snap, error, i, hlen;
3274 	struct ether_header *eh1, eh2;
3275 	struct ip *ip;
3276 	struct llc llc1;
3277 	u_int16_t ether_type;
3278 	pfil_return_t rv;
3279 
3280 	snap = 0;
3281 	error = -1;	/* Default error if not error == 0 */
3282 
3283 #if 0
3284 	/* we may return with the IP fields swapped, ensure its not shared */
3285 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3286 #endif
3287 
3288 	if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3289 		return (0); /* filtering is disabled */
3290 
3291 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3292 	if ((*mp)->m_len < i) {
3293 	    *mp = m_pullup(*mp, i);
3294 	    if (*mp == NULL) {
3295 		printf("%s: m_pullup failed\n", __func__);
3296 		return (-1);
3297 	    }
3298 	}
3299 
3300 	eh1 = mtod(*mp, struct ether_header *);
3301 	ether_type = ntohs(eh1->ether_type);
3302 
3303 	/*
3304 	 * Check for SNAP/LLC.
3305 	 */
3306 	if (ether_type < ETHERMTU) {
3307 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3308 
3309 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3310 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3311 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3312 		    llc2->llc_control == LLC_UI) {
3313 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3314 			snap = 1;
3315 		}
3316 	}
3317 
3318 	/*
3319 	 * If we're trying to filter bridge traffic, don't look at anything
3320 	 * other than IP and ARP traffic.  If the filter doesn't understand
3321 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3322 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3323 	 * but of course we don't have an AppleTalk filter to begin with.
3324 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3325 	 * ARP traffic.)
3326 	 */
3327 	switch (ether_type) {
3328 		case ETHERTYPE_ARP:
3329 		case ETHERTYPE_REVARP:
3330 			if (V_pfil_ipfw_arp == 0)
3331 				return (0); /* Automatically pass */
3332 			break;
3333 
3334 		case ETHERTYPE_IP:
3335 #ifdef INET6
3336 		case ETHERTYPE_IPV6:
3337 #endif /* INET6 */
3338 			break;
3339 		default:
3340 			/*
3341 			 * Check to see if the user wants to pass non-ip
3342 			 * packets, these will not be checked by pfil(9) and
3343 			 * passed unconditionally so the default is to drop.
3344 			 */
3345 			if (V_pfil_onlyip)
3346 				goto bad;
3347 	}
3348 
3349 	/* Run the packet through pfil before stripping link headers */
3350 	if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3351 	    dir == PFIL_OUT && ifp != NULL) {
3352 		switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) {
3353 		case PFIL_DROPPED:
3354 			return (EACCES);
3355 		case PFIL_CONSUMED:
3356 			return (0);
3357 		}
3358 	}
3359 
3360 	/* Strip off the Ethernet header and keep a copy. */
3361 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3362 	m_adj(*mp, ETHER_HDR_LEN);
3363 
3364 	/* Strip off snap header, if present */
3365 	if (snap) {
3366 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3367 		m_adj(*mp, sizeof(struct llc));
3368 	}
3369 
3370 	/*
3371 	 * Check the IP header for alignment and errors
3372 	 */
3373 	if (dir == PFIL_IN) {
3374 		switch (ether_type) {
3375 			case ETHERTYPE_IP:
3376 				error = bridge_ip_checkbasic(mp);
3377 				break;
3378 #ifdef INET6
3379 			case ETHERTYPE_IPV6:
3380 				error = bridge_ip6_checkbasic(mp);
3381 				break;
3382 #endif /* INET6 */
3383 			default:
3384 				error = 0;
3385 		}
3386 		if (error)
3387 			goto bad;
3388 	}
3389 
3390 	error = 0;
3391 
3392 	/*
3393 	 * Run the packet through pfil
3394 	 */
3395 	rv = PFIL_PASS;
3396 	switch (ether_type) {
3397 	case ETHERTYPE_IP:
3398 		/*
3399 		 * Run pfil on the member interface and the bridge, both can
3400 		 * be skipped by clearing pfil_member or pfil_bridge.
