xref: /freebsd/sys/net/if_bridge.c (revision 2f02600abfddfc4e9f20dd384a2e729b451e16bd)
1 /*	$NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $	*/
2 
3 /*
4  * Copyright 2001 Wasabi Systems, Inc.
5  * All rights reserved.
6  *
7  * Written by Jason R. Thorpe for Wasabi Systems, Inc.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. All advertising materials mentioning features or use of this software
18  *    must display the following acknowledgement:
19  *	This product includes software developed for the NetBSD Project by
20  *	Wasabi Systems, Inc.
21  * 4. The name of Wasabi Systems, Inc. may not be used to endorse
22  *    or promote products derived from this software without specific prior
23  *    written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
27  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
28  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL WASABI SYSTEMS, INC
29  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35  * POSSIBILITY OF SUCH DAMAGE.
36  */
37 
38 /*
39  * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
40  * All rights reserved.
41  *
42  * Redistribution and use in source and binary forms, with or without
43  * modification, are permitted provided that the following conditions
44  * are met:
45  * 1. Redistributions of source code must retain the above copyright
46  *    notice, this list of conditions and the following disclaimer.
47  * 2. Redistributions in binary form must reproduce the above copyright
48  *    notice, this list of conditions and the following disclaimer in the
49  *    documentation and/or other materials provided with the distribution.
50  *
51  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
52  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
53  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
54  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
55  * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
56  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
57  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
58  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
59  * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
60  * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
61  * POSSIBILITY OF SUCH DAMAGE.
62  *
63  * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
64  */
65 
66 /*
67  * Network interface bridge support.
68  *
69  * TODO:
70  *
71  *	- Currently only supports Ethernet-like interfaces (Ethernet,
72  *	  802.11, VLANs on Ethernet, etc.)  Figure out a nice way
73  *	  to bridge other types of interfaces (FDDI-FDDI, and maybe
74  *	  consider heterogenous bridges).
75  */
76 
77 #include <sys/cdefs.h>
78 __FBSDID("$FreeBSD$");
79 
80 #include "opt_inet.h"
81 #include "opt_inet6.h"
82 
83 #include <sys/param.h>
84 #include <sys/eventhandler.h>
85 #include <sys/mbuf.h>
86 #include <sys/malloc.h>
87 #include <sys/protosw.h>
88 #include <sys/systm.h>
89 #include <sys/jail.h>
90 #include <sys/time.h>
91 #include <sys/socket.h> /* for net/if.h */
92 #include <sys/sockio.h>
93 #include <sys/ctype.h>  /* string functions */
94 #include <sys/kernel.h>
95 #include <sys/random.h>
96 #include <sys/syslog.h>
97 #include <sys/sysctl.h>
98 #include <vm/uma.h>
99 #include <sys/module.h>
100 #include <sys/priv.h>
101 #include <sys/proc.h>
102 #include <sys/lock.h>
103 #include <sys/mutex.h>
104 
105 #include <net/bpf.h>
106 #include <net/if.h>
107 #include <net/if_clone.h>
108 #include <net/if_dl.h>
109 #include <net/if_types.h>
110 #include <net/if_var.h>
111 #include <net/pfil.h>
112 #include <net/vnet.h>
113 
114 #include <netinet/in.h> /* for struct arpcom */
115 #include <netinet/in_systm.h>
116 #include <netinet/in_var.h>
117 #include <netinet/ip.h>
118 #include <netinet/ip_var.h>
119 #ifdef INET6
120 #include <netinet/ip6.h>
121 #include <netinet6/ip6_var.h>
122 #include <netinet6/in6_ifattach.h>
123 #endif
124 #if defined(INET) || defined(INET6)
125 #include <netinet/ip_carp.h>
126 #endif
127 #include <machine/in_cksum.h>
128 #include <netinet/if_ether.h> /* for struct arpcom */
129 #include <net/bridgestp.h>
130 #include <net/if_bridgevar.h>
131 #include <net/if_llc.h>
132 #include <net/if_vlan_var.h>
133 
134 #include <net/route.h>
135 
136 /*
137  * Size of the route hash table.  Must be a power of two.
138  */
139 #ifndef BRIDGE_RTHASH_SIZE
140 #define	BRIDGE_RTHASH_SIZE		1024
141 #endif
142 
143 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
144 
145 /*
146  * Default maximum number of addresses to cache.
147  */
148 #ifndef BRIDGE_RTABLE_MAX
149 #define	BRIDGE_RTABLE_MAX		2000
150 #endif
151 
152 /*
153  * Timeout (in seconds) for entries learned dynamically.
154  */
155 #ifndef BRIDGE_RTABLE_TIMEOUT
156 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
157 #endif
158 
159 /*
160  * Number of seconds between walks of the route list.
161  */
162 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
163 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
164 #endif
165 
166 /*
167  * List of capabilities to possibly mask on the member interface.
168  */
169 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
170 
171 /*
172  * List of capabilities to strip
173  */
174 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
175 
176 /*
177  * Bridge interface list entry.
178  */
179 struct bridge_iflist {
180 	LIST_ENTRY(bridge_iflist) bif_next;
181 	struct ifnet		*bif_ifp;	/* member if */
182 	struct bstp_port	bif_stp;	/* STP state */
183 	uint32_t		bif_flags;	/* member if flags */
184 	int			bif_savedcaps;	/* saved capabilities */
185 	uint32_t		bif_addrmax;	/* max # of addresses */
186 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
187 	uint32_t		bif_addrexceeded;/* # of address violations */
188 };
189 
190 /*
191  * Bridge route node.
192  */
193 struct bridge_rtnode {
194 	LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
195 	LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
196 	struct bridge_iflist	*brt_dst;	/* destination if */
197 	unsigned long		brt_expire;	/* expiration time */
198 	uint8_t			brt_flags;	/* address flags */
199 	uint8_t			brt_addr[ETHER_ADDR_LEN];
200 	uint16_t		brt_vlan;	/* vlan id */
201 };
202 #define	brt_ifp			brt_dst->bif_ifp
203 
204 /*
205  * Software state for each bridge.
206  */
207 struct bridge_softc {
208 	struct ifnet		*sc_ifp;	/* make this an interface */
209 	LIST_ENTRY(bridge_softc) sc_list;
210 	struct mtx		sc_mtx;
211 	struct cv		sc_cv;
212 	uint32_t		sc_brtmax;	/* max # of addresses */
213 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
214 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
215 	struct callout		sc_brcallout;	/* bridge callout */
216 	uint32_t		sc_iflist_ref;	/* refcount for sc_iflist */
217 	uint32_t		sc_iflist_xcnt;	/* refcount for sc_iflist */
218 	LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
219 	LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
220 	LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
221 	uint32_t		sc_rthash_key;	/* key for hash */
222 	LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
223 	struct bstp_state	sc_stp;		/* STP state */
224 	uint32_t		sc_brtexceeded;	/* # of cache drops */
225 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
226 	u_char			sc_defaddr[6];	/* Default MAC address */
227 };
228 
229 static struct mtx 	bridge_list_mtx;
230 eventhandler_tag	bridge_detach_cookie = NULL;
231 
232 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
233 
234 uma_zone_t bridge_rtnode_zone;
235 
236 static int	bridge_clone_create(struct if_clone *, int, caddr_t);
237 static void	bridge_clone_destroy(struct ifnet *);
238 
239 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
240 static void	bridge_mutecaps(struct bridge_softc *);
241 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
242 		    int);
243 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
244 static void	bridge_init(void *);
245 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
246 static void	bridge_stop(struct ifnet *, int);
247 static int	bridge_transmit(struct ifnet *, struct mbuf *);
248 static void	bridge_qflush(struct ifnet *);
249 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
250 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
251 		    struct rtentry *);
252 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
253 		    struct mbuf *);
254 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
255 
256 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
257 		    struct mbuf *m);
258 
259 static void	bridge_timer(void *);
260 
261 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
262 		    struct mbuf *, int);
263 static void	bridge_span(struct bridge_softc *, struct mbuf *);
264 
265 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
266 		    uint16_t, struct bridge_iflist *, int, uint8_t);
267 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
268 		    uint16_t);
269 static void	bridge_rttrim(struct bridge_softc *);
270 static void	bridge_rtage(struct bridge_softc *);
271 static void	bridge_rtflush(struct bridge_softc *, int);
272 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
273 		    uint16_t);
274 
275 static void	bridge_rtable_init(struct bridge_softc *);
276 static void	bridge_rtable_fini(struct bridge_softc *);
277 
278 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
279 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
280 		    const uint8_t *, uint16_t);
281 static int	bridge_rtnode_insert(struct bridge_softc *,
282 		    struct bridge_rtnode *);
283 static void	bridge_rtnode_destroy(struct bridge_softc *,
284 		    struct bridge_rtnode *);
285 static void	bridge_rtable_expire(struct ifnet *, int);
286 static void	bridge_state_change(struct ifnet *, int);
287 
288 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
289 		    const char *name);
290 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
291 		    struct ifnet *ifp);
292 static void	bridge_delete_member(struct bridge_softc *,
293 		    struct bridge_iflist *, int);
294 static void	bridge_delete_span(struct bridge_softc *,
295 		    struct bridge_iflist *);
296 
297 static int	bridge_ioctl_add(struct bridge_softc *, void *);
298 static int	bridge_ioctl_del(struct bridge_softc *, void *);
299 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
300 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
301 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
302 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
303 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
304 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
305 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
306 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
307 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
308 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
309 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
310 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
311 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
312 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
313 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
314 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
315 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
316 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
317 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
318 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
319 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
320 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
321 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
322 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
323 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
324 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
325 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
326 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
327 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
328 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
329 		    int);
330 static int	bridge_ip_checkbasic(struct mbuf **mp);
331 #ifdef INET6
332 static int	bridge_ip6_checkbasic(struct mbuf **mp);
333 #endif /* INET6 */
334 static int	bridge_fragment(struct ifnet *, struct mbuf *,
335 		    struct ether_header *, int, struct llc *);
336 static void	bridge_linkstate(struct ifnet *ifp);
337 static void	bridge_linkcheck(struct bridge_softc *sc);
338 
339 extern void (*bridge_linkstate_p)(struct ifnet *ifp);
340 
341 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
342 #define	VLANTAGOF(_m)	\
343     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
344 
345 static struct bstp_cb_ops bridge_ops = {
346 	.bcb_state = bridge_state_change,
347 	.bcb_rtage = bridge_rtable_expire
348 };
349 
350 SYSCTL_DECL(_net_link);
351 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
352 
353 static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */
354 static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */
355 static int pfil_member = 1; /* run pfil hooks on the member interface */
356 static int pfil_ipfw = 0;   /* layer2 filter with ipfw */
357 static int pfil_ipfw_arp = 0;   /* layer2 filter with ipfw */
358 static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for
359                                    locally destined packets */
360 static int log_stp   = 0;   /* log STP state changes */
361 static int bridge_inherit_mac = 0;   /* share MAC with first bridge member */
362 TUNABLE_INT("net.link.bridge.pfil_onlyip", &pfil_onlyip);
363 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
364     &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
365 TUNABLE_INT("net.link.bridge.ipfw_arp", &pfil_ipfw_arp);
366 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
367     &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
368 TUNABLE_INT("net.link.bridge.pfil_bridge", &pfil_bridge);
369 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
370     &pfil_bridge, 0, "Packet filter on the bridge interface");
371 TUNABLE_INT("net.link.bridge.pfil_member", &pfil_member);
372 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
373     &pfil_member, 0, "Packet filter on the member interface");
374 TUNABLE_INT("net.link.bridge.pfil_local_phys", &pfil_local_phys);
375 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
376     &pfil_local_phys, 0,
377     "Packet filter on the physical interface for locally destined packets");
378 TUNABLE_INT("net.link.bridge.log_stp", &log_stp);
379 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
380     &log_stp, 0, "Log STP state changes");
381 TUNABLE_INT("net.link.bridge.inherit_mac", &bridge_inherit_mac);
382 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, CTLFLAG_RW,
383     &bridge_inherit_mac, 0,
384     "Inherit MAC address from the first bridge member");
385 
386 static VNET_DEFINE(int, allow_llz_overlap) = 0;
387 #define	V_allow_llz_overlap	VNET(allow_llz_overlap)
388 SYSCTL_VNET_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap, CTLFLAG_RW,
389     &VNET_NAME(allow_llz_overlap), 0, "Allow overlap of link-local scope "
390     "zones of a bridge interface and the member interfaces");
391 
392 struct bridge_control {
393 	int	(*bc_func)(struct bridge_softc *, void *);
394 	int	bc_argsize;
395 	int	bc_flags;
396 };
397 
398 #define	BC_F_COPYIN		0x01	/* copy arguments in */
399 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
400 #define	BC_F_SUSER		0x04	/* do super-user check */
401 
402 const struct bridge_control bridge_control_table[] = {
403 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
404 	  BC_F_COPYIN|BC_F_SUSER },
405 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
406 	  BC_F_COPYIN|BC_F_SUSER },
407 
408 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
409 	  BC_F_COPYIN|BC_F_COPYOUT },
410 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
411 	  BC_F_COPYIN|BC_F_SUSER },
412 
413 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
414 	  BC_F_COPYIN|BC_F_SUSER },
415 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
416 	  BC_F_COPYOUT },
417 
418 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
419 	  BC_F_COPYIN|BC_F_COPYOUT },
420 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
421 	  BC_F_COPYIN|BC_F_COPYOUT },
422 
423 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
424 	  BC_F_COPYIN|BC_F_SUSER },
425 
426 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
427 	  BC_F_COPYIN|BC_F_SUSER },
428 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
429 	  BC_F_COPYOUT },
430 
431 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
432 	  BC_F_COPYIN|BC_F_SUSER },
433 
434 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
435 	  BC_F_COPYIN|BC_F_SUSER },
436 
437 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
438 	  BC_F_COPYOUT },
439 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
440 	  BC_F_COPYIN|BC_F_SUSER },
441 
442 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
443 	  BC_F_COPYOUT },
444 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
445 	  BC_F_COPYIN|BC_F_SUSER },
446 
447 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
448 	  BC_F_COPYOUT },
449 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
450 	  BC_F_COPYIN|BC_F_SUSER },
451 
452 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
453 	  BC_F_COPYOUT },
454 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
455 	  BC_F_COPYIN|BC_F_SUSER },
456 
457 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
458 	  BC_F_COPYIN|BC_F_SUSER },
459 
460 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
461 	  BC_F_COPYIN|BC_F_SUSER },
462 
463 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
464 	  BC_F_COPYIN|BC_F_SUSER },
465 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
466 	  BC_F_COPYIN|BC_F_SUSER },
467 
468 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
469 	  BC_F_COPYOUT },
470 
471 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
472 	  BC_F_COPYOUT },
473 
474 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
475 	  BC_F_COPYIN|BC_F_COPYOUT },
476 
477 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
478 	  BC_F_COPYIN|BC_F_SUSER },
479 
480 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
481 	  BC_F_COPYIN|BC_F_SUSER },
482 
483 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
484 	  BC_F_COPYIN|BC_F_SUSER },
485 
486 };
487 const int bridge_control_table_size =
488     sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
489 
490 LIST_HEAD(, bridge_softc) bridge_list;
491 
492 static struct if_clone *bridge_cloner;
493 static const char bridge_name[] = "bridge";
494 
495 static int
496 bridge_modevent(module_t mod, int type, void *data)
497 {
498 
499 	switch (type) {
500 	case MOD_LOAD:
501 		mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
502 		bridge_cloner = if_clone_simple(bridge_name,
503 		    bridge_clone_create, bridge_clone_destroy, 0);
504 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
505 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
506 		    UMA_ALIGN_PTR, 0);
507 		LIST_INIT(&bridge_list);
508 		bridge_input_p = bridge_input;
509 		bridge_output_p = bridge_output;
510 		bridge_dn_p = bridge_dummynet;
511 		bridge_linkstate_p = bridge_linkstate;
512 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
513 		    ifnet_departure_event, bridge_ifdetach, NULL,
514 		    EVENTHANDLER_PRI_ANY);
515 		break;
516 	case MOD_UNLOAD:
517 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
518 		    bridge_detach_cookie);
519 		if_clone_detach(bridge_cloner);
520 		uma_zdestroy(bridge_rtnode_zone);
521 		bridge_input_p = NULL;
522 		bridge_output_p = NULL;
523 		bridge_dn_p = NULL;
524 		bridge_linkstate_p = NULL;
525 		mtx_destroy(&bridge_list_mtx);
526 		break;
527 	default:
528 		return (EOPNOTSUPP);
529 	}
530 	return (0);
531 }
532 
533 static moduledata_t bridge_mod = {
534 	"if_bridge",
535 	bridge_modevent,
536 	0
537 };
538 
539 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
540 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
541 
542 /*
543  * handler for net.link.bridge.ipfw
544  */
545 static int
546 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
547 {
548 	int enable = pfil_ipfw;
549 	int error;
550 
551 	error = sysctl_handle_int(oidp, &enable, 0, req);
552 	enable = (enable) ? 1 : 0;
553 
554 	if (enable != pfil_ipfw) {
555 		pfil_ipfw = enable;
556 
557 		/*
558 		 * Disable pfil so that ipfw doesnt run twice, if the user
559 		 * really wants both then they can re-enable pfil_bridge and/or
560 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
561 		 * layer2 type.
562 		 */
563 		if (pfil_ipfw) {
564 			pfil_onlyip = 0;
565 			pfil_bridge = 0;
566 			pfil_member = 0;
567 		}
568 	}
569 
570 	return (error);
571 }
572 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
573 	    &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
574 
575 /*
576  * bridge_clone_create:
577  *
578  *	Create a new bridge instance.
579  */
580 static int
581 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
582 {
583 	struct bridge_softc *sc, *sc2;
584 	struct ifnet *bifp, *ifp;
585 	int fb, retry;
586 	unsigned long hostid;
587 
588 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
589 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
590 	if (ifp == NULL) {
591 		free(sc, M_DEVBUF);
592 		return (ENOSPC);
593 	}
594 
595 	BRIDGE_LOCK_INIT(sc);
596 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
597 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
598 
599 	/* Initialize our routing table. */
600 	bridge_rtable_init(sc);
601 
602 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
603 
604 	LIST_INIT(&sc->sc_iflist);
605 	LIST_INIT(&sc->sc_spanlist);
606 
607 	ifp->if_softc = sc;
608 	if_initname(ifp, bridge_name, unit);
609 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
610 	ifp->if_ioctl = bridge_ioctl;
611 	ifp->if_transmit = bridge_transmit;
612 	ifp->if_qflush = bridge_qflush;
613 	ifp->if_init = bridge_init;
614 	ifp->if_type = IFT_BRIDGE;
615 
616 	/*
617 	 * Generate an ethernet address with a locally administered address.
618 	 *
619 	 * Since we are using random ethernet addresses for the bridge, it is
620 	 * possible that we might have address collisions, so make sure that
621 	 * this hardware address isn't already in use on another bridge.
622 	 * The first try uses the hostid and falls back to arc4rand().
623 	 */
624 	fb = 0;
625 	getcredhostid(curthread->td_ucred, &hostid);
626 	do {
627 		if (fb || hostid == 0) {
628 			arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
629 			sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
630 			sc->sc_defaddr[0] |= 2;	/* set the LAA bit */
631 		} else {
632 			sc->sc_defaddr[0] = 0x2;
633 			sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
634 			sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
635 			sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
636 			sc->sc_defaddr[4] =  hostid        & 0xff;
637 			sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
638 		}
639 
640 		fb = 1;
641 		retry = 0;
642 		mtx_lock(&bridge_list_mtx);
643 		LIST_FOREACH(sc2, &bridge_list, sc_list) {
644 			bifp = sc2->sc_ifp;
645 			if (memcmp(sc->sc_defaddr,
646 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
647 				retry = 1;
648 				break;
649 			}
650 		}
651 		mtx_unlock(&bridge_list_mtx);
652 	} while (retry == 1);
653 
654 	bstp_attach(&sc->sc_stp, &bridge_ops);
655 	ether_ifattach(ifp, sc->sc_defaddr);
656 	/* Now undo some of the damage... */
657 	ifp->if_baudrate = 0;
658 	ifp->if_type = IFT_BRIDGE;
659 
660 	mtx_lock(&bridge_list_mtx);
661 	LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
662 	mtx_unlock(&bridge_list_mtx);
663 
664 	return (0);
665 }
666 
667 /*
668  * bridge_clone_destroy:
669  *
670  *	Destroy a bridge instance.
671  */
672 static void
673 bridge_clone_destroy(struct ifnet *ifp)
674 {
675 	struct bridge_softc *sc = ifp->if_softc;
676 	struct bridge_iflist *bif;
677 
678 	BRIDGE_LOCK(sc);
679 
680 	bridge_stop(ifp, 1);
681 	ifp->if_flags &= ~IFF_UP;
682 
683 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
684 		bridge_delete_member(sc, bif, 0);
685 
686 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
687 		bridge_delete_span(sc, bif);
688 	}
689 
690 	BRIDGE_UNLOCK(sc);
691 
692 	callout_drain(&sc->sc_brcallout);
693 
694 	mtx_lock(&bridge_list_mtx);
695 	LIST_REMOVE(sc, sc_list);
696 	mtx_unlock(&bridge_list_mtx);
697 
698 	bstp_detach(&sc->sc_stp);
699 	ether_ifdetach(ifp);
700 	if_free(ifp);
701 
702 	/* Tear down the routing table. */
703 	bridge_rtable_fini(sc);
704 
705 	BRIDGE_LOCK_DESTROY(sc);
706 	free(sc, M_DEVBUF);
707 }
708 
709 /*
710  * bridge_ioctl:
711  *
712  *	Handle a control request from the operator.
713  */
714 static int
715 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
716 {
717 	struct bridge_softc *sc = ifp->if_softc;
718 	struct ifreq *ifr = (struct ifreq *)data;
719 	struct bridge_iflist *bif;
720 	struct thread *td = curthread;
721 	union {
722 		struct ifbreq ifbreq;
723 		struct ifbifconf ifbifconf;
724 		struct ifbareq ifbareq;
725 		struct ifbaconf ifbaconf;
726 		struct ifbrparam ifbrparam;
727 		struct ifbropreq ifbropreq;
728 	} args;
729 	struct ifdrv *ifd = (struct ifdrv *) data;
730 	const struct bridge_control *bc;
731 	int error = 0;
732 
733 	switch (cmd) {
734 
735 	case SIOCADDMULTI:
736 	case SIOCDELMULTI:
737 		break;
738 
739 	case SIOCGDRVSPEC:
740 	case SIOCSDRVSPEC:
741 		if (ifd->ifd_cmd >= bridge_control_table_size) {
742 			error = EINVAL;
743 			break;
744 		}
745 		bc = &bridge_control_table[ifd->ifd_cmd];
746 
747 		if (cmd == SIOCGDRVSPEC &&
748 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
749 			error = EINVAL;
750 			break;
751 		}
752 		else if (cmd == SIOCSDRVSPEC &&
753 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
754 			error = EINVAL;
755 			break;
756 		}
757 
758 		if (bc->bc_flags & BC_F_SUSER) {
759 			error = priv_check(td, PRIV_NET_BRIDGE);
760 			if (error)
761 				break;
762 		}
763 
764 		if (ifd->ifd_len != bc->bc_argsize ||
765 		    ifd->ifd_len > sizeof(args)) {
766 			error = EINVAL;
767 			break;
768 		}
769 
770 		bzero(&args, sizeof(args));
771 		if (bc->bc_flags & BC_F_COPYIN) {
772 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
773 			if (error)
774 				break;
775 		}
776 
777 		BRIDGE_LOCK(sc);
778 		error = (*bc->bc_func)(sc, &args);
779 		BRIDGE_UNLOCK(sc);
780 		if (error)
781 			break;
782 
783 		if (bc->bc_flags & BC_F_COPYOUT)
784 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
785 
786 		break;
787 
788 	case SIOCSIFFLAGS:
789 		if (!(ifp->if_flags & IFF_UP) &&
790 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
791 			/*
792 			 * If interface is marked down and it is running,
793 			 * then stop and disable it.
794 			 */
795 			BRIDGE_LOCK(sc);
796 			bridge_stop(ifp, 1);
797 			BRIDGE_UNLOCK(sc);
798 		} else if ((ifp->if_flags & IFF_UP) &&
799 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
800 			/*
801 			 * If interface is marked up and it is stopped, then
802 			 * start it.
803 			 */
804 			(*ifp->if_init)(sc);
805 		}
806 		break;
807 
808 	case SIOCSIFMTU:
809 		if (ifr->ifr_mtu < 576) {
810 			error = EINVAL;
811 			break;
812 		}
813 		if (LIST_EMPTY(&sc->sc_iflist)) {
814 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
815 			break;
816 		}
817 		BRIDGE_LOCK(sc);
818 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
819 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
820 				log(LOG_NOTICE, "%s: invalid MTU: %lu(%s)"
821 				    " != %d\n", sc->sc_ifp->if_xname,
822 				    bif->bif_ifp->if_mtu,
823 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
824 				error = EINVAL;
825 				break;
826 			}
827 		}
828 		if (!error)
829 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
830 		BRIDGE_UNLOCK(sc);
831 		break;
832 	default:
833 		/*
834 		 * drop the lock as ether_ioctl() will call bridge_start() and
835 		 * cause the lock to be recursed.
836 		 */
837 		error = ether_ioctl(ifp, cmd, data);
838 		break;
839 	}
840 
841 	return (error);
842 }
843 
844 /*
845  * bridge_mutecaps:
846  *
847  *	Clear or restore unwanted capabilities on the member interface
848  */
849 static void
850 bridge_mutecaps(struct bridge_softc *sc)
851 {
852 	struct bridge_iflist *bif;
853 	int enabled, mask;
854 
855 	/* Initial bitmask of capabilities to test */
856 	mask = BRIDGE_IFCAPS_MASK;
857 
858 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
859 		/* Every member must support it or its disabled */
860 		mask &= bif->bif_savedcaps;
861 	}
862 
863 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
864 		enabled = bif->bif_ifp->if_capenable;
865 		enabled &= ~BRIDGE_IFCAPS_STRIP;
866 		/* strip off mask bits and enable them again if allowed */
867 		enabled &= ~BRIDGE_IFCAPS_MASK;
868 		enabled |= mask;
869 		bridge_set_ifcap(sc, bif, enabled);
870 	}
871 
872 }
873 
874 static void
875 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
876 {
877 	struct ifnet *ifp = bif->bif_ifp;
878 	struct ifreq ifr;
879 	int error;
880 
881 	bzero(&ifr, sizeof(ifr));
882 	ifr.ifr_reqcap = set;
883 
884 	if (ifp->if_capenable != set) {
885 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
886 		if (error)
887 			if_printf(sc->sc_ifp,
888 			    "error setting interface capabilities on %s\n",
889 			    ifp->if_xname);
890 	}
891 }
892 
893 /*
894  * bridge_lookup_member:
895  *
896  *	Lookup a bridge member interface.
897  */
898 static struct bridge_iflist *
899 bridge_lookup_member(struct bridge_softc *sc, const char *name)
900 {
901 	struct bridge_iflist *bif;
902 	struct ifnet *ifp;
903 
904 	BRIDGE_LOCK_ASSERT(sc);
905 
906 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
907 		ifp = bif->bif_ifp;
908 		if (strcmp(ifp->if_xname, name) == 0)
909 			return (bif);
910 	}
911 
912 	return (NULL);
913 }
914 
915 /*
916  * bridge_lookup_member_if:
917  *
918  *	Lookup a bridge member interface by ifnet*.
919  */
920 static struct bridge_iflist *
921 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
922 {
923 	struct bridge_iflist *bif;
924 
925 	BRIDGE_LOCK_ASSERT(sc);
926 
927 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
928 		if (bif->bif_ifp == member_ifp)
929 			return (bif);
930 	}
931 
932 	return (NULL);
933 }
934 
935 /*
936  * bridge_delete_member:
937  *
938  *	Delete the specified member interface.
939  */
940 static void
941 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
942     int gone)
943 {
944 	struct ifnet *ifs = bif->bif_ifp;
945 	struct ifnet *fif = NULL;
946 
947 	BRIDGE_LOCK_ASSERT(sc);
948 
949 	if (bif->bif_flags & IFBIF_STP)
950 		bstp_disable(&bif->bif_stp);
951 
952 	ifs->if_bridge = NULL;
953 	BRIDGE_XLOCK(sc);
954 	LIST_REMOVE(bif, bif_next);
955 	BRIDGE_XDROP(sc);
956 
957 	/*
958 	 * If removing the interface that gave the bridge its mac address, set
959 	 * the mac address of the bridge to the address of the next member, or
960 	 * to its default address if no members are left.
961 	 */
962 	if (bridge_inherit_mac && sc->sc_ifaddr == ifs) {
963 		if (LIST_EMPTY(&sc->sc_iflist)) {
964 			bcopy(sc->sc_defaddr,
965 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
966 			sc->sc_ifaddr = NULL;
967 		} else {
968 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
969 			bcopy(IF_LLADDR(fif),
970 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
971 			sc->sc_ifaddr = fif;
972 		}
973 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
974 	}
975 
976 	bridge_linkcheck(sc);
977 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
978 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
979 	KASSERT(bif->bif_addrcnt == 0,
980 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
981 
982 	BRIDGE_UNLOCK(sc);
983 	if (!gone) {
984 		switch (ifs->if_type) {
985 		case IFT_ETHER:
986 		case IFT_L2VLAN:
987 			/*
988 			 * Take the interface out of promiscuous mode.
989 			 */
990 			(void) ifpromisc(ifs, 0);
991 			break;
992 
993 		case IFT_GIF:
994 			break;
995 
996 		default:
997 #ifdef DIAGNOSTIC
998 			panic("bridge_delete_member: impossible");
999 #endif
1000 			break;
1001 		}
1002 		/* reneable any interface capabilities */
1003 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1004 	}
1005 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1006 	BRIDGE_LOCK(sc);
1007 	free(bif, M_DEVBUF);
1008 }
1009 
1010 /*
1011  * bridge_delete_span:
1012  *
1013  *	Delete the specified span interface.
1014  */
1015 static void
1016 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1017 {
1018 	BRIDGE_LOCK_ASSERT(sc);
1019 
1020 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1021 	    ("%s: not a span interface", __func__));
1022 
1023 	LIST_REMOVE(bif, bif_next);
1024 	free(bif, M_DEVBUF);
1025 }
1026 
1027 static int
1028 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1029 {
1030 	struct ifbreq *req = arg;
1031 	struct bridge_iflist *bif = NULL;
1032 	struct ifnet *ifs;
1033 	int error = 0;
1034 
1035 	ifs = ifunit(req->ifbr_ifsname);
1036 	if (ifs == NULL)
1037 		return (ENOENT);
1038 	if (ifs->if_ioctl == NULL)	/* must be supported */
1039 		return (EINVAL);
1040 
1041 	/* If it's in the span list, it can't be a member. */
1042 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1043 		if (ifs == bif->bif_ifp)
1044 			return (EBUSY);
1045 
1046 	if (ifs->if_bridge == sc)
1047 		return (EEXIST);
1048 
1049 	if (ifs->if_bridge != NULL)
1050 		return (EBUSY);
1051 
1052 	switch (ifs->if_type) {
1053 	case IFT_ETHER:
1054 	case IFT_L2VLAN:
1055 	case IFT_GIF:
1056 		/* permitted interface types */
1057 		break;
1058 	default:
1059 		return (EINVAL);
1060 	}
1061 
1062 #ifdef INET6
1063 	/*
1064 	 * Two valid inet6 addresses with link-local scope must not be
1065 	 * on the parent interface and the member interfaces at the
1066 	 * same time.  This restriction is needed to prevent violation
1067 	 * of link-local scope zone.  Attempts to add a member
1068 	 * interface which has inet6 addresses when the parent has
1069 	 * inet6 triggers removal of all inet6 addresses on the member
1070 	 * interface.
1071 	 */
1072 
1073 	/* Check if the parent interface has a link-local scope addr. */
1074 	if (V_allow_llz_overlap == 0 &&
1075 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1076 		/*
1077 		 * If any, remove all inet6 addresses from the member
1078 		 * interfaces.
1079 		 */
1080 		BRIDGE_XLOCK(sc);
1081 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1082  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1083 				BRIDGE_UNLOCK(sc);
1084 				in6_ifdetach(bif->bif_ifp);
1085 				BRIDGE_LOCK(sc);
1086 				if_printf(sc->sc_ifp,
1087 				    "IPv6 addresses on %s have been removed "
1088 				    "before adding it as a member to prevent "
1089 				    "IPv6 address scope violation.\n",
1090 				    bif->bif_ifp->if_xname);
1091 			}
1092 		}
1093 		BRIDGE_XDROP(sc);
1094 		if (in6ifa_llaonifp(ifs)) {
1095 			BRIDGE_UNLOCK(sc);
1096 			in6_ifdetach(ifs);
1097 			BRIDGE_LOCK(sc);
1098 			if_printf(sc->sc_ifp,
1099 			    "IPv6 addresses on %s have been removed "
1100 			    "before adding it as a member to prevent "
1101 			    "IPv6 address scope violation.\n",
1102 			    ifs->if_xname);
1103 		}
1104 	}
1105 #endif
1106 	/* Allow the first Ethernet member to define the MTU */
1107 	if (LIST_EMPTY(&sc->sc_iflist))
1108 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1109 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1110 		if_printf(sc->sc_ifp, "invalid MTU: %lu(%s) != %lu\n",
1111 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1112 		return (EINVAL);
1113 	}
1114 
1115 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1116 	if (bif == NULL)
1117 		return (ENOMEM);
1118 
1119 	bif->bif_ifp = ifs;
1120 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1121 	bif->bif_savedcaps = ifs->if_capenable;
1122 
1123 	/*
1124 	 * Assign the interface's MAC address to the bridge if it's the first
1125 	 * member and the MAC address of the bridge has not been changed from
1126 	 * the default randomly generated one.
1127 	 */
1128 	if (bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1129 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1130 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1131 		sc->sc_ifaddr = ifs;
1132 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1133 	}
1134 
1135 	ifs->if_bridge = sc;
1136 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1137 	/*
1138 	 * XXX: XLOCK HERE!?!
1139 	 *
1140 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1141 	 */
1142 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1143 
1144 	/* Set interface capabilities to the intersection set of all members */
1145 	bridge_mutecaps(sc);
1146 	bridge_linkcheck(sc);
1147 
1148 	/* Place the interface into promiscuous mode */
1149 	switch (ifs->if_type) {
1150 		case IFT_ETHER:
1151 		case IFT_L2VLAN:
1152 			BRIDGE_UNLOCK(sc);
1153 			error = ifpromisc(ifs, 1);
1154 			BRIDGE_LOCK(sc);
1155 			break;
1156 	}
1157 
1158 	if (error) {
1159 		bridge_delete_member(sc, bif, 0);
1160 		free(bif, M_DEVBUF);
1161 	}
1162 	return (error);
1163 }
1164 
1165 static int
1166 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1167 {
1168 	struct ifbreq *req = arg;
1169 	struct bridge_iflist *bif;
1170 
1171 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1172 	if (bif == NULL)
1173 		return (ENOENT);
1174 
1175 	bridge_delete_member(sc, bif, 0);
1176 
1177 	return (0);
1178 }
1179 
1180 static int
1181 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1182 {
1183 	struct ifbreq *req = arg;
1184 	struct bridge_iflist *bif;
1185 	struct bstp_port *bp;
1186 
1187 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1188 	if (bif == NULL)
1189 		return (ENOENT);
1190 
1191 	bp = &bif->bif_stp;
1192 	req->ifbr_ifsflags = bif->bif_flags;
1193 	req->ifbr_state = bp->bp_state;
1194 	req->ifbr_priority = bp->bp_priority;
1195 	req->ifbr_path_cost = bp->bp_path_cost;
1196 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1197 	req->ifbr_proto = bp->bp_protover;
1198 	req->ifbr_role = bp->bp_role;
1199 	req->ifbr_stpflags = bp->bp_flags;
1200 	req->ifbr_addrcnt = bif->bif_addrcnt;
1201 	req->ifbr_addrmax = bif->bif_addrmax;
1202 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1203 
1204 	/* Copy STP state options as flags */
1205 	if (bp->bp_operedge)
1206 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1207 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1208 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1209 	if (bp->bp_ptp_link)
1210 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1211 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1212 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1213 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1214 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1215 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1216 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1217 	return (0);
1218 }
1219 
1220 static int
1221 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1222 {
1223 	struct ifbreq *req = arg;
1224 	struct bridge_iflist *bif;
1225 	struct bstp_port *bp;
1226 	int error;
1227 
1228 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1229 	if (bif == NULL)
1230 		return (ENOENT);
1231 	bp = &bif->bif_stp;
1232 
1233 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1234 		/* SPAN is readonly */
1235 		return (EINVAL);
1236 
1237 	if (req->ifbr_ifsflags & IFBIF_STP) {
1238 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1239 			error = bstp_enable(&bif->bif_stp);
1240 			if (error)
1241 				return (error);
1242 		}
1243 	} else {
1244 		if ((bif->bif_flags & IFBIF_STP) != 0)
1245 			bstp_disable(&bif->bif_stp);
1246 	}
1247 
1248 	/* Pass on STP flags */
1249 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1250 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1251 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1252 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1253 
1254 	/* Save the bits relating to the bridge */
1255 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1256 
1257 	return (0);
1258 }
1259 
1260 static int
1261 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1262 {
1263 	struct ifbrparam *param = arg;
1264 
1265 	sc->sc_brtmax = param->ifbrp_csize;
1266 	bridge_rttrim(sc);
1267 
1268 	return (0);
1269 }
1270 
1271 static int
1272 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1273 {
1274 	struct ifbrparam *param = arg;
1275 
1276 	param->ifbrp_csize = sc->sc_brtmax;
1277 
1278 	return (0);
1279 }
1280 
1281 static int
1282 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1283 {
1284 	struct ifbifconf *bifc = arg;
1285 	struct bridge_iflist *bif;
1286 	struct ifbreq breq;
1287 	char *buf, *outbuf;
1288 	int count, buflen, len, error = 0;
1289 
1290 	count = 0;
1291 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1292 		count++;
1293 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1294 		count++;
1295 
1296 	buflen = sizeof(breq) * count;
1297 	if (bifc->ifbic_len == 0) {
1298 		bifc->ifbic_len = buflen;
1299 		return (0);
1300 	}
1301 	BRIDGE_UNLOCK(sc);
1302 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1303 	BRIDGE_LOCK(sc);
1304 
1305 	count = 0;
1306 	buf = outbuf;
1307 	len = min(bifc->ifbic_len, buflen);
1308 	bzero(&breq, sizeof(breq));
1309 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1310 		if (len < sizeof(breq))
1311 			break;
1312 
1313 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1314 		    sizeof(breq.ifbr_ifsname));
1315 		/* Fill in the ifbreq structure */
1316 		error = bridge_ioctl_gifflags(sc, &breq);
1317 		if (error)
1318 			break;
1319 		memcpy(buf, &breq, sizeof(breq));
1320 		count++;
1321 		buf += sizeof(breq);
1322 		len -= sizeof(breq);
1323 	}
1324 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1325 		if (len < sizeof(breq))
1326 			break;
1327 
1328 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1329 		    sizeof(breq.ifbr_ifsname));
1330 		breq.ifbr_ifsflags = bif->bif_flags;
1331 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1332 		memcpy(buf, &breq, sizeof(breq));
1333 		count++;
1334 		buf += sizeof(breq);
1335 		len -= sizeof(breq);
1336 	}
1337 
1338 	BRIDGE_UNLOCK(sc);
1339 	bifc->ifbic_len = sizeof(breq) * count;
1340 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1341 	BRIDGE_LOCK(sc);
1342 	free(outbuf, M_TEMP);
1343 	return (error);
1344 }
1345 
1346 static int
1347 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1348 {
1349 	struct ifbaconf *bac = arg;
1350 	struct bridge_rtnode *brt;
1351 	struct ifbareq bareq;
1352 	char *buf, *outbuf;
1353 	int count, buflen, len, error = 0;
1354 
1355 	if (bac->ifbac_len == 0)
1356 		return (0);
1357 
1358 	count = 0;
1359 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1360 		count++;
1361 	buflen = sizeof(bareq) * count;
1362 
1363 	BRIDGE_UNLOCK(sc);
1364 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1365 	BRIDGE_LOCK(sc);
1366 
1367 	count = 0;
1368 	buf = outbuf;
1369 	len = min(bac->ifbac_len, buflen);
1370 	bzero(&bareq, sizeof(bareq));
1371 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1372 		if (len < sizeof(bareq))
1373 			goto out;
1374 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1375 		    sizeof(bareq.ifba_ifsname));
1376 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1377 		bareq.ifba_vlan = brt->brt_vlan;
1378 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1379 				time_uptime < brt->brt_expire)
1380 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1381 		else
1382 			bareq.ifba_expire = 0;
1383 		bareq.ifba_flags = brt->brt_flags;
1384 
1385 		memcpy(buf, &bareq, sizeof(bareq));
1386 		count++;
1387 		buf += sizeof(bareq);
1388 		len -= sizeof(bareq);
1389 	}
1390 out:
1391 	BRIDGE_UNLOCK(sc);
1392 	bac->ifbac_len = sizeof(bareq) * count;
1393 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1394 	BRIDGE_LOCK(sc);
1395 	free(outbuf, M_TEMP);
1396 	return (error);
1397 }
1398 
1399 static int
1400 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1401 {
1402 	struct ifbareq *req = arg;
1403 	struct bridge_iflist *bif;
1404 	int error;
1405 
1406 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1407 	if (bif == NULL)
1408 		return (ENOENT);
1409 
1410 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1411 	    req->ifba_flags);
1412 
1413 	return (error);
1414 }
1415 
1416 static int
1417 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1418 {
1419 	struct ifbrparam *param = arg;
1420 
1421 	sc->sc_brttimeout = param->ifbrp_ctime;
1422 	return (0);
1423 }
1424 
1425 static int
1426 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1427 {
1428 	struct ifbrparam *param = arg;
1429 
1430 	param->ifbrp_ctime = sc->sc_brttimeout;
1431 	return (0);
1432 }
1433 
1434 static int
1435 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1436 {
1437 	struct ifbareq *req = arg;
1438 
1439 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1440 }
1441 
1442 static int
1443 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1444 {
1445 	struct ifbreq *req = arg;
1446 
1447 	bridge_rtflush(sc, req->ifbr_ifsflags);
1448 	return (0);
1449 }
1450 
1451 static int
1452 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1453 {
1454 	struct ifbrparam *param = arg;
1455 	struct bstp_state *bs = &sc->sc_stp;
1456 
1457 	param->ifbrp_prio = bs->bs_bridge_priority;
1458 	return (0);
1459 }
1460 
1461 static int
1462 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1463 {
1464 	struct ifbrparam *param = arg;
1465 
1466 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1467 }
1468 
1469 static int
1470 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1471 {
1472 	struct ifbrparam *param = arg;
1473 	struct bstp_state *bs = &sc->sc_stp;
1474 
1475 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1476 	return (0);
1477 }
1478 
1479 static int
1480 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1481 {
1482 	struct ifbrparam *param = arg;
1483 
1484 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1485 }
1486 
1487 static int
1488 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1489 {
1490 	struct ifbrparam *param = arg;
1491 	struct bstp_state *bs = &sc->sc_stp;
1492 
1493 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1494 	return (0);
1495 }
1496 
1497 static int
1498 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1499 {
1500 	struct ifbrparam *param = arg;
1501 
1502 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1503 }
1504 
1505 static int
1506 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1507 {
1508 	struct ifbrparam *param = arg;
1509 	struct bstp_state *bs = &sc->sc_stp;
1510 
1511 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1512 	return (0);
1513 }
1514 
1515 static int
1516 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1517 {
1518 	struct ifbrparam *param = arg;
1519 
1520 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1521 }
1522 
1523 static int
1524 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1525 {
1526 	struct ifbreq *req = arg;
1527 	struct bridge_iflist *bif;
1528 
1529 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1530 	if (bif == NULL)
1531 		return (ENOENT);
1532 
1533 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1534 }
1535 
1536 static int
1537 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1538 {
1539 	struct ifbreq *req = arg;
1540 	struct bridge_iflist *bif;
1541 
1542 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1543 	if (bif == NULL)
1544 		return (ENOENT);
1545 
1546 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1547 }
1548 
1549 static int
1550 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1551 {
1552 	struct ifbreq *req = arg;
1553 	struct bridge_iflist *bif;
1554 
1555 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1556 	if (bif == NULL)
1557 		return (ENOENT);
1558 
1559 	bif->bif_addrmax = req->ifbr_addrmax;
1560 	return (0);
1561 }
1562 
1563 static int
1564 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1565 {
1566 	struct ