xref: /freebsd/sys/net/if_bridge.c (revision 657729a89dd578d8cfc70d6616f5c65a48a8b33a)
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 *, char *, size_t,
290 		    struct ifc_data *, struct ifnet **);
291 static int	bridge_clone_destroy(struct if_clone *, struct ifnet *, uint32_t);
292 
293 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
294 static void	bridge_mutecaps(struct bridge_softc *);
295 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
296 		    int);
297 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
298 static void	bridge_init(void *);
299 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
300 static void	bridge_stop(struct ifnet *, int);
301 static int	bridge_transmit(struct ifnet *, struct mbuf *);
302 #ifdef ALTQ
303 static void	bridge_altq_start(if_t);
304 static int	bridge_altq_transmit(if_t, struct mbuf *);
305 #endif
306 static void	bridge_qflush(struct ifnet *);
307 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
308 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
309 		    struct rtentry *);
310 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
311 		    struct mbuf *);
312 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
313 
314 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
315 		    struct mbuf *m);
316 
317 static void	bridge_timer(void *);
318 
319 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
320 		    struct mbuf *, int);
321 static void	bridge_span(struct bridge_softc *, struct mbuf *);
322 
323 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
324 		    uint16_t, struct bridge_iflist *, int, uint8_t);
325 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
326 		    uint16_t);
327 static void	bridge_rttrim(struct bridge_softc *);
328 static void	bridge_rtage(struct bridge_softc *);
329 static void	bridge_rtflush(struct bridge_softc *, int);
330 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
331 		    uint16_t);
332 
333 static void	bridge_rtable_init(struct bridge_softc *);
334 static void	bridge_rtable_fini(struct bridge_softc *);
335 
336 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
337 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
338 		    const uint8_t *, uint16_t);
339 static int	bridge_rtnode_insert(struct bridge_softc *,
340 		    struct bridge_rtnode *);
341 static void	bridge_rtnode_destroy(struct bridge_softc *,
342 		    struct bridge_rtnode *);
343 static void	bridge_rtable_expire(struct ifnet *, int);
344 static void	bridge_state_change(struct ifnet *, int);
345 
346 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
347 		    const char *name);
348 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
349 		    struct ifnet *ifp);
350 static void	bridge_delete_member(struct bridge_softc *,
351 		    struct bridge_iflist *, int);
352 static void	bridge_delete_span(struct bridge_softc *,
353 		    struct bridge_iflist *);
354 
355 static int	bridge_ioctl_add(struct bridge_softc *, void *);
356 static int	bridge_ioctl_del(struct bridge_softc *, void *);
357 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
358 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
359 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
360 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
361 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
362 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
363 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
364 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
365 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
366 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
367 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
368 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
369 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
370 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
371 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
372 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
373 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
374 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
375 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
376 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
377 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
378 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
379 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
380 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
381 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
382 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
383 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
384 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
385 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
386 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
387 		    int);
388 static int	bridge_ip_checkbasic(struct mbuf **mp);
389 #ifdef INET6
390 static int	bridge_ip6_checkbasic(struct mbuf **mp);
391 #endif /* INET6 */
392 static int	bridge_fragment(struct ifnet *, struct mbuf **mp,
393 		    struct ether_header *, int, struct llc *);
394 static void	bridge_linkstate(struct ifnet *ifp);
395 static void	bridge_linkcheck(struct bridge_softc *sc);
396 
397 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
398 #define	VLANTAGOF(_m)	\
399     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
400 
401 static struct bstp_cb_ops bridge_ops = {
402 	.bcb_state = bridge_state_change,
403 	.bcb_rtage = bridge_rtable_expire
404 };
405 
406 SYSCTL_DECL(_net_link);
407 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
408     "Bridge");
409 
410 /* only pass IP[46] packets when pfil is enabled */
411 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
412 #define	V_pfil_onlyip	VNET(pfil_onlyip)
413 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
414     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
415     "Only pass IP packets when pfil is enabled");
416 
417 /* run pfil hooks on the bridge interface */
418 VNET_DEFINE_STATIC(int, pfil_bridge) = 0;
419 #define	V_pfil_bridge	VNET(pfil_bridge)
420 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
421     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
422     "Packet filter on the bridge interface");
423 
424 /* layer2 filter with ipfw */
425 VNET_DEFINE_STATIC(int, pfil_ipfw);
426 #define	V_pfil_ipfw	VNET(pfil_ipfw)
427 
428 /* layer2 ARP filter with ipfw */
429 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
430 #define	V_pfil_ipfw_arp	VNET(pfil_ipfw_arp)
431 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
432     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
433     "Filter ARP packets through IPFW layer2");
434 
435 /* run pfil hooks on the member interface */
436 VNET_DEFINE_STATIC(int, pfil_member) = 0;
437 #define	V_pfil_member	VNET(pfil_member)
438 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
439     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
440     "Packet filter on the member interface");
441 
442 /* run pfil hooks on the physical interface for locally destined packets */
443 VNET_DEFINE_STATIC(int, pfil_local_phys);
444 #define	V_pfil_local_phys	VNET(pfil_local_phys)
445 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
446     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
447     "Packet filter on the physical interface for locally destined packets");
448 
449 /* log STP state changes */
450 VNET_DEFINE_STATIC(int, log_stp);
451 #define	V_log_stp	VNET(log_stp)
452 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
453     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
454     "Log STP state changes");
455 
456 /* share MAC with first bridge member */
457 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
458 #define	V_bridge_inherit_mac	VNET(bridge_inherit_mac)
459 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
460     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
461     "Inherit MAC address from the first bridge member");
462 
463 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
464 #define	V_allow_llz_overlap	VNET(allow_llz_overlap)
465 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
466     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
467     "Allow overlap of link-local scope "
468     "zones of a bridge interface and the member interfaces");
469 
470 struct bridge_control {
471 	int	(*bc_func)(struct bridge_softc *, void *);
472 	int	bc_argsize;
473 	int	bc_flags;
474 };
475 
476 #define	BC_F_COPYIN		0x01	/* copy arguments in */
477 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
478 #define	BC_F_SUSER		0x04	/* do super-user check */
479 
480 const struct bridge_control bridge_control_table[] = {
481 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
482 	  BC_F_COPYIN|BC_F_SUSER },
483 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
484 	  BC_F_COPYIN|BC_F_SUSER },
485 
486 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
487 	  BC_F_COPYIN|BC_F_COPYOUT },
488 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
489 	  BC_F_COPYIN|BC_F_SUSER },
490 
491 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
492 	  BC_F_COPYIN|BC_F_SUSER },
493 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
494 	  BC_F_COPYOUT },
495 
496 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
497 	  BC_F_COPYIN|BC_F_COPYOUT },
498 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
499 	  BC_F_COPYIN|BC_F_COPYOUT },
500 
501 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
502 	  BC_F_COPYIN|BC_F_SUSER },
503 
504 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
505 	  BC_F_COPYIN|BC_F_SUSER },
506 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
507 	  BC_F_COPYOUT },
508 
509 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
510 	  BC_F_COPYIN|BC_F_SUSER },
511 
512 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
513 	  BC_F_COPYIN|BC_F_SUSER },
514 
515 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
516 	  BC_F_COPYOUT },
517 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
518 	  BC_F_COPYIN|BC_F_SUSER },
519 
520 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
521 	  BC_F_COPYOUT },
522 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
523 	  BC_F_COPYIN|BC_F_SUSER },
524 
525 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
526 	  BC_F_COPYOUT },
527 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
528 	  BC_F_COPYIN|BC_F_SUSER },
529 
530 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
531 	  BC_F_COPYOUT },
532 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
533 	  BC_F_COPYIN|BC_F_SUSER },
534 
535 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
536 	  BC_F_COPYIN|BC_F_SUSER },
537 
538 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
539 	  BC_F_COPYIN|BC_F_SUSER },
540 
541 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
542 	  BC_F_COPYIN|BC_F_SUSER },
543 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
544 	  BC_F_COPYIN|BC_F_SUSER },
545 
546 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
547 	  BC_F_COPYOUT },
548 
549 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
550 	  BC_F_COPYOUT },
551 
552 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
553 	  BC_F_COPYIN|BC_F_COPYOUT },
554 
555 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
556 	  BC_F_COPYIN|BC_F_SUSER },
557 
558 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
559 	  BC_F_COPYIN|BC_F_SUSER },
560 
561 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
562 	  BC_F_COPYIN|BC_F_SUSER },
563 
564 };
565 const int bridge_control_table_size = nitems(bridge_control_table);
566 
567 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
568 #define	V_bridge_list	VNET(bridge_list)
569 #define	BRIDGE_LIST_LOCK_INIT(x)	sx_init(&V_bridge_list_sx,	\
570 					    "if_bridge list")
571 #define	BRIDGE_LIST_LOCK_DESTROY(x)	sx_destroy(&V_bridge_list_sx)
572 #define	BRIDGE_LIST_LOCK(x)		sx_xlock(&V_bridge_list_sx)
573 #define	BRIDGE_LIST_UNLOCK(x)		sx_xunlock(&V_bridge_list_sx)
574 
575 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
576 #define	V_bridge_cloner	VNET(bridge_cloner)
577 
578 static const char bridge_name[] = "bridge";
579 
580 static void
581 vnet_bridge_init(const void *unused __unused)
582 {
583 
584 	V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
585 	    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
586 	    UMA_ALIGN_PTR, 0);
587 	BRIDGE_LIST_LOCK_INIT();
588 	LIST_INIT(&V_bridge_list);
589 
590 	struct if_clone_addreq req = {
591 		.create_f = bridge_clone_create,
592 		.destroy_f = bridge_clone_destroy,
593 		.flags = IFC_F_AUTOUNIT,
594 	};
595 	V_bridge_cloner = ifc_attach_cloner(bridge_name, &req);
596 }
597 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
598     vnet_bridge_init, NULL);
599 
600 static void
601 vnet_bridge_uninit(const void *unused __unused)
602 {
603 
604 	ifc_detach_cloner(V_bridge_cloner);
605 	V_bridge_cloner = NULL;
606 	BRIDGE_LIST_LOCK_DESTROY();
607 
608 	/* Callbacks may use the UMA zone. */
609 	NET_EPOCH_DRAIN_CALLBACKS();
610 
611 	uma_zdestroy(V_bridge_rtnode_zone);
612 }
613 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
614     vnet_bridge_uninit, NULL);
615 
616 static int
617 bridge_modevent(module_t mod, int type, void *data)
618 {
619 
620 	switch (type) {
621 	case MOD_LOAD:
622 		bridge_dn_p = bridge_dummynet;
623 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
624 		    ifnet_departure_event, bridge_ifdetach, NULL,
625 		    EVENTHANDLER_PRI_ANY);
626 		break;
627 	case MOD_UNLOAD:
628 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
629 		    bridge_detach_cookie);
630 		bridge_dn_p = NULL;
631 		break;
632 	default:
633 		return (EOPNOTSUPP);
634 	}
635 	return (0);
636 }
637 
638 static moduledata_t bridge_mod = {
639 	"if_bridge",
640 	bridge_modevent,
641 	0
642 };
643 
644 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
645 MODULE_VERSION(if_bridge, 1);
646 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
647 
648 /*
649  * handler for net.link.bridge.ipfw
650  */
651 static int
652 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
653 {
654 	int enable = V_pfil_ipfw;
655 	int error;
656 
657 	error = sysctl_handle_int(oidp, &enable, 0, req);
658 	enable &= 1;
659 
660 	if (enable != V_pfil_ipfw) {
661 		V_pfil_ipfw = enable;
662 
663 		/*
664 		 * Disable pfil so that ipfw doesnt run twice, if the user
665 		 * really wants both then they can re-enable pfil_bridge and/or
666 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
667 		 * layer2 type.
668 		 */
669 		if (V_pfil_ipfw) {
670 			V_pfil_onlyip = 0;
671 			V_pfil_bridge = 0;
672 			V_pfil_member = 0;
673 		}
674 	}
675 
676 	return (error);
677 }
678 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
679     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT,
680     &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
681     "Layer2 filter with IPFW");
682 
683 #ifdef VIMAGE
684 static void
685 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg)
686 {
687 	struct bridge_softc *sc = ifp->if_softc;
688 	struct bridge_iflist *bif;
689 
690 	BRIDGE_LOCK(sc);
691 
692 	while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
693 		bridge_delete_member(sc, bif, 0);
694 
695 	while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
696 		bridge_delete_span(sc, bif);
697 	}
698 
699 	BRIDGE_UNLOCK(sc);
700 
701 	ether_reassign(ifp, newvnet, arg);
702 }
703 #endif
704 
705 /*
706  * bridge_clone_create:
707  *
708  *	Create a new bridge instance.
709  */
710 static int
711 bridge_clone_create(struct if_clone *ifc, char *name, size_t len,
712     struct ifc_data *ifd, struct ifnet **ifpp)
713 {
714 	struct bridge_softc *sc;
715 	struct ifnet *ifp;
716 
717 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
718 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
719 	if (ifp == NULL) {
720 		free(sc, M_DEVBUF);
721 		return (ENOSPC);
722 	}
723 
724 	BRIDGE_LOCK_INIT(sc);
725 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
726 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
727 
728 	/* Initialize our routing table. */
729 	bridge_rtable_init(sc);
730 
731 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
732 
733 	CK_LIST_INIT(&sc->sc_iflist);
734 	CK_LIST_INIT(&sc->sc_spanlist);
735 
736 	ifp->if_softc = sc;
737 	if_initname(ifp, bridge_name, ifd->unit);
738 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
739 	ifp->if_ioctl = bridge_ioctl;
740 #ifdef ALTQ
741 	ifp->if_start = bridge_altq_start;
742 	ifp->if_transmit = bridge_altq_transmit;
743 	IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
744 	ifp->if_snd.ifq_drv_maxlen = 0;
745 	IFQ_SET_READY(&ifp->if_snd);
746 #else
747 	ifp->if_transmit = bridge_transmit;
748 #endif
749 	ifp->if_qflush = bridge_qflush;
750 	ifp->if_init = bridge_init;
751 	ifp->if_type = IFT_BRIDGE;
752 
753 	ether_gen_addr(ifp, &sc->sc_defaddr);
754 
755 	bstp_attach(&sc->sc_stp, &bridge_ops);
756 	ether_ifattach(ifp, sc->sc_defaddr.octet);
757 	/* Now undo some of the damage... */
758 	ifp->if_baudrate = 0;
759 	ifp->if_type = IFT_BRIDGE;
760 #ifdef VIMAGE
761 	ifp->if_reassign = bridge_reassign;
762 #endif
763 
764 	BRIDGE_LIST_LOCK();
765 	LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
766 	BRIDGE_LIST_UNLOCK();
767 	*ifpp = ifp;
768 
769 	return (0);
770 }
771 
772 static void
773 bridge_clone_destroy_cb(struct epoch_context *ctx)
774 {
775 	struct bridge_softc *sc;
776 
777 	sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
778 
779 	BRIDGE_LOCK_DESTROY(sc);
780 	free(sc, M_DEVBUF);
781 }
782 
783 /*
784  * bridge_clone_destroy:
785  *
786  *	Destroy a bridge instance.
787  */
788 static int
789 bridge_clone_destroy(struct if_clone *ifc, struct ifnet *ifp, uint32_t flags)
790 {
791 	struct bridge_softc *sc = ifp->if_softc;
792 	struct bridge_iflist *bif;
793 	struct epoch_tracker et;
794 
795 	BRIDGE_LOCK(sc);
796 
797 	bridge_stop(ifp, 1);
798 	ifp->if_flags &= ~IFF_UP;
799 
800 	while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
801 		bridge_delete_member(sc, bif, 0);
802 
803 	while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
804 		bridge_delete_span(sc, bif);
805 	}
806 
807 	/* Tear down the routing table. */
808 	bridge_rtable_fini(sc);
809 
810 	BRIDGE_UNLOCK(sc);
811 
812 	NET_EPOCH_ENTER(et);
813 
814 	callout_drain(&sc->sc_brcallout);
815 
816 	BRIDGE_LIST_LOCK();
817 	LIST_REMOVE(sc, sc_list);
818 	BRIDGE_LIST_UNLOCK();
819 
820 	bstp_detach(&sc->sc_stp);
821 #ifdef ALTQ
822 	IFQ_PURGE(&ifp->if_snd);
823 #endif
824 	NET_EPOCH_EXIT(et);
825 
826 	ether_ifdetach(ifp);
827 	if_free(ifp);
828 
829 	NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx);
830 
831 	return (0);
832 }
833 
834 /*
835  * bridge_ioctl:
836  *
837  *	Handle a control request from the operator.
838  */
839 static int
840 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
841 {
842 	struct bridge_softc *sc = ifp->if_softc;
843 	struct ifreq *ifr = (struct ifreq *)data;
844 	struct bridge_iflist *bif;
845 	struct thread *td = curthread;
846 	union {
847 		struct ifbreq ifbreq;
848 		struct ifbifconf ifbifconf;
849 		struct ifbareq ifbareq;
850 		struct ifbaconf ifbaconf;
851 		struct ifbrparam ifbrparam;
852 		struct ifbropreq ifbropreq;
853 	} args;
854 	struct ifdrv *ifd = (struct ifdrv *) data;
855 	const struct bridge_control *bc;
856 	int error = 0, oldmtu;
857 
858 	BRIDGE_LOCK(sc);
859 
860 	switch (cmd) {
861 	case SIOCADDMULTI:
862 	case SIOCDELMULTI:
863 		break;
864 
865 	case SIOCGDRVSPEC:
866 	case SIOCSDRVSPEC:
867 		if (ifd->ifd_cmd >= bridge_control_table_size) {
868 			error = EINVAL;
869 			break;
870 		}
871 		bc = &bridge_control_table[ifd->ifd_cmd];
872 
873 		if (cmd == SIOCGDRVSPEC &&
874 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
875 			error = EINVAL;
876 			break;
877 		}
878 		else if (cmd == SIOCSDRVSPEC &&
879 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
880 			error = EINVAL;
881 			break;
882 		}
883 
884 		if (bc->bc_flags & BC_F_SUSER) {
885 			error = priv_check(td, PRIV_NET_BRIDGE);
886 			if (error)
887 				break;
888 		}
889 
890 		if (ifd->ifd_len != bc->bc_argsize ||
891 		    ifd->ifd_len > sizeof(args)) {
892 			error = EINVAL;
893 			break;
894 		}
895 
896 		bzero(&args, sizeof(args));
897 		if (bc->bc_flags & BC_F_COPYIN) {
898 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
899 			if (error)
900 				break;
901 		}
902 
903 		oldmtu = ifp->if_mtu;
904 		error = (*bc->bc_func)(sc, &args);
905 		if (error)
906 			break;
907 
908 		/*
909 		 * Bridge MTU may change during addition of the first port.
910 		 * If it did, do network layer specific procedure.
911 		 */
912 		if (ifp->if_mtu != oldmtu) {
913 #ifdef INET6
914 			nd6_setmtu(ifp);
915 #endif
916 			rt_updatemtu(ifp);
917 		}
918 
919 		if (bc->bc_flags & BC_F_COPYOUT)
920 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
921 
922 		break;
923 
924 	case SIOCSIFFLAGS:
925 		if (!(ifp->if_flags & IFF_UP) &&
926 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
927 			/*
928 			 * If interface is marked down and it is running,
929 			 * then stop and disable it.
930 			 */
931 			bridge_stop(ifp, 1);
932 		} else if ((ifp->if_flags & IFF_UP) &&
933 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
934 			/*
935 			 * If interface is marked up and it is stopped, then
936 			 * start it.
937 			 */
938 			BRIDGE_UNLOCK(sc);
939 			(*ifp->if_init)(sc);
940 			BRIDGE_LOCK(sc);
941 		}
942 		break;
943 
944 	case SIOCSIFMTU:
945 		oldmtu = sc->sc_ifp->if_mtu;
946 
947 		if (ifr->ifr_mtu < 576) {
948 			error = EINVAL;
949 			break;
950 		}
951 		if (CK_LIST_EMPTY(&sc->sc_iflist)) {
952 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
953 			break;
954 		}
955 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
956 			error = (*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
957 			    SIOCSIFMTU, (caddr_t)ifr);
958 			if (error != 0) {
959 				log(LOG_NOTICE, "%s: invalid MTU: %u for"
960 				    " member %s\n", sc->sc_ifp->if_xname,
961 				    ifr->ifr_mtu,
962 				    bif->bif_ifp->if_xname);
963 				error = EINVAL;
964 				break;
965 			}
966 		}
967 		if (error) {
968 			/* Restore the previous MTU on all member interfaces. */
969 			ifr->ifr_mtu = oldmtu;
970 			CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
971 				(*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
972 				    SIOCSIFMTU, (caddr_t)ifr);
973 			}
974 		} else {
975 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
976 		}
977 		break;
978 	default:
979 		/*
980 		 * drop the lock as ether_ioctl() will call bridge_start() and
981 		 * cause the lock to be recursed.
982 		 */
983 		BRIDGE_UNLOCK(sc);
984 		error = ether_ioctl(ifp, cmd, data);
985 		BRIDGE_LOCK(sc);
986 		break;
987 	}
988 
989 	BRIDGE_UNLOCK(sc);
990 
991 	return (error);
992 }
993 
994 /*
995  * bridge_mutecaps:
996  *
997  *	Clear or restore unwanted capabilities on the member interface
998  */
999 static void
1000 bridge_mutecaps(struct bridge_softc *sc)
1001 {
1002 	struct bridge_iflist *bif;
1003 	int enabled, mask;
1004 
1005 	BRIDGE_LOCK_ASSERT(sc);
1006 
1007 	/* Initial bitmask of capabilities to test */
1008 	mask = BRIDGE_IFCAPS_MASK;
1009 
1010 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1011 		/* Every member must support it or its disabled */
1012 		mask &= bif->bif_savedcaps;
1013 	}
1014 
1015 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1016 		enabled = bif->bif_ifp->if_capenable;
1017 		enabled &= ~BRIDGE_IFCAPS_STRIP;
1018 		/* strip off mask bits and enable them again if allowed */
1019 		enabled &= ~BRIDGE_IFCAPS_MASK;
1020 		enabled |= mask;
1021 		bridge_set_ifcap(sc, bif, enabled);
1022 	}
1023 }
1024 
1025 static void
1026 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1027 {
1028 	struct ifnet *ifp = bif->bif_ifp;
1029 	struct ifreq ifr;
1030 	int error, mask, stuck;
1031 
1032 	bzero(&ifr, sizeof(ifr));
1033 	ifr.ifr_reqcap = set;
1034 
1035 	if (ifp->if_capenable != set) {
1036 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1037 		if (error)
1038 			if_printf(sc->sc_ifp,
1039 			    "error setting capabilities on %s: %d\n",
1040 			    ifp->if_xname, error);
1041 		mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1042 		stuck = ifp->if_capenable & mask & ~set;
1043 		if (stuck != 0)
1044 			if_printf(sc->sc_ifp,
1045 			    "can't disable some capabilities on %s: 0x%x\n",
1046 			    ifp->if_xname, stuck);
1047 	}
1048 }
1049 
1050 /*
1051  * bridge_lookup_member:
1052  *
1053  *	Lookup a bridge member interface.
1054  */
1055 static struct bridge_iflist *
1056 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1057 {
1058 	struct bridge_iflist *bif;
1059 	struct ifnet *ifp;
1060 
1061 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1062 
1063 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1064 		ifp = bif->bif_ifp;
1065 		if (strcmp(ifp->if_xname, name) == 0)
1066 			return (bif);
1067 	}
1068 
1069 	return (NULL);
1070 }
1071 
1072 /*
1073  * bridge_lookup_member_if:
1074  *
1075  *	Lookup a bridge member interface by ifnet*.
1076  */
1077 static struct bridge_iflist *
1078 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1079 {
1080 	struct bridge_iflist *bif;
1081 
1082 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1083 
1084 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1085 		if (bif->bif_ifp == member_ifp)
1086 			return (bif);
1087 	}
1088 
1089 	return (NULL);
1090 }
1091 
1092 static void
1093 bridge_delete_member_cb(struct epoch_context *ctx)
1094 {
1095 	struct bridge_iflist *bif;
1096 
1097 	bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1098 
1099 	free(bif, M_DEVBUF);
1100 }
1101 
1102 /*
1103  * bridge_delete_member:
1104  *
1105  *	Delete the specified member interface.
1106  */
1107 static void
1108 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1109     int gone)
1110 {
1111 	struct ifnet *ifs = bif->bif_ifp;
1112 	struct ifnet *fif = NULL;
1113 	struct bridge_iflist *bifl;
1114 
1115 	BRIDGE_LOCK_ASSERT(sc);
1116 
1117 	if (bif->bif_flags & IFBIF_STP)
1118 		bstp_disable(&bif->bif_stp);
1119 
1120 	ifs->if_bridge = NULL;
1121 	CK_LIST_REMOVE(bif, bif_next);
1122 
1123 	/*
1124 	 * If removing the interface that gave the bridge its mac address, set
1125 	 * the mac address of the bridge to the address of the next member, or
1126 	 * to its default address if no members are left.
1127 	 */
1128 	if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1129 		if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1130 			bcopy(&sc->sc_defaddr,
1131 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1132 			sc->sc_ifaddr = NULL;
1133 		} else {
1134 			bifl = CK_LIST_FIRST(&sc->sc_iflist);
1135 			fif = bifl->bif_ifp;
1136 			bcopy(IF_LLADDR(fif),
1137 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1138 			sc->sc_ifaddr = fif;
1139 		}
1140 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1141 	}
1142 
1143 	bridge_linkcheck(sc);
1144 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
1145 	BRIDGE_RT_LOCK(sc);
1146 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1147 	BRIDGE_RT_UNLOCK(sc);
1148 	KASSERT(bif->bif_addrcnt == 0,
1149 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1150 
1151 	ifs->if_bridge_output = NULL;
1152 	ifs->if_bridge_input = NULL;
1153 	ifs->if_bridge_linkstate = NULL;
1154 	if (!gone) {
1155 		switch (ifs->if_type) {
1156 		case IFT_ETHER:
1157 		case IFT_L2VLAN:
1158 			/*
1159 			 * Take the interface out of promiscuous mode, but only
1160 			 * if it was promiscuous in the first place. It might
1161 			 * not be if we're in the bridge_ioctl_add() error path.
1162 			 */
1163 			if (ifs->if_flags & IFF_PROMISC)
1164 				(void) ifpromisc(ifs, 0);
1165 			break;
1166 
1167 		case IFT_GIF:
1168 			break;
1169 
1170 		default:
1171 #ifdef DIAGNOSTIC
1172 			panic("bridge_delete_member: impossible");
1173 #endif
1174 			break;
1175 		}
1176 		/* reneable any interface capabilities */
1177 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1178 	}
1179 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1180 
1181 	NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1182 }
1183 
1184 /*
1185  * bridge_delete_span:
1186  *
1187  *	Delete the specified span interface.
1188  */
1189 static void
1190 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1191 {
1192 	BRIDGE_LOCK_ASSERT(sc);
1193 
1194 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1195 	    ("%s: not a span interface", __func__));
1196 
1197 	CK_LIST_REMOVE(bif, bif_next);
1198 
1199 	NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1200 }
1201 
1202 static int
1203 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1204 {
1205 	struct ifbreq *req = arg;
1206 	struct bridge_iflist *bif = NULL;
1207 	struct ifnet *ifs;
1208 	int error = 0;
1209 
1210 	ifs = ifunit(req->ifbr_ifsname);
1211 	if (ifs == NULL)
1212 		return (ENOENT);
1213 	if (ifs->if_ioctl == NULL)	/* must be supported */
1214 		return (EINVAL);
1215 
1216 	/* If it's in the span list, it can't be a member. */
1217 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1218 		if (ifs == bif->bif_ifp)
1219 			return (EBUSY);
1220 
1221 	if (ifs->if_bridge == sc)
1222 		return (EEXIST);
1223 
1224 	if (ifs->if_bridge != NULL)
1225 		return (EBUSY);
1226 
1227 	switch (ifs->if_type) {
1228 	case IFT_ETHER:
1229 	case IFT_L2VLAN:
1230 	case IFT_GIF:
1231 		/* permitted interface types */
1232 		break;
1233 	default:
1234 		return (EINVAL);
1235 	}
1236 
1237 #ifdef INET6
1238 	/*
1239 	 * Two valid inet6 addresses with link-local scope must not be
1240 	 * on the parent interface and the member interfaces at the
1241 	 * same time.  This restriction is needed to prevent violation
1242 	 * of link-local scope zone.  Attempts to add a member
1243 	 * interface which has inet6 addresses when the parent has
1244 	 * inet6 triggers removal of all inet6 addresses on the member
1245 	 * interface.
1246 	 */
1247 
1248 	/* Check if the parent interface has a link-local scope addr. */
1249 	if (V_allow_llz_overlap == 0 &&
1250 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1251 		/*
1252 		 * If any, remove all inet6 addresses from the member
1253 		 * interfaces.
1254 		 */
1255 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1256  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1257 				in6_ifdetach(bif->bif_ifp);
1258 				if_printf(sc->sc_ifp,
1259 				    "IPv6 addresses on %s have been removed "
1260 				    "before adding it as a member to prevent "
1261 				    "IPv6 address scope violation.\n",
1262 				    bif->bif_ifp->if_xname);
1263 			}
1264 		}
1265 		if (in6ifa_llaonifp(ifs)) {
1266 			in6_ifdetach(ifs);
1267 			if_printf(sc->sc_ifp,
1268 			    "IPv6 addresses on %s have been removed "
1269 			    "before adding it as a member to prevent "
1270 			    "IPv6 address scope violation.\n",
1271 			    ifs->if_xname);
1272 		}
1273 	}
1274 #endif
1275 	/* Allow the first Ethernet member to define the MTU */
1276 	if (CK_LIST_EMPTY(&sc->sc_iflist))
1277 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1278 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1279 		struct ifreq ifr;
1280 
1281 		snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s",
1282 		    ifs->if_xname);
1283 		ifr.ifr_mtu = sc->sc_ifp->if_mtu;
1284 
1285 		error = (*ifs->if_ioctl)(ifs,
1286 		    SIOCSIFMTU, (caddr_t)&ifr);
1287 		if (error != 0) {
1288 			log(LOG_NOTICE, "%s: invalid MTU: %u for"
1289 			    " new member %s\n", sc->sc_ifp->if_xname,
1290 			    ifr.ifr_mtu,
1291 			    ifs->if_xname);
1292 			return (EINVAL);
1293 		}
1294 	}
1295 
1296 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1297 	if (bif == NULL)
1298 		return (ENOMEM);
1299 
1300 	bif->bif_ifp = ifs;
1301 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1302 	bif->bif_savedcaps = ifs->if_capenable;
1303 
1304 	/*
1305 	 * Assign the interface's MAC address to the bridge if it's the first
1306 	 * member and the MAC address of the bridge has not been changed from
1307 	 * the default randomly generated one.
1308 	 */
1309 	if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1310 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1311 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1312 		sc->sc_ifaddr = ifs;
1313 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1314 	}
1315 
1316 	ifs->if_bridge = sc;
1317 	ifs->if_bridge_output = bridge_output;
1318 	ifs->if_bridge_input = bridge_input;
1319 	ifs->if_bridge_linkstate = bridge_linkstate;
1320 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1321 	/*
1322 	 * XXX: XLOCK HERE!?!
1323 	 *
1324 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1325 	 */
1326 	CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1327 
1328 	/* Set interface capabilities to the intersection set of all members */
1329 	bridge_mutecaps(sc);
1330 	bridge_linkcheck(sc);
1331 
1332 	/* Place the interface into promiscuous mode */
1333 	switch (ifs->if_type) {
1334 		case IFT_ETHER:
1335 		case IFT_L2VLAN:
1336 			error = ifpromisc(ifs, 1);
1337 			break;
1338 	}
1339 
1340 	if (error)
1341 		bridge_delete_member(sc, bif, 0);
1342 	return (error);
1343 }
1344 
1345 static int
1346 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1347 {
1348 	struct ifbreq *req = arg;
1349 	struct bridge_iflist *bif;
1350 
1351 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1352 	if (bif == NULL)
1353 		return (ENOENT);
1354 
1355 	bridge_delete_member(sc, bif, 0);
1356 
1357 	return (0);
1358 }
1359 
1360 static int
1361 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1362 {
1363 	struct ifbreq *req = arg;
1364 	struct bridge_iflist *bif;
1365 	struct bstp_port *bp;
1366 
1367 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1368 	if (bif == NULL)
1369 		return (ENOENT);
1370 
1371 	bp = &bif->bif_stp;
1372 	req->ifbr_ifsflags = bif->bif_flags;
1373 	req->ifbr_state = bp->bp_state;
1374 	req->ifbr_priority = bp->bp_priority;
1375 	req->ifbr_path_cost = bp->bp_path_cost;
1376 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1377 	req->ifbr_proto = bp->bp_protover;
1378 	req->ifbr_role = bp->bp_role;
1379 	req->ifbr_stpflags = bp->bp_flags;
1380 	req->ifbr_addrcnt = bif->bif_addrcnt;
1381 	req->ifbr_addrmax = bif->bif_addrmax;
1382 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1383 
1384 	/* Copy STP state options as flags */
1385 	if (bp->bp_operedge)
1386 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1387 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1388 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1389 	if (bp->bp_ptp_link)
1390 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1391 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1392 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1393 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1394 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1395 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1396 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1397 	return (0);
1398 }
1399 
1400 static int
1401 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1402 {
1403 	struct epoch_tracker et;
1404 	struct ifbreq *req = arg;
1405 	struct bridge_iflist *bif;
1406 	struct bstp_port *bp;
1407 	int error;
1408 
1409 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1410 	if (bif == NULL)
1411 		return (ENOENT);
1412 	bp = &bif->bif_stp;
1413 
1414 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1415 		/* SPAN is readonly */
1416 		return (EINVAL);
1417 
1418 	NET_EPOCH_ENTER(et);
1419 
1420 	if (req->ifbr_ifsflags & IFBIF_STP) {
1421 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1422 			error = bstp_enable(&bif->bif_stp);
1423 			if (error) {
1424 				NET_EPOCH_EXIT(et);
1425 				return (error);
1426 			}
1427 		}
1428 	} else {
1429 		if ((bif->bif_flags & IFBIF_STP) != 0)
1430 			bstp_disable(&bif->bif_stp);
1431 	}
1432 
1433 	/* Pass on STP flags */
1434 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1435 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1436 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1437 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1438 
1439 	/* Save the bits relating to the bridge */
1440 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1441 
1442 	NET_EPOCH_EXIT(et);
1443 
1444 	return (0);
1445 }
1446 
1447 static int
1448 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1449 {
1450 	struct ifbrparam *param = arg;
1451 
1452 	sc->sc_brtmax = param->ifbrp_csize;
1453 	bridge_rttrim(sc);
1454 
1455 	return (0);
1456 }
1457 
1458 static int
1459 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1460 {
1461 	struct ifbrparam *param = arg;
1462 
1463 	param->ifbrp_csize = sc->sc_brtmax;
1464 
1465 	return (0);
1466 }
1467 
1468 static int
1469 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1470 {
1471 	struct ifbifconf *bifc = arg;
1472 	struct bridge_iflist *bif;
1473 	struct ifbreq breq;
1474 	char *buf, *outbuf;
1475 	int count, buflen, len, error = 0;
1476 
1477 	count = 0;
1478 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1479 		count++;
1480 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1481 		count++;
1482 
1483 	buflen = sizeof(breq) * count;
1484 	if (bifc->ifbic_len == 0) {
1485 		bifc->ifbic_len = buflen;
1486 		return (0);
1487 	}
1488 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1489 	if (outbuf == NULL)
1490 		return (ENOMEM);
1491 
1492 	count = 0;
1493 	buf = outbuf;
1494 	len = min(bifc->ifbic_len, buflen);
1495 	bzero(&breq, sizeof(breq));
1496 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1497 		if (len < sizeof(breq))
1498 			break;
1499 
1500 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1501 		    sizeof(breq.ifbr_ifsname));
1502 		/* Fill in the ifbreq structure */
1503 		error = bridge_ioctl_gifflags(sc, &breq);
1504 		if (error)
1505 			break;
1506 		memcpy(buf, &breq, sizeof(breq));
1507 		count++;
1508 		buf += sizeof(breq);
1509 		len -= sizeof(breq);
1510 	}
1511 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1512 		if (len < sizeof(breq))
1513 			break;
1514 
1515 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1516 		    sizeof(breq.ifbr_ifsname));
1517 		breq.ifbr_ifsflags = bif->bif_flags;
1518 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1519 		memcpy(buf, &breq, sizeof(breq));
1520 		count++;
1521 		buf += sizeof(breq);
1522 		len -= sizeof(breq);
1523 	}
1524 
1525 	bifc->ifbic_len = sizeof(breq) * count;
1526 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1527 	free(outbuf, M_TEMP);
1528 	return (error);
1529 }
1530 
1531 static int
1532 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1533 {
1534 	struct ifbaconf *bac = arg;
1535 	struct bridge_rtnode *brt;
1536 	struct ifbareq bareq;
1537 	char *buf, *outbuf;
1538 	int count, buflen, len, error = 0;
1539 
1540 	if (bac->ifbac_len == 0)
1541 		return (0);
1542 
1543 	count = 0;
1544 	CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1545 		count++;
1546 	buflen = sizeof(bareq) * count;
1547 
1548 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1549 	if (outbuf == NULL)
1550 		return (ENOMEM);
1551 
1552 	count = 0;
1553 	buf = outbuf;
1554 	len = min(bac->ifbac_len, buflen);
1555 	bzero(&bareq, sizeof(bareq));
1556 	CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1557 		if (len < sizeof(bareq))
1558 			goto out;
1559 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1560 		    sizeof(bareq.ifba_ifsname));
1561 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1562 		bareq.ifba_vlan = brt->brt_vlan;
1563 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1564 				time_uptime < brt->brt_expire)
1565 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1566 		else
1567 			bareq.ifba_expire = 0;
1568 		bareq.ifba_flags = brt->brt_flags;
1569 
1570 		memcpy(buf, &bareq, sizeof(bareq));
1571 		count++;
1572 		buf += sizeof(bareq);
1573 		len -= sizeof(bareq);
1574 	}
1575 out:
1576 	bac->ifbac_len = sizeof(bareq) * count;
1577 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1578 	free(outbuf, M_TEMP);
1579 	return (error);
1580 }
1581 
1582 static int
1583 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1584 {
1585 	struct ifbareq *req = arg;
1586 	struct bridge_iflist *bif;
1587 	struct epoch_tracker et;
1588 	int error;
1589 
1590 	NET_EPOCH_ENTER(et);
1591 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1592 	if (bif == NULL) {
1593 		NET_EPOCH_EXIT(et);
1594 		return (ENOENT);
1595 	}
1596 
1597 	/* bridge_rtupdate() may acquire the lock. */
1598 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1599 	    req->ifba_flags);
1600 	NET_EPOCH_EXIT(et);
1601 
1602 	return (error);
1603 }
1604 
1605 static int
1606 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1607 {
1608 	struct ifbrparam *param = arg;
1609 
1610 	sc->sc_brttimeout = param->ifbrp_ctime;
1611 	return (0);
1612 }
1613 
1614 static int
1615 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1616 {
1617 	struct ifbrparam *param = arg;
1618 
1619 	param->ifbrp_ctime = sc->sc_brttimeout;
1620 	return (0);
1621 }
1622 
1623 static int
1624 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1625 {
1626 	struct ifbareq *req = arg;
1627 
1628 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1629 }
1630 
1631 static int
1632 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1633 {
1634 	struct ifbreq *req = arg;
1635 
1636 	BRIDGE_RT_LOCK(sc);
1637 	bridge_rtflush(sc, req->ifbr_ifsflags);
1638 	BRIDGE_RT_UNLOCK(sc);
1639 
1640 	return (0);
1641 }
1642 
1643 static int
1644 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1645 {
1646 	struct ifbrparam *param = arg;
1647 	struct bstp_state *bs = &sc->sc_stp;
1648 
1649 	param->ifbrp_prio = bs->bs_bridge_priority;
1650 	return (0);
1651 }
1652 
1653 static int
1654 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1655 {
1656 	struct ifbrparam *param = arg;
1657 
1658 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1659 }
1660 
1661 static int
1662 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1663 {
1664 	struct ifbrparam *param = arg;
1665 	struct bstp_state *bs = &sc->sc_stp;
1666 
1667 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1668 	return (0);
1669 }
1670 
1671 static int
1672 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1673 {
1674 	struct ifbrparam *param = arg;
1675 
1676 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1677 }
1678 
1679 static int
1680 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1681 {
1682 	struct ifbrparam *param = arg;
1683 	struct bstp_state *bs = &sc->sc_stp;
1684 
1685 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1686 	return (0);
1687 }
1688 
1689 static int
1690 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1691 {
1692 	struct ifbrparam *param = arg;
1693 
1694 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1695 }
1696 
1697 static int
1698 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1699 {
1700 	struct ifbrparam *param = arg;
1701 	struct bstp_state *bs = &sc->sc_stp;
1702 
1703 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1704 	return (0);
1705 }
1706 
1707 static int
1708 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1709 {
1710 	struct ifbrparam *param = arg;
1711 
1712 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1713 }
1714 
1715 static int
1716 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1717 {
1718 	struct ifbreq *req = arg;
1719 	struct bridge_iflist *bif;
1720 
1721 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1722 	if (bif == NULL)
1723 		return (ENOENT);
1724 
1725 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1726 }
1727 
1728 static int
1729 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1730 {
1731 	struct ifbreq *req = arg;
1732 	struct bridge_iflist *bif;
1733 
1734 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1735 	if (bif == NULL)
1736 		return (ENOENT);
1737 
1738 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1739 }
1740 
1741 static int
1742 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1743 {
1744 	struct ifbreq *req = arg;
1745 	struct bridge_iflist *bif;
1746 
1747 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1748 	if (bif == NULL)
1749 		return (ENOENT);
1750 
1751 	bif->bif_addrmax = req->ifbr_addrmax;
1752 	return (0);
1753 }
1754 
1755 static int
1756 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1757 {
1758 	struct ifbreq *req = arg;
1759 	struct bridge_iflist *bif = NULL;
1760 	struct ifnet *ifs;
1761 
1762 	ifs = ifunit(req->ifbr_ifsname);
1763 	if (ifs == NULL)
1764 		return (ENOENT);
1765 
1766 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1767 		if (ifs == bif->bif_ifp)
1768 			return (EBUSY);
1769 
1770 	if (ifs->if_bridge != NULL)
1771 		return (EBUSY);
1772 
1773 	switch (ifs->if_type) {
1774 		case IFT_ETHER:
1775 		case IFT_GIF:
1776 		case IFT_L2VLAN:
1777 			break;
1778 		default:
1779 			return (EINVAL);
1780 	}
1781 
1782 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1783 	if (bif == NULL)
1784 		return (ENOMEM);
1785 
1786 	bif->bif_ifp = ifs;
1787 	bif->bif_flags = IFBIF_SPAN;
1788 
1789 	CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1790 
1791 	return (0);
1792 }
1793 
1794 static int
1795 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1796 {
1797 	struct ifbreq *req = arg;
1798 	struct bridge_iflist *bif;
1799 	struct ifnet *ifs;
1800 
1801 	ifs = ifunit(req->ifbr_ifsname);
1802 	if (ifs == NULL)
1803 		return (ENOENT);
1804 
1805 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1806 		if (ifs == bif->bif_ifp)
1807 			break;
1808 
1809 	if (bif == NULL)
1810 		return (ENOENT);
1811 
1812 	bridge_delete_span(sc, bif);
1813 
1814 	return (0);
1815 }
1816 
1817 static int
1818 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1819 {
1820 	struct ifbropreq *req = arg;
1821 	struct bstp_state *bs = &sc->sc_stp;
1822 	struct bstp_port *root_port;
1823 
1824 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1825 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1826 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1827 
1828 	root_port = bs->bs_root_port;
1829 	if (root_port == NULL)
1830 		req->ifbop_root_port = 0;
1831 	else
1832 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1833 
1834 	req->ifbop_holdcount = bs->bs_txholdcount;
1835 	req->ifbop_priority = bs->bs_bridge_priority;
1836 	req->ifbop_protocol = bs->bs_protover;
1837 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1838 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1839 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1840 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1841 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1842 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1843 
1844 	return (0);
1845 }
1846 
1847 static int
1848 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1849 {
1850 	struct ifbrparam *param = arg;
1851 
1852 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1853 	return (0);
1854 }
1855 
1856 static int
1857 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1858 {
1859 	struct ifbpstpconf *bifstp = arg;
1860 	struct bridge_iflist *bif;
1861 	struct bstp_port *bp;
1862 	struct ifbpstpreq bpreq;
1863 	char *buf, *outbuf;
1864 	int count, buflen, len, error = 0;
1865 
1866 	count = 0;
1867 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1868 		if ((bif->bif_flags & IFBIF_STP) != 0)
1869 			count++;
1870 	}
1871 
1872 	buflen = sizeof(bpreq) * count;
1873 	if (bifstp->ifbpstp_len == 0) {
1874 		bifstp->ifbpstp_len = buflen;
1875 		return (0);
1876 	}
1877 
1878 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1879 	if (outbuf == NULL)
1880 		return (ENOMEM);
1881 
1882 	count = 0;
1883 	buf = outbuf;
1884 	len = min(bifstp->ifbpstp_len, buflen);
1885 	bzero(&bpreq, sizeof(bpreq));
1886 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1887 		if (len < sizeof(bpreq))
1888 			break;
1889 
1890 		if ((bif->bif_flags & IFBIF_STP) == 0)
1891 			continue;
1892 
1893 		bp = &bif->bif_stp;
1894 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1895 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1896 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1897 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1898 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1899 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1900 
1901 		memcpy(buf, &bpreq, sizeof(bpreq));
1902 		count++;
1903 		buf += sizeof(bpreq);
1904 		len -= sizeof(bpreq);
1905 	}
1906 
1907 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1908 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1909 	free(outbuf, M_TEMP);
1910 	return (error);
1911 }
1912 
1913 static int
1914 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1915 {
1916 	struct ifbrparam *param = arg;
1917 
1918 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1919 }
1920 
1921 static int
1922 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1923 {
1924 	struct ifbrparam *param = arg;
1925 
1926 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1927 }
1928 
1929 /*
1930  * bridge_ifdetach:
1931  *
1932  *	Detach an interface from a bridge.  Called when a member
1933  *	interface is detaching.
1934  */
1935 static void
1936 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1937 {
1938 	struct bridge_softc *sc = ifp->if_bridge;
1939 	struct bridge_iflist *bif;
1940 
1941 	if (ifp->if_flags & IFF_RENAMING)
1942 		return;
1943 	if (V_bridge_cloner == NULL) {
1944 		/*
1945 		 * This detach handler can be called after
1946 		 * vnet_bridge_uninit().  Just return in that case.
1947 		 */
1948 		return;
1949 	}
1950 	/* Check if the interface is a bridge member */
1951 	if (sc != NULL) {
1952 		BRIDGE_LOCK(sc);
1953 
1954 		bif = bridge_lookup_member_if(sc, ifp);
1955 		if (bif != NULL)
1956 			bridge_delete_member(sc, bif, 1);
1957 
1958 		BRIDGE_UNLOCK(sc);
1959 		return;
1960 	}
1961 
1962 	/* Check if the interface is a span port */
1963 	BRIDGE_LIST_LOCK();
1964 	LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1965 		BRIDGE_LOCK(sc);
1966 		CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1967 			if (ifp == bif->bif_ifp) {
1968 				bridge_delete_span(sc, bif);
1969 				break;
1970 			}
1971 
1972 		BRIDGE_UNLOCK(sc);
1973 	}
1974 	BRIDGE_LIST_UNLOCK();
1975 }
1976 
1977 /*
1978  * bridge_init:
1979  *
1980  *	Initialize a bridge interface.
1981  */
1982 static void
1983 bridge_init(void *xsc)
1984 {
1985 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1986 	struct ifnet *ifp = sc->sc_ifp;
1987 
1988 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1989 		return;
1990 
1991 	BRIDGE_LOCK(sc);
1992 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1993 	    bridge_timer, sc);
1994 
1995 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1996 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1997 
1998 	BRIDGE_UNLOCK(sc);
1999 }
2000 
2001 /*
2002  * bridge_stop:
2003  *
2004  *	Stop the bridge interface.
2005  */
2006 static void
2007 bridge_stop(struct ifnet *ifp, int disable)
2008 {
2009 	struct bridge_softc *sc = ifp->if_softc;
2010 
2011 	BRIDGE_LOCK_ASSERT(sc);
2012 
2013 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2014 		return;
2015 
2016 	BRIDGE_RT_LOCK(sc);
2017 	callout_stop(&sc->sc_brcallout);
2018 
2019 	bstp_stop(&sc->sc_stp);
2020 
2021 	bridge_rtflush(sc, IFBF_FLUSHDYN);
2022 	BRIDGE_RT_UNLOCK(sc);
2023 
2024 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
2025 }
2026 
2027 /*
2028  * bridge_enqueue:
2029  *
2030  *	Enqueue a packet on a bridge member interface.
2031  *
2032  */
2033 static int
2034 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
2035 {
2036 	int len, err = 0;
2037 	short mflags;
2038 	struct mbuf *m0;
2039 
2040 	/* We may be sending a fragment so traverse the mbuf */
2041 	for (; m; m = m0) {
2042 		m0 = m->m_nextpkt;
2043 		m->m_nextpkt = NULL;
2044 		len = m->m_pkthdr.len;
2045 		mflags = m->m_flags;
2046 
2047 		/*
2048 		 * If underlying interface can not do VLAN tag insertion itself
2049 		 * then attach a packet tag that holds it.
2050 		 */
2051 		if ((m->m_flags & M_VLANTAG) &&
2052 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
2053 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
2054 			if (m == NULL) {
2055 				if_printf(dst_ifp,
2056 				    "unable to prepend VLAN header\n");
2057 				if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2058 				continue;
2059 			}
2060 			m->m_flags &= ~M_VLANTAG;
2061 		}
2062 
2063 		M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2064 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2065 			m_freem(m0);
2066 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2067 			break;
2068 		}
2069 
2070 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2071 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2072 		if (mflags & M_MCAST)
2073 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2074 	}
2075 
2076 	return (err);
2077 }
2078 
2079 /*
2080  * bridge_dummynet:
2081  *
2082  * 	Receive a queued packet from dummynet and pass it on to the output
2083  * 	interface.
2084  *
2085  *	The mbuf has the Ethernet header already attached.
2086  */
2087 static void
2088 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2089 {
2090 	struct bridge_softc *sc;
2091 
2092 	sc = ifp->if_bridge;
2093 
2094 	/*
2095 	 * The packet didnt originate from a member interface. This should only
2096 	 * ever happen if a member interface is removed while packets are
2097 	 * queued for it.
2098 	 */
2099 	if (sc == NULL) {
2100 		m_freem(m);
2101 		return;
2102 	}
2103 
2104 	if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2105 #ifdef INET6
2106 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2107 #endif
2108 	    ) {
2109 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2110 			return;
2111 		if (m == NULL)
2112 			return;
2113 	}
2114 
2115 	bridge_enqueue(sc, ifp, m);
2116 }
2117 
2118 /*
2119  * bridge_output:
2120  *
2121  *	Send output from a bridge member interface.  This
2122  *	performs the bridging function for locally originated
2123  *	packets.
2124  *
2125  *	The mbuf has the Ethernet header already attached.  We must
2126  *	enqueue or free the mbuf before returning.
2127  */
2128 static int
2129 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2130     struct rtentry *rt)
2131 {
2132 	struct ether_header *eh;
2133 	struct ifnet *bifp, *dst_if;
2134 	struct bridge_softc *sc;
2135 	uint16_t vlan;
2136 
2137 	NET_EPOCH_ASSERT();
2138 
2139 	if (m->m_len < ETHER_HDR_LEN) {
2140 		m = m_pullup(m, ETHER_HDR_LEN);
2141 		if (m == NULL)
2142 			return (0);
2143 	}
2144 
2145 	eh = mtod(m, struct ether_header *);
2146 	sc = ifp->if_bridge;
2147 	vlan = VLANTAGOF(m);
2148 
2149 	bifp = sc->sc_ifp;
2150 
2151 	/*
2152 	 * If bridge is down, but the original output interface is up,
2153 	 * go ahead and send out that interface.  Otherwise, the packet
2154 	 * is dropped below.
2155 	 */
2156 	if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2157 		dst_if = ifp;
2158 		goto sendunicast;
2159 	}
2160 
2161 	/*
2162 	 * If the packet is a multicast, or we don't know a better way to
2163 	 * get there, send to all interfaces.
2164 	 */
2165 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
2166 		dst_if = NULL;
2167 	else
2168 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2169 	/* Tap any traffic not passing back out the originating interface */
2170 	if (dst_if != ifp)
2171 		ETHER_BPF_MTAP(bifp, m);
2172 	if (dst_if == NULL) {
2173 		struct bridge_iflist *bif;
2174 		struct mbuf *mc;
2175 		int used = 0;
2176 
2177 		bridge_span(sc, m);
2178 
2179 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2180 			dst_if = bif->bif_ifp;
2181 
2182 			if (dst_if->if_type == IFT_GIF)
2183 				continue;
2184 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2185 				continue;
2186 
2187 			/*
2188 			 * If this is not the original output interface,
2189 			 * and the interface is participating in spanning
2190 			 * tree, make sure the port is in a state that
2191 			 * allows forwarding.
2192 			 */
2193 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2194 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2195 				continue;
2196 
2197 			if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2198 				used = 1;
2199 				mc = m;
2200 			} else {
2201 				mc = m_dup(m, M_NOWAIT);
2202 				if (mc == NULL) {
2203 					if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2204 					continue;
2205 				}
2206 			}
2207 
2208 			bridge_enqueue(sc, dst_if, mc);
2209 		}
2210 		if (used == 0)
2211 			m_freem(m);
2212 		return (0);
2213 	}
2214 
2215 sendunicast:
2216 	/*
2217 	 * XXX Spanning tree consideration here?
2218 	 */
2219 
2220 	bridge_span(sc, m);
2221 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2222 		m_freem(m);
2223 		return (0);
2224 	}
2225 
2226 	bridge_enqueue(sc, dst_if, m);
2227 	return (0);
2228 }
2229 
2230 /*
2231  * bridge_transmit:
2232  *
2233  *	Do output on a bridge.
2234  *
2235  */
2236 static int
2237 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2238 {
2239 	struct bridge_softc *sc;
2240 	struct ether_header *eh;
2241 	struct ifnet *dst_if;
2242 	int error = 0;
2243 
2244 	sc = ifp->if_softc;
2245 
2246 	ETHER_BPF_MTAP(ifp, m);
2247 
2248 	eh = mtod(m, struct ether_header *);
2249 
2250 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2251 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2252 		error = bridge_enqueue(sc, dst_if, m);
2253 	} else
2254 		bridge_broadcast(sc, ifp, m, 0);
2255 
2256 	return (error);
2257 }
2258 
2259 #ifdef ALTQ
2260 static void
2261 bridge_altq_start(if_t ifp)
2262 {
2263 	struct ifaltq *ifq = &ifp->if_snd;
2264 	struct mbuf *m;
2265 
2266 	IFQ_LOCK(ifq);
2267 	IFQ_DEQUEUE_NOLOCK(ifq, m);
2268 	while (m != NULL) {
2269 		bridge_transmit(ifp, m);
2270 		IFQ_DEQUEUE_NOLOCK(ifq, m);
2271 	}
2272 	IFQ_UNLOCK(ifq);
2273 }
2274 
2275 static int
2276 bridge_altq_transmit(if_t ifp, struct mbuf *m)
2277 {
2278 	int err;
2279 
2280 	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
2281 		IFQ_ENQUEUE(&ifp->if_snd, m, err);
2282 		if (err == 0)
2283 			bridge_altq_start(ifp);
2284 	} else
2285 		err = bridge_transmit(ifp, m);
2286 
2287 	return (err);
2288 }
2289 #endif	/* ALTQ */
2290 
2291 /*
2292  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2293  */
2294 static void
2295 bridge_qflush(struct ifnet *ifp __unused)
2296 {
2297 }
2298 
2299 /*
2300  * bridge_forward:
2301  *
2302  *	The forwarding function of the bridge.
2303  *
2304  *	NOTE: Releases the lock on return.
2305  */
2306 static void
2307 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2308     struct mbuf *m)
2309 {
2310 	struct bridge_iflist *dbif;
2311 	struct ifnet *src_if, *dst_if, *ifp;
2312 	struct ether_header *eh;
2313 	uint16_t vlan;
2314 	uint8_t *dst;
2315 	int error;
2316 
2317 	NET_EPOCH_ASSERT();
2318 
2319 	src_if = m->m_pkthdr.rcvif;
2320 	ifp = sc->sc_ifp;
2321 
2322 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2323 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2324 	vlan = VLANTAGOF(m);
2325 
2326 	if ((sbif->bif_flags & IFBIF_STP) &&
2327 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2328 		goto drop;
2329 
2330 	eh = mtod(m, struct ether_header *);
2331 	dst = eh->ether_dhost;
2332 
2333 	/* If the interface is learning, record the address. */
2334 	if (sbif->bif_flags & IFBIF_LEARNING) {
2335 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2336 		    sbif, 0, IFBAF_DYNAMIC);
2337 		/*
2338 		 * If the interface has addresses limits then deny any source
2339 		 * that is not in the cache.
2340 		 */
2341 		if (error && sbif->bif_addrmax)
2342 			goto drop;
2343 	}
2344 
2345 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2346 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2347 		goto drop;
2348 
2349 	/*
2350 	 * At this point, the port either doesn't participate
2351 	 * in spanning tree or it is in the forwarding state.
2352 	 */
2353 
2354 	/*
2355 	 * If the packet is unicast, destined for someone on
2356 	 * "this" side of the bridge, drop it.
2357 	 */
2358 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2359 		dst_if = bridge_rtlookup(sc, dst, vlan);
2360 		if (src_if == dst_if)
2361 			goto drop;
2362 	} else {
2363 		/*
2364 		 * Check if its a reserved multicast address, any address
2365 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2366 		 * bridge.
2367 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2368 		 */
2369 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2370 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2371 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2372 			goto drop;
2373 
2374 		/* ...forward it to all interfaces. */
2375 		if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2376 		dst_if = NULL;
2377 	}
2378 
2379 	/*
2380 	 * If we have a destination interface which is a member of our bridge,
2381 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2382 	 * For broadcast or multicast packets, don't bother because it will
2383 	 * be reinjected into ether_input. We do this before we pass the packets
2384 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2385 	 * drop the packet, or possibly modify it, making it difficult to debug
2386 	 * firewall issues on the bridge.
2387 	 */
2388 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2389 		ETHER_BPF_MTAP(ifp, m);
2390 
2391 	/* run the packet filter */
2392 	if (PFIL_HOOKED_IN(V_inet_pfil_head)
2393 #ifdef INET6
2394 	    || PFIL_HOOKED_IN(V_inet6_pfil_head)
2395 #endif
2396 	    ) {
2397 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2398 			return;
2399 		if (m == NULL)
2400 			return;
2401 	}
2402 
2403 	if (dst_if == NULL) {
2404 		bridge_broadcast(sc, src_if, m, 1);
2405 		return;
2406 	}
2407 
2408 	/*
2409 	 * At this point, we're dealing with a unicast frame
2410 	 * going to a different interface.
2411 	 */
2412 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2413 		goto drop;
2414 
2415 	dbif = bridge_lookup_member_if(sc, dst_if);
2416 	if (dbif == NULL)
2417 		/* Not a member of the bridge (anymore?) */
2418 		goto drop;
2419 
2420 	/* Private segments can not talk to each other */
2421 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2422 		goto drop;
2423 
2424 	if ((dbif->bif_flags & IFBIF_STP) &&
2425 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2426 		goto drop;
2427 
2428 	if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2429 #ifdef INET6
2430 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2431 #endif
2432 	    ) {
2433 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2434 			return;
2435 		if (m == NULL)
2436 			return;
2437 	}
2438 
2439 	bridge_enqueue(sc, dst_if, m);
2440 	return;
2441 
2442 drop:
2443 	m_freem(m);
2444 }
2445 
2446 /*
2447  * bridge_input:
2448  *
2449  *	Receive input from a member interface.  Queue the packet for
2450  *	bridging if it is not for us.
2451  */
2452 static struct mbuf *
2453 bridge_input(struct ifnet *ifp, struct mbuf *m)
2454 {
2455 	struct bridge_softc *sc = ifp->if_bridge;
2456 	struct bridge_iflist *bif, *bif2;
2457 	struct ifnet *bifp;
2458 	struct ether_header *eh;
2459 	struct mbuf *mc, *mc2;
2460 	uint16_t vlan;
2461 	int error;
2462 
2463 	NET_EPOCH_ASSERT();
2464 
2465 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2466 		return (m);
2467 
2468 	bifp = sc->sc_ifp;
2469 	vlan = VLANTAGOF(m);
2470 
2471 	/*
2472 	 * Implement support for bridge monitoring. If this flag has been
2473 	 * set on this interface, discard the packet once we push it through
2474 	 * the bpf(4) machinery, but before we do, increment the byte and
2475 	 * packet counters associated with this interface.
2476 	 */
2477 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2478 		m->m_pkthdr.rcvif  = bifp;
2479 		ETHER_BPF_MTAP(bifp, m);
2480 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2481 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2482 		m_freem(m);
2483 		return (NULL);
2484 	}
2485 	bif = bridge_lookup_member_if(sc, ifp);
2486 	if (bif == NULL) {
2487 		return (m);
2488 	}
2489 
2490 	eh = mtod(m, struct ether_header *);
2491 
2492 	bridge_span(sc, m);
2493 
2494 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2495 		/* Tap off 802.1D packets; they do not get forwarded. */
2496 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2497 		    ETHER_ADDR_LEN) == 0) {
2498 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2499 			return (NULL);
2500 		}
2501 
2502 		if ((bif->bif_flags & IFBIF_STP) &&
2503 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2504 			return (m);
2505 		}
2506 
2507 		/*
2508 		 * Make a deep copy of the packet and enqueue the copy
2509 		 * for bridge processing; return the original packet for
2510 		 * local processing.
2511 		 */
2512 		mc = m_dup(m, M_NOWAIT);
2513 		if (mc == NULL) {
2514 			return (m);
2515 		}
2516 
2517 		/* Perform the bridge forwarding function with the copy. */
2518 		bridge_forward(sc, bif, mc);
2519 
2520 		/*
2521 		 * Reinject the mbuf as arriving on the bridge so we have a
2522 		 * chance at claiming multicast packets. We can not loop back
2523 		 * here from ether_input as a bridge is never a member of a
2524 		 * bridge.
2525 		 */
2526 		KASSERT(bifp->if_bridge == NULL,
2527 		    ("loop created in bridge_input"));
2528 		mc2 = m_dup(m, M_NOWAIT);
2529 		if (mc2 != NULL) {
2530 			/* Keep the layer3 header aligned */
2531 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2532 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2533 		}
2534 		if (mc2 != NULL) {
2535 			mc2->m_pkthdr.rcvif = bifp;
2536 			(*bifp->if_input)(bifp, mc2);
2537 		}
2538 
2539 		/* Return the original packet for local processing. */
2540 		return (m);
2541 	}
2542 
2543 	if ((bif->bif_flags & IFBIF_STP) &&
2544 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2545 		return (m);
2546 	}
2547 
2548 #if (defined(INET) || defined(INET6))
2549 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2550 	|| ((iface)->if_carp \
2551 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2552 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2553 	|| ((iface)->if_carp \
2554 	    && (*carp_forus_p)((iface), eh->ether_shost))
2555 #else
2556 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2557 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2558 #endif
2559 
2560 #ifdef INET6
2561 #   define OR_PFIL_HOOKED_INET6 \
2562 	|| PFIL_HOOKED_IN(V_inet6_pfil_head)
2563 #else
2564 #   define OR_PFIL_HOOKED_INET6
2565 #endif
2566 
2567 #define GRAB_OUR_PACKETS(iface) \
2568 	if ((iface)->if_type == IFT_GIF) \
2569 		continue; \
2570 	/* It is destined for us. */ \
2571 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2572 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2573 	    ) {								\
2574 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2575 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2576 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2577 			if (error && bif->bif_addrmax) {		\
2578 				m_freem(m);				\
2579 				return (NULL);				\
2580 			}						\
2581 		}							\
2582 		m->m_pkthdr.rcvif = iface;				\
2583 		if ((iface) == ifp) {					\
2584 			/* Skip bridge processing... src == dest */	\
2585 			return (m);					\
2586 		}							\
2587 		/* It's passing over or to the bridge, locally. */	\
2588 		ETHER_BPF_MTAP(bifp, m);				\
2589 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);		\
2590 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2591 		/* Filter on the physical interface. */			\
2592 		if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \
2593 		     OR_PFIL_HOOKED_INET6)) {				\
2594 			if (bridge_pfil(&m, NULL, ifp,			\
2595 			    PFIL_IN) != 0 || m == NULL) {		\
2596 				return (NULL);				\
2597 			}						\
2598 		}							\
2599 		if ((iface) != bifp)					\
2600 			ETHER_BPF_MTAP(iface, m);			\
2601 		return (m);						\
2602 	}								\
2603 									\
2604 	/* We just received a packet that we sent out. */		\
2605 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2606 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2607 	    ) {								\
2608 		m_freem(m);						\
2609 		return (NULL);						\
2610 	}
2611 
2612 	/*
2613 	 * Unicast.  Make sure it's not for the bridge.
2614 	 */
2615 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2616 
2617 	/*
2618 	 * Give a chance for ifp at first priority. This will help when	the
2619 	 * packet comes through the interface like VLAN's with the same MACs
2620 	 * on several interfaces from the same bridge. This also will save
2621 	 * some CPU cycles in case the destination interface and the input
2622 	 * interface (eq ifp) are the same.
2623 	 */
2624 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2625 
2626 	/* Now check the all bridge members. */
2627 	CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2628 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2629 	}
2630 
2631 #undef OR_CARP_CHECK_WE_ARE_DST
2632 #undef OR_CARP_CHECK_WE_ARE_SRC
2633 #undef OR_PFIL_HOOKED_INET6
2634 #undef GRAB_OUR_PACKETS
2635 
2636 	/* Perform the bridge forwarding function. */
2637 	bridge_forward(sc, bif, m);
2638 
2639 	return (NULL);
2640 }
2641 
2642 /*
2643  * bridge_broadcast:
2644  *
2645  *	Send a frame to all interfaces that are members of
2646  *	the bridge, except for the one on which the packet
2647  *	arrived.
2648  *
2649  *	NOTE: Releases the lock on return.
2650  */
2651 static void
2652 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2653     struct mbuf *m, int runfilt)
2654 {
2655 	struct bridge_iflist *dbif, *sbif;
2656 	struct mbuf *mc;
2657 	struct ifnet *dst_if;
2658 	int used = 0, i;
2659 
2660 	NET_EPOCH_ASSERT();
2661 
2662 	sbif = bridge_lookup_member_if(sc, src_if);
2663 
2664 	/* Filter on the bridge interface before broadcasting */
2665 	if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2666 #ifdef INET6
2667 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2668 #endif
2669 	    )) {
2670 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2671 			return;
2672 		if (m == NULL)
2673 			return;
2674 	}
2675 
2676 	CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2677 		dst_if = dbif->bif_ifp;
2678 		if (dst_if == src_if)
2679 			continue;
2680 
2681 		/* Private segments can not talk to each other */
2682 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2683 			continue;
2684 
2685 		if ((dbif->bif_flags & IFBIF_STP) &&
2686 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2687 			continue;
2688 
2689 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2690 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2691 			continue;
2692 
2693 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2694 			continue;
2695 
2696 		if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2697 			mc = m;
2698 			used = 1;
2699 		} else {
2700 			mc = m_dup(m, M_NOWAIT);
2701 			if (mc == NULL) {
2702 				if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2703 				continue;
2704 			}
2705 		}
2706 
2707 		/*
2708 		 * Filter on the output interface. Pass a NULL bridge interface
2709 		 * pointer so we do not redundantly filter on the bridge for
2710 		 * each interface we broadcast on.
2711 		 */
2712 		if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2713 #ifdef INET6
2714 		    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2715 #endif
2716 		    )) {
2717 			if (used == 0) {
2718 				/* Keep the layer3 header aligned */
2719 				i = min(mc->m_pkthdr.len, max_protohdr);
2720 				mc = m_copyup(mc, i, ETHER_ALIGN);
2721 				if (mc == NULL) {
2722 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2723 					continue;
2724 				}
2725 			}
2726 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2727 				continue;
2728 			if (mc == NULL)
2729 				continue;
2730 		}
2731 
2732 		bridge_enqueue(sc, dst_if, mc);
2733 	}
2734 	if (used == 0)
2735 		m_freem(m);
2736 }
2737 
2738 /*
2739  * bridge_span:
2740  *
2741  *	Duplicate a packet out one or more interfaces that are in span mode,
2742  *	the original mbuf is unmodified.
2743  */
2744 static void
2745 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2746 {
2747 	struct bridge_iflist *bif;
2748 	struct ifnet *dst_if;
2749 	struct mbuf *mc;
2750 
2751 	NET_EPOCH_ASSERT();
2752 
2753 	if (CK_LIST_EMPTY(&sc->sc_spanlist))
2754 		return;
2755 
2756 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2757 		dst_if = bif->bif_ifp;
2758 
2759 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2760 			continue;
2761 
2762 		mc = m_dup(m, M_NOWAIT);
2763 		if (mc == NULL) {
2764 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2765 			continue;
2766 		}
2767 
2768 		bridge_enqueue(sc, dst_if, mc);
2769 	}
2770 }
2771 
2772 /*
2773  * bridge_rtupdate:
2774  *
2775  *	Add a bridge routing entry.
2776  */
2777 static int
2778 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2779     struct bridge_iflist *bif, int setflags, uint8_t flags)
2780 {
2781 	struct bridge_rtnode *brt;
2782 	int error;
2783 
2784 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
2785 
2786 	/* Check the source address is valid and not multicast. */
2787 	if (ETHER_IS_MULTICAST(dst) ||
2788 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2789 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2790 		return (EINVAL);
2791 
2792 	/* 802.1p frames map to vlan 1 */
2793 	if (vlan == 0)
2794 		vlan = 1;
2795 
2796 	/*
2797 	 * A route for this destination might already exist.  If so,
2798 	 * update it, otherwise create a new one.
2799 	 */
2800 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2801 		BRIDGE_RT_LOCK(sc);
2802 
2803 		/* Check again, now that we have the lock. There could have
2804 		 * been a race and we only want to insert this once. */
2805 		if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) {
2806 			BRIDGE_RT_UNLOCK(sc);
2807 			return (0);
2808 		}
2809 
2810 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2811 			sc->sc_brtexceeded++;
2812 			BRIDGE_RT_UNLOCK(sc);
2813 			return (ENOSPC);
2814 		}
2815 		/* Check per interface address limits (if enabled) */
2816 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2817 			bif->bif_addrexceeded++;
2818 			BRIDGE_RT_UNLOCK(sc);
2819 			return (ENOSPC);
2820 		}
2821 
2822 		/*
2823 		 * Allocate a new bridge forwarding node, and
2824 		 * initialize the expiration time and Ethernet
2825 		 * address.
2826 		 */
2827 		brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2828 		if (brt == NULL) {
2829 			BRIDGE_RT_UNLOCK(sc);
2830 			return (ENOMEM);
2831 		}
2832 		brt->brt_vnet = curvnet;
2833 
2834 		if (bif->bif_flags & IFBIF_STICKY)
2835 			brt->brt_flags = IFBAF_STICKY;
2836 		else
2837 			brt->brt_flags = IFBAF_DYNAMIC;
2838 
2839 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2840 		brt->brt_vlan = vlan;
2841 
2842 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2843 			uma_zfree(V_bridge_rtnode_zone, brt);
2844 			BRIDGE_RT_UNLOCK(sc);
2845 			return (error);
2846 		}
2847 		brt->brt_dst = bif;
2848 		bif->bif_addrcnt++;
2849 
2850 		BRIDGE_RT_UNLOCK(sc);
2851 	}
2852 
2853 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2854 	    brt->brt_dst != bif) {
2855 		BRIDGE_RT_LOCK(sc);
2856 		brt->brt_dst->bif_addrcnt--;
2857 		brt->brt_dst = bif;
2858 		brt->brt_dst->bif_addrcnt++;
2859 		BRIDGE_RT_UNLOCK(sc);
2860 	}
2861 
2862 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2863 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2864 	if (setflags)
2865 		brt->brt_flags = flags;
2866 
2867 	return (0);
2868 }
2869 
2870 /*
2871  * bridge_rtlookup:
2872  *
2873  *	Lookup the destination interface for an address.
2874  */
2875 static struct ifnet *
2876 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2877 {
2878 	struct bridge_rtnode *brt;
2879 
2880 	NET_EPOCH_ASSERT();
2881 
2882 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2883 		return (NULL);
2884 
2885 	return (brt->brt_ifp);
2886 }
2887 
2888 /*
2889  * bridge_rttrim:
2890  *
2891  *	Trim the routine table so that we have a number
2892  *	of routing entries less than or equal to the
2893  *	maximum number.
2894  */
2895 static void
2896 bridge_rttrim(struct bridge_softc *sc)
2897 {
2898 	struct bridge_rtnode *brt, *nbrt;
2899 
2900 	NET_EPOCH_ASSERT();
2901 	BRIDGE_RT_LOCK_ASSERT(sc);
2902 
2903 	/* Make sure we actually need to do this. */
2904 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2905 		return;
2906 
2907 	/* Force an aging cycle; this might trim enough addresses. */
2908 	bridge_rtage(sc);
2909 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2910 		return;
2911 
2912 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2913 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2914 			bridge_rtnode_destroy(sc, brt);
2915 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2916 				return;
2917 		}
2918 	}
2919 }
2920 
2921 /*
2922  * bridge_timer:
2923  *
2924  *	Aging timer for the bridge.
2925  */
2926 static void
2927 bridge_timer(void *arg)
2928 {
2929 	struct bridge_softc *sc = arg;
2930 
2931 	BRIDGE_RT_LOCK_ASSERT(sc);
2932 
2933 	/* Destruction of rtnodes requires a proper vnet context */
2934 	CURVNET_SET(sc->sc_ifp->if_vnet);
2935 	bridge_rtage(sc);
2936 
2937 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2938 		callout_reset(&sc->sc_brcallout,
2939 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2940 	CURVNET_RESTORE();
2941 }
2942 
2943 /*
2944  * bridge_rtage:
2945  *
2946  *	Perform an aging cycle.
2947  */
2948 static void
2949 bridge_rtage(struct bridge_softc *sc)
2950 {
2951 	struct bridge_rtnode *brt, *nbrt;
2952 
2953 	BRIDGE_RT_LOCK_ASSERT(sc);
2954 
2955 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2956 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2957 			if (time_uptime >= brt->brt_expire)
2958 				bridge_rtnode_destroy(sc, brt);
2959 		}
2960 	}
2961 }
2962 
2963 /*
2964  * bridge_rtflush:
2965  *
2966  *	Remove all dynamic addresses from the bridge.
2967  */
2968 static void
2969 bridge_rtflush(struct bridge_softc *sc, int full)
2970 {
2971 	struct bridge_rtnode *brt, *nbrt;
2972 
2973 	BRIDGE_RT_LOCK_ASSERT(sc);
2974 
2975 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2976 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2977 			bridge_rtnode_destroy(sc, brt);
2978 	}
2979 }
2980 
2981 /*
2982  * bridge_rtdaddr:
2983  *
2984  *	Remove an address from the table.
2985  */
2986 static int
2987 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2988 {
2989 	struct bridge_rtnode *brt;
2990 	int found = 0;
2991 
2992 	BRIDGE_RT_LOCK(sc);
2993 
2994 	/*
2995 	 * If vlan is zero then we want to delete for all vlans so the lookup
2996 	 * may return more than one.
2997 	 */
2998 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2999 		bridge_rtnode_destroy(sc, brt);
3000 		found = 1;
3001 	}
3002 
3003 	BRIDGE_RT_UNLOCK(sc);
3004 
3005 	return (found ? 0 : ENOENT);
3006 }
3007 
3008 /*
3009  * bridge_rtdelete:
3010  *
3011  *	Delete routes to a speicifc member interface.
3012  */
3013 static void
3014 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
3015 {
3016 	struct bridge_rtnode *brt, *nbrt;
3017 
3018 	BRIDGE_RT_LOCK_ASSERT(sc);
3019 
3020 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3021 		if (brt->brt_ifp == ifp && (full ||
3022 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
3023 			bridge_rtnode_destroy(sc, brt);
3024 	}
3025 }
3026 
3027 /*
3028  * bridge_rtable_init:
3029  *
3030  *	Initialize the route table for this bridge.
3031  */
3032 static void
3033 bridge_rtable_init(struct bridge_softc *sc)
3034 {
3035 	int i;
3036 
3037 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
3038 	    M_DEVBUF, M_WAITOK);
3039 
3040 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
3041 		CK_LIST_INIT(&sc->sc_rthash[i]);
3042 
3043 	sc->sc_rthash_key = arc4random();
3044 	CK_LIST_INIT(&sc->sc_rtlist);
3045 }
3046 
3047 /*
3048  * bridge_rtable_fini:
3049  *
3050  *	Deconstruct the route table for this bridge.
3051  */
3052 static void
3053 bridge_rtable_fini(struct bridge_softc *sc)
3054 {
3055 
3056 	KASSERT(sc->sc_brtcnt == 0,
3057 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3058 	free(sc->sc_rthash, M_DEVBUF);
3059 }
3060 
3061 /*
3062  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3063  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3064  */
3065 #define	mix(a, b, c)							\
3066 do {									\
3067 	a -= b; a -= c; a ^= (c >> 13);					\
3068 	b -= c; b -= a; b ^= (a << 8);					\
3069 	c -= a; c -= b; c ^= (b >> 13);					\
3070 	a -= b; a -= c; a ^= (c >> 12);					\
3071 	b -= c; b -= a; b ^= (a << 16);					\
3072 	c -= a; c -= b; c ^= (b >> 5);					\
3073 	a -= b; a -= c; a ^= (c >> 3);					\
3074 	b -= c; b -= a; b ^= (a << 10);					\
3075 	c -= a; c -= b; c ^= (b >> 15);					\
3076 } while (/*CONSTCOND*/0)
3077 
3078 static __inline uint32_t
3079 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3080 {
3081 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3082 
3083 	b += addr[5] << 8;
3084 	b += addr[4];
3085 	a += addr[3] << 24;
3086 	a += addr[2] << 16;
3087 	a += addr[1] << 8;
3088 	a += addr[0];
3089 
3090 	mix(a, b, c);
3091 
3092 	return (c & BRIDGE_RTHASH_MASK);
3093 }
3094 
3095 #undef mix
3096 
3097 static int
3098 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3099 {
3100 	int i, d;
3101 
3102 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3103 		d = ((int)a[i]) - ((int)b[i]);
3104 	}
3105 
3106 	return (d);
3107 }
3108 
3109 /*
3110  * bridge_rtnode_lookup:
3111  *
3112  *	Look up a bridge route node for the specified destination. Compare the
3113  *	vlan id or if zero then just return the first match.
3114  */
3115 static struct bridge_rtnode *
3116 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3117 {
3118 	struct bridge_rtnode *brt;
3119 	uint32_t hash;
3120 	int dir;
3121 
3122 	BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3123 
3124 	hash = bridge_rthash(sc, addr);
3125 	CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3126 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3127 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3128 			return (brt);
3129 		if (dir > 0)
3130 			return (NULL);
3131 	}
3132 
3133 	return (NULL);
3134 }
3135 
3136 /*
3137  * bridge_rtnode_insert:
3138  *
3139  *	Insert the specified bridge node into the route table.  We
3140  *	assume the entry is not already in the table.
3141  */
3142 static int
3143 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3144 {
3145 	struct bridge_rtnode *lbrt;
3146 	uint32_t hash;
3147 	int dir;
3148 
3149 	BRIDGE_RT_LOCK_ASSERT(sc);
3150 
3151 	hash = bridge_rthash(sc, brt->brt_addr);
3152 
3153 	lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3154 	if (lbrt == NULL) {
3155 		CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3156 		goto out;
3157 	}
3158 
3159 	do {
3160 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3161 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3162 			return (EEXIST);
3163 		if (dir > 0) {
3164 			CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3165 			goto out;
3166 		}
3167 		if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3168 			CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3169 			goto out;
3170 		}
3171 		lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3172 	} while (lbrt != NULL);
3173 
3174 #ifdef DIAGNOSTIC
3175 	panic("bridge_rtnode_insert: impossible");
3176 #endif
3177 
3178 out:
3179 	CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3180 	sc->sc_brtcnt++;
3181 
3182 	return (0);
3183 }
3184 
3185 static void
3186 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3187 {
3188 	struct bridge_rtnode *brt;
3189 
3190 	brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3191 
3192 	CURVNET_SET(brt->brt_vnet);
3193 	uma_zfree(V_bridge_rtnode_zone, brt);
3194 	CURVNET_RESTORE();
3195 }
3196 
3197 /*
3198  * bridge_rtnode_destroy:
3199  *
3200  *	Destroy a bridge rtnode.
3201  */
3202 static void
3203 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3204 {
3205 	BRIDGE_RT_LOCK_ASSERT(sc);
3206 
3207 	CK_LIST_REMOVE(brt, brt_hash);
3208 
3209 	CK_LIST_REMOVE(brt, brt_list);
3210 	sc->sc_brtcnt--;
3211 	brt->brt_dst->bif_addrcnt--;
3212 
3213 	NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx);
3214 }
3215 
3216 /*
3217  * bridge_rtable_expire:
3218  *
3219  *	Set the expiry time for all routes on an interface.
3220  */
3221 static void
3222 bridge_rtable_expire(struct ifnet *ifp, int age)
3223 {
3224 	struct bridge_softc *sc = ifp->if_bridge;
3225 	struct bridge_rtnode *brt;
3226 
3227 	CURVNET_SET(ifp->if_vnet);
3228 	BRIDGE_RT_LOCK(sc);
3229 
3230 	/*
3231 	 * If the age is zero then flush, otherwise set all the expiry times to
3232 	 * age for the interface
3233 	 */
3234 	if (age == 0)
3235 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3236 	else {
3237 		CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3238 			/* Cap the expiry time to 'age' */
3239 			if (brt->brt_ifp == ifp &&
3240 			    brt->brt_expire > time_uptime + age &&
3241 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3242 				brt->brt_expire = time_uptime + age;
3243 		}
3244 	}
3245 	BRIDGE_RT_UNLOCK(sc);
3246 	CURVNET_RESTORE();
3247 }
3248 
3249 /*
3250  * bridge_state_change:
3251  *
3252  *	Callback from the bridgestp code when a port changes states.
3253  */
3254 static void
3255 bridge_state_change(struct ifnet *ifp, int state)
3256 {
3257 	struct bridge_softc *sc = ifp->if_bridge;
3258 	static const char *stpstates[] = {
3259 		"disabled",
3260 		"listening",
3261 		"learning",
3262 		"forwarding",
3263 		"blocking",
3264 		"discarding"
3265 	};
3266 
3267 	CURVNET_SET(ifp->if_vnet);
3268 	if (V_log_stp)
3269 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3270 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3271 	CURVNET_RESTORE();
3272 }
3273 
3274 /*
3275  * Send bridge packets through pfil if they are one of the types pfil can deal
3276  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3277  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3278  * that interface.
3279  */
3280 static int
3281 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3282 {
3283 	int snap, error, i, hlen;
3284 	struct ether_header *eh1, eh2;
3285 	struct ip *ip;
3286 	struct llc llc1;
3287 	u_int16_t ether_type;
3288 	pfil_return_t rv;
3289 
3290 	snap = 0;
3291 	error = -1;	/* Default error if not error == 0 */
3292 
3293 #if 0
3294 	/* we may return with the IP fields swapped, ensure its not shared */
3295 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3296 #endif
3297 
3298 	if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3299 		return (0); /* filtering is disabled */
3300 
3301 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3302 	if ((*mp)->m_len < i) {
3303 	    *mp = m_pullup(*mp, i);
3304 	    if (*mp == NULL) {
3305 		printf("%s: m_pullup failed\n", __func__);
3306 		return (-1);
3307 	    }
3308 	}
3309 
3310 	eh1 = mtod(*mp, struct ether_header *);
3311 	ether_type = ntohs(eh1->ether_type);
3312 
3313 	/*
3314 	 * Check for SNAP/LLC.
3315 	 */
3316 	if (ether_type < ETHERMTU) {
3317 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3318 
3319 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3320 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3321 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3322 		    llc2->llc_control == LLC_UI) {
3323 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3324 			snap = 1;
3325 		}
3326 	}
3327 
3328 	/*
3329 	 * If we're trying to filter bridge traffic, don't look at anything
3330 	 * other than IP and ARP traffic.  If the filter doesn't understand
3331 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3332 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3333 	 * but of course we don't have an AppleTalk filter to begin with.
3334 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3335 	 * ARP traffic.)
3336 	 */
3337 	switch (ether_type) {
3338 		case ETHERTYPE_ARP:
3339 		case ETHERTYPE_REVARP:
3340 			if (V_pfil_ipfw_arp == 0)
3341 				return (0); /* Automatically pass */
3342 			break;
3343 
3344 		case ETHERTYPE_IP:
3345 #ifdef INET6
3346 		case ETHERTYPE_IPV6:
3347 #endif /* INET6 */
3348 			break;
3349 		default:
3350 			/*
3351 			 * Check to see if the user wants to pass non-ip
3352 			 * packets, these will not be checked by pfil(9) and
3353 			 * passed unconditionally so the default is to drop.
3354 			 */
3355 			if (V_pfil_onlyip)
3356 				goto bad;
3357 	}
3358 
3359 	/* Run the packet through pfil before stripping link headers */
3360 	if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3361 	    dir == PFIL_OUT && ifp != NULL) {
3362 		switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) {
3363 		case PFIL_DROPPED:
3364 			return (EACCES);
3365 		case PFIL_CONSUMED:
3366 			return (0);
3367 		}
3368 	}
3369 
3370 	/* Strip off the Ethernet header and keep a copy. */
3371 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3372 	m_adj(*mp, ETHER_HDR_LEN);
3373 
3374 	/* Strip off snap header, if present */
3375 	if (snap) {
3376 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3377 		m_adj(*mp, sizeof(struct llc));
3378 	}
3379 
3380 	/*
3381 	 * Check the IP header for alignment and errors
3382 	 */
3383 	if (dir == PFIL_IN) {
3384 		switch (ether_type) {
3385 			case ETHERTYPE_IP:
3386 				error = bridge_ip_checkbasic(mp);
3387 				break;
3388 #ifdef INET6
3389 			case ETHERTYPE_IPV6:
3390 				error = bridge_ip6_checkbasic(mp);
3391 				break;
3392 #endif /* INET6 */
3393 			default:
3394 				error = 0;
3395 		}
3396 		if (error)
3397 			goto bad;
3398 	}
3399 
3400 	error = 0;
3401 
3402 	/*
3403 	 * Run the packet through pfil
3404 	 */
3405 	rv = PFIL_PASS;
3406 	switch (ether_type) {
3407 	case ETHERTYPE_IP:
3408 		/*
3409 		 * Run pfil on the member interface and the bridge, both can
3410 		 * be skipped by clearing pfil_member or pfil_bridge.
3411 		 *
3412 		 * Keep the order:
3413 		 *   in_if -> bridge_if -> out_if
3414 		 */
3415 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3416 		    pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3417 		    PFIL_PASS)
3418 			break;
3419 
3420 		if (V_pfil_member && ifp != NULL && (rv =
3421 		    pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) !=
3422 		    PFIL_PASS)
3423 			break;
3424 
3425 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3426 		    pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3427 		    PFIL_PASS)
3428 			break;
3429 
3430 		/* check if we need to fragment the packet */
3431 		/* bridge_fragment generates a mbuf chain of packets */
3432 		/* that already include eth headers */
3433 		if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3434 			i = (*mp)->m_pkthdr.len;
3435 			if (i > ifp->if_mtu) {
3436 				error = bridge_fragment(ifp, mp, &eh2, snap,
3437 					    &llc1);
3438 				return (error);
3439 			}
3440 		}
3441 
3442 		/* Recalculate the ip checksum. */
3443 		ip = mtod(*mp, struct ip *);
3444 		hlen = ip->ip_hl << 2;
3445 		if (hlen < sizeof(struct ip))
3446 			goto bad;
3447 		if (hlen > (*mp)->m_len) {
3448 			if ((*mp = m_pullup(*mp, hlen)) == NULL)
3449 				goto bad;
3450 			ip = mtod(*mp, struct ip *);
3451 			if (ip == NULL)
3452 				goto bad;
3453 		}
3454 		ip->ip_sum = 0;
3455 		if (hlen == sizeof(struct ip))
3456 			ip->ip_sum = in_cksum_hdr(ip);
3457 		else
3458 			ip->ip_sum = in_cksum(*mp, hlen);
3459 
3460 		break;
3461 #ifdef INET6
3462 	case ETHERTYPE_IPV6:
3463 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3464 		    pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3465 		    PFIL_PASS)
3466 			break;
3467 
3468 		if (V_pfil_member && ifp != NULL && (rv =
3469 		    pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) !=
3470 		    PFIL_PASS)
3471 			break;
3472 
3473 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3474 		    pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3475 		    PFIL_PASS)
3476 			break;
3477 		break;
3478 #endif
3479 	}
3480 
3481 	switch (rv) {
3482 	case PFIL_CONSUMED:
3483 		return (0);
3484 	case PFIL_DROPPED:
3485 		return (EACCES);
3486 	default:
3487 		break;
3488 	}
3489 
3490 	error = -1;
3491 
3492 	/*
3493 	 * Finally, put everything back the way it was and return
3494 	 */
3495 	if (snap) {
3496 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3497 		if (*mp == NULL)
3498 			return (error);
3499 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3500 	}
3501 
3502 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3503 	if (*mp == NULL)
3504 		return (error);
3505 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3506 
3507 	return (0);
3508 
3509 bad:
3510 	m_freem(*mp);
3511 	*mp = NULL;
3512 	return (error);
3513 }
3514 
3515 /*
3516  * Perform basic checks on header size since
3517  * pfil assumes ip_input has already processed
3518  * it for it.  Cut-and-pasted from ip_input.c.
3519  * Given how simple the IPv6 version is,
3520  * does the IPv4 version really need to be
3521  * this complicated?
3522  *
3523  * XXX Should we update ipstat here, or not?
3524  * XXX Right now we update ipstat but not
3525  * XXX csum_counter.
3526  */
3527 static int
3528 bridge_ip_checkbasic(struct mbuf **mp)
3529 {
3530 	struct mbuf *m = *mp;
3531 	struct ip *ip;
3532 	int len, hlen;
3533 	u_short sum;
3534 
3535 	if (*mp == NULL)
3536 		return (-1);
3537 
3538 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3539 		if ((m = m_copyup(m, sizeof(struct ip),
3540 			(max_linkhdr + 3) & ~3)) == NULL) {
3541 			/* XXXJRT new stat, please */
3542 			KMOD_IPSTAT_INC(ips_toosmall);
3543 			goto bad;
3544 		}
3545 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3546 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3547 			KMOD_IPSTAT_INC(ips_toosmall);
3548 			goto bad;
3549 		}
3550 	}
3551 	ip = mtod(m, struct ip *);
3552 	if (ip == NULL) goto bad;
3553 
3554 	if (ip->ip_v != IPVERSION) {
3555 		KMOD_IPSTAT_INC(ips_badvers);
3556 		goto bad;
3557 	}
3558 	hlen = ip->ip_hl << 2;
3559 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3560 		KMOD_IPSTAT_INC(ips_badhlen);
3561 		goto bad;
3562 	}
3563 	if (hlen > m->m_len) {
3564 		if ((m = m_pullup(m, hlen)) == NULL) {
3565 			KMOD_IPSTAT_INC(ips_badhlen);
3566 			goto bad;
3567 		}
3568 		ip = mtod(m, struct ip *);
3569 		if (ip == NULL) goto bad;
3570 	}
3571 
3572 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3573 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3574 	} else {
3575 		if (hlen == sizeof(struct ip)) {
3576 			sum = in_cksum_hdr(ip);
3577 		} else {
3578 			sum = in_cksum(m, hlen);
3579 		}
3580 	}
3581 	if (sum) {
3582 		KMOD_IPSTAT_INC(ips_badsum);
3583 		goto bad;
3584 	}
3585 
3586 	/* Retrieve the packet length. */
3587 	len = ntohs(ip->ip_len);
3588 
3589 	/*
3590 	 * Check for additional length bogosity
3591 	 */
3592 	if (len < hlen) {
3593 		KMOD_IPSTAT_INC(ips_badlen);
3594 		goto bad;
3595 	}
3596 
3597 	/*
3598 	 * Check that the amount of data in the buffers
3599 	 * is as at least much as the IP header would have us expect.
3600 	 * Drop packet if shorter than we expect.
3601 	 */
3602 	if (m->m_pkthdr.len < len) {
3603 		KMOD_IPSTAT_INC(ips_tooshort);
3604 		goto bad;
3605 	}
3606 
3607 	/* Checks out, proceed */
3608 	*mp = m;
3609 	return (0);
3610 
3611 bad:
3612 	*mp = m;
3613 	return (-1);
3614 }
3615 
3616 #ifdef INET6
3617 /*
3618  * Same as above, but for IPv6.
3619  * Cut-and-pasted from ip6_input.c.
3620  * XXX Should we update ip6stat, or not?
3621  */
3622 static int
3623 bridge_ip6_checkbasic(struct mbuf **mp)
3624 {
3625 	struct mbuf *m = *mp;
3626 	struct ip6_hdr *ip6;
3627 
3628 	/*
3629 	 * If the IPv6 header is not aligned, slurp it up into a new
3630 	 * mbuf with space for link headers, in the event we forward
3631 	 * it.  Otherwise, if it is aligned, make sure the entire base
3632 	 * IPv6 header is in the first mbuf of the chain.
3633 	 */
3634 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3635 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3636 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3637 			    (max_linkhdr + 3) & ~3)) == NULL) {
3638 			/* XXXJRT new stat, please */
3639 			IP6STAT_INC(ip6s_toosmall);
3640 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3641 			goto bad;
3642 		}
3643 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3644 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3645 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3646 			IP6STAT_INC(ip6s_toosmall);
3647 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3648 			goto bad;
3649 		}
3650 	}
3651 
3652 	ip6 = mtod(m, struct ip6_hdr *);
3653 
3654 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3655 		IP6STAT_INC(ip6s_badvers);
3656 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3657 		goto bad;
3658 	}
3659 
3660 	/* Checks out, proceed */
3661 	*mp = m;
3662 	return (0);
3663 
3664 bad:
3665 	*mp = m;
3666 	return (-1);
3667 }
3668 #endif /* INET6 */
3669 
3670 /*
3671  * bridge_fragment:
3672  *
3673  *	Fragment mbuf chain in multiple packets and prepend ethernet header.
3674  */
3675 static int
3676 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3677     int snap, struct llc *llc)
3678 {
3679 	struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3680 	struct ip *ip;
3681 	int error = -1;
3682 
3683 	if (m->m_len < sizeof(struct ip) &&
3684 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3685 		goto dropit;
3686 	ip = mtod(m, struct ip *);
3687 
3688 	m->m_pkthdr.csum_flags |= CSUM_IP;
3689 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3690 	if (error)
3691 		goto dropit;
3692 
3693 	/*
3694 	 * Walk the chain and re-add the Ethernet header for
3695 	 * each mbuf packet.
3696 	 */
3697 	for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3698 		nextpkt = mcur->m_nextpkt;
3699 		mcur->m_nextpkt = NULL;
3700 		if (snap) {
3701 			M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3702 			if (mcur == NULL) {
3703 				error = ENOBUFS;
3704 				if (mprev != NULL)
3705 					mprev->m_nextpkt = nextpkt;
3706 				goto dropit;
3707 			}
3708 			bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3709 		}
3710 
3711 		M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3712 		if (mcur == NULL) {
3713 			error = ENOBUFS;
3714 			if (mprev != NULL)
3715 				mprev->m_nextpkt = nextpkt;
3716 			goto dropit;
3717 		}
3718 		bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3719 
3720 		/*
3721 		 * The previous two M_PREPEND could have inserted one or two
3722 		 * mbufs in front so we have to update the previous packet's
3723 		 * m_nextpkt.
3724 		 */
3725 		mcur->m_nextpkt = nextpkt;
3726 		if (mprev != NULL)
3727 			mprev->m_nextpkt = mcur;
3728 		else {
3729 			/* The first mbuf in the original chain needs to be
3730 			 * updated. */
3731 			*mp = mcur;
3732 		}
3733 		mprev = mcur;
3734 	}
3735 
3736 	KMOD_IPSTAT_INC(ips_fragmented);
3737 	return (error);
3738 
3739 dropit:
3740 	for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3741 		m = mcur->m_nextpkt;
3742 		m_freem(mcur);
3743 	}
3744 	return (error);
3745 }
3746 
3747 static void
3748 bridge_linkstate(struct ifnet *ifp)
3749 {
3750 	struct bridge_softc *sc = ifp->if_bridge;
3751 	struct bridge_iflist *bif;
3752 	struct epoch_tracker et;
3753 
3754 	NET_EPOCH_ENTER(et);
3755 
3756 	bif = bridge_lookup_member_if(sc, ifp);
3757 	if (bif == NULL) {
3758 		NET_EPOCH_EXIT(et);
3759 		return;
3760 	}
3761 	bridge_linkcheck(sc);
3762 
3763 	bstp_linkstate(&bif->bif_stp);
3764 
3765 	NET_EPOCH_EXIT(et);
3766 }
3767 
3768 static void
3769 bridge_linkcheck(struct bridge_softc *sc)
3770 {
3771 	struct bridge_iflist *bif;
3772 	int new_link, hasls;
3773 
3774 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3775 
3776 	new_link = LINK_STATE_DOWN;
3777 	hasls = 0;
3778 	/* Our link is considered up if at least one of our ports is active */
3779 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3780 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3781 			hasls++;
3782 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3783 			new_link = LINK_STATE_UP;
3784 			break;
3785 		}
3786 	}
3787 	if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3788 		/* If no interfaces support link-state then we default to up */
3789 		new_link = LINK_STATE_UP;
3790 	}
3791 	if_link_state_change(sc->sc_ifp, new_link);
3792 }
3793