3401 		 *
3402 		 * Keep the order:
3403 		 *   in_if -> bridge_if -> out_if
3404 		 */
3405 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3406 		    pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3407 		    PFIL_PASS)
3408 			break;
3409 
3410 		if (V_pfil_member && ifp != NULL && (rv =
3411 		    pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) !=
3412 		    PFIL_PASS)
3413 			break;
3414 
3415 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3416 		    pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3417 		    PFIL_PASS)
3418 			break;
3419 
3420 		/* check if we need to fragment the packet */
3421 		/* bridge_fragment generates a mbuf chain of packets */
3422 		/* that already include eth headers */
3423 		if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3424 			i = (*mp)->m_pkthdr.len;
3425 			if (i > ifp->if_mtu) {
3426 				error = bridge_fragment(ifp, mp, &eh2, snap,
3427 					    &llc1);
3428 				return (error);
3429 			}
3430 		}
3431 
3432 		/* Recalculate the ip checksum. */
3433 		ip = mtod(*mp, struct ip *);
3434 		hlen = ip->ip_hl << 2;
3435 		if (hlen < sizeof(struct ip))
3436 			goto bad;
3437 		if (hlen > (*mp)->m_len) {
3438 			if ((*mp = m_pullup(*mp, hlen)) == NULL)
3439 				goto bad;
3440 			ip = mtod(*mp, struct ip *);
3441 			if (ip == NULL)
3442 				goto bad;
3443 		}
3444 		ip->ip_sum = 0;
3445 		if (hlen == sizeof(struct ip))
3446 			ip->ip_sum = in_cksum_hdr(ip);
3447 		else
3448 			ip->ip_sum = in_cksum(*mp, hlen);
3449 
3450 		break;
3451 #ifdef INET6
3452 	case ETHERTYPE_IPV6:
3453 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3454 		    pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3455 		    PFIL_PASS)
3456 			break;
3457 
3458 		if (V_pfil_member && ifp != NULL && (rv =
3459 		    pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) !=
3460 		    PFIL_PASS)
3461 			break;
3462 
3463 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3464 		    pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3465 		    PFIL_PASS)
3466 			break;
3467 		break;
3468 #endif
3469 	}
3470 
3471 	switch (rv) {
3472 	case PFIL_CONSUMED:
3473 		return (0);
3474 	case PFIL_DROPPED:
3475 		return (EACCES);
3476 	default:
3477 		break;
3478 	}
3479 
3480 	error = -1;
3481 
3482 	/*
3483 	 * Finally, put everything back the way it was and return
3484 	 */
3485 	if (snap) {
3486 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3487 		if (*mp == NULL)
3488 			return (error);
3489 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3490 	}
3491 
3492 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3493 	if (*mp == NULL)
3494 		return (error);
3495 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3496 
3497 	return (0);
3498 
3499 bad:
3500 	m_freem(*mp);
3501 	*mp = NULL;
3502 	return (error);
3503 }
3504 
3505 /*
3506  * Perform basic checks on header size since
3507  * pfil assumes ip_input has already processed
3508  * it for it.  Cut-and-pasted from ip_input.c.
3509  * Given how simple the IPv6 version is,
3510  * does the IPv4 version really need to be
3511  * this complicated?
3512  *
3513  * XXX Should we update ipstat here, or not?
3514  * XXX Right now we update ipstat but not
3515  * XXX csum_counter.
3516  */
3517 static int
3518 bridge_ip_checkbasic(struct mbuf **mp)
3519 {
3520 	struct mbuf *m = *mp;
3521 	struct ip *ip;
3522 	int len, hlen;
3523 	u_short sum;
3524 
3525 	if (*mp == NULL)
3526 		return (-1);
3527 
3528 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3529 		if ((m = m_copyup(m, sizeof(struct ip),
3530 			(max_linkhdr + 3) & ~3)) == NULL) {
3531 			/* XXXJRT new stat, please */
3532 			KMOD_IPSTAT_INC(ips_toosmall);
3533 			goto bad;
3534 		}
3535 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3536 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3537 			KMOD_IPSTAT_INC(ips_toosmall);
3538 			goto bad;
3539 		}
3540 	}
3541 	ip = mtod(m, struct ip *);
3542 	if (ip == NULL) goto bad;
3543 
3544 	if (ip->ip_v != IPVERSION) {
3545 		KMOD_IPSTAT_INC(ips_badvers);
3546 		goto bad;
3547 	}
3548 	hlen = ip->ip_hl << 2;
3549 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3550 		KMOD_IPSTAT_INC(ips_badhlen);
3551 		goto bad;
3552 	}
3553 	if (hlen > m->m_len) {
3554 		if ((m = m_pullup(m, hlen)) == NULL) {
3555 			KMOD_IPSTAT_INC(ips_badhlen);
3556 			goto bad;
3557 		}
3558 		ip = mtod(m, struct ip *);
3559 		if (ip == NULL) goto bad;
3560 	}
3561 
3562 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3563 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3564 	} else {
3565 		if (hlen == sizeof(struct ip)) {
3566 			sum = in_cksum_hdr(ip);
3567 		} else {
3568 			sum = in_cksum(m, hlen);
3569 		}
3570 	}
3571 	if (sum) {
3572 		KMOD_IPSTAT_INC(ips_badsum);
3573 		goto bad;
3574 	}
3575 
3576 	/* Retrieve the packet length. */
3577 	len = ntohs(ip->ip_len);
3578 
3579 	/*
3580 	 * Check for additional length bogosity
3581 	 */
3582 	if (len < hlen) {
3583 		KMOD_IPSTAT_INC(ips_badlen);
3584 		goto bad;
3585 	}
3586 
3587 	/*
3588 	 * Check that the amount of data in the buffers
3589 	 * is as at least much as the IP header would have us expect.
3590 	 * Drop packet if shorter than we expect.
3591 	 */
3592 	if (m->m_pkthdr.len < len) {
3593 		KMOD_IPSTAT_INC(ips_tooshort);
3594 		goto bad;
3595 	}
3596 
3597 	/* Checks out, proceed */
3598 	*mp = m;
3599 	return (0);
3600 
3601 bad:
3602 	*mp = m;
3603 	return (-1);
3604 }
3605 
3606 #ifdef INET6
3607 /*
3608  * Same as above, but for IPv6.
3609  * Cut-and-pasted from ip6_input.c.
3610  * XXX Should we update ip6stat, or not?
3611  */
3612 static int
3613 bridge_ip6_checkbasic(struct mbuf **mp)
3614 {
3615 	struct mbuf *m = *mp;
3616 	struct ip6_hdr *ip6;
3617 
3618 	/*
3619 	 * If the IPv6 header is not aligned, slurp it up into a new
3620 	 * mbuf with space for link headers, in the event we forward
3621 	 * it.  Otherwise, if it is aligned, make sure the entire base
3622 	 * IPv6 header is in the first mbuf of the chain.
3623 	 */
3624 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3625 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3626 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3627 			    (max_linkhdr + 3) & ~3)) == NULL) {
3628 			/* XXXJRT new stat, please */
3629 			IP6STAT_INC(ip6s_toosmall);
3630 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3631 			goto bad;
3632 		}
3633 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3634 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3635 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3636 			IP6STAT_INC(ip6s_toosmall);
3637 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3638 			goto bad;
3639 		}
3640 	}
3641 
3642 	ip6 = mtod(m, struct ip6_hdr *);
3643 
3644 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3645 		IP6STAT_INC(ip6s_badvers);
3646 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3647 		goto bad;
3648 	}
3649 
3650 	/* Checks out, proceed */
3651 	*mp = m;
3652 	return (0);
3653 
3654 bad:
3655 	*mp = m;
3656 	return (-1);
3657 }
3658 #endif /* INET6 */
3659 
3660 /*
3661  * bridge_fragment:
3662  *
3663  *	Fragment mbuf chain in multiple packets and prepend ethernet header.
3664  */
3665 static int
3666 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3667     int snap, struct llc *llc)
3668 {
3669 	struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3670 	struct ip *ip;
3671 	int error = -1;
3672 
3673 	if (m->m_len < sizeof(struct ip) &&
3674 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3675 		goto dropit;
3676 	ip = mtod(m, struct ip *);
3677 
3678 	m->m_pkthdr.csum_flags |= CSUM_IP;
3679 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3680 	if (error)
3681 		goto dropit;
3682 
3683 	/*
3684 	 * Walk the chain and re-add the Ethernet header for
3685 	 * each mbuf packet.
3686 	 */
3687 	for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3688 		nextpkt = mcur->m_nextpkt;
3689 		mcur->m_nextpkt = NULL;
3690 		if (snap) {
3691 			M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3692 			if (mcur == NULL) {
3693 				error = ENOBUFS;
3694 				if (mprev != NULL)
3695 					mprev->m_nextpkt = nextpkt;
3696 				goto dropit;
3697 			}
3698 			bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3699 		}
3700 
3701 		M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3702 		if (mcur == NULL) {
3703 			error = ENOBUFS;
3704 			if (mprev != NULL)
3705 				mprev->m_nextpkt = nextpkt;
3706 			goto dropit;
3707 		}
3708 		bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3709 
3710 		/*
3711 		 * The previous two M_PREPEND could have inserted one or two
3712 		 * mbufs in front so we have to update the previous packet's
3713 		 * m_nextpkt.
3714 		 */
3715 		mcur->m_nextpkt = nextpkt;
3716 		if (mprev != NULL)
3717 			mprev->m_nextpkt = mcur;
3718 		else {
3719 			/* The first mbuf in the original chain needs to be
3720 			 * updated. */
3721 			*mp = mcur;
3722 		}
3723 		mprev = mcur;
3724 	}
3725 
3726 	KMOD_IPSTAT_INC(ips_fragmented);
3727 	return (error);
3728 
3729 dropit:
3730 	for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3731 		m = mcur->m_nextpkt;
3732 		m_freem(mcur);
3733 	}
3734 	return (error);
3735 }
3736 
3737 static void
3738 bridge_linkstate(struct ifnet *ifp)
3739 {
3740 	struct bridge_softc *sc = ifp->if_bridge;
3741 	struct bridge_iflist *bif;
3742 	struct epoch_tracker et;
3743 
3744 	NET_EPOCH_ENTER(et);
3745 
3746 	bif = bridge_lookup_member_if(sc, ifp);
3747 	if (bif == NULL) {
3748 		NET_EPOCH_EXIT(et);
3749 		return;
3750 	}
3751 	bridge_linkcheck(sc);
3752 
3753 	bstp_linkstate(&bif->bif_stp);
3754 
3755 	NET_EPOCH_EXIT(et);
3756 }
3757 
3758 static void
3759 bridge_linkcheck(struct bridge_softc *sc)
3760 {
3761 	struct bridge_iflist *bif;
3762 	int new_link, hasls;
3763 
3764 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3765 
3766 	new_link = LINK_STATE_DOWN;
3767 	hasls = 0;
3768 	/* Our link is considered up if at least one of our ports is active */
3769 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3770 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3771 			hasls++;
3772 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3773 			new_link = LINK_STATE_UP;
3774 			break;
3775 		}
3776 	}
3777 	if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3778 		/* If no interfaces support link-state then we default to up */
3779 		new_link = LINK_STATE_UP;
3780 	}
3781 	if_link_state_change(sc->sc_ifp, new_link);
3782 }
3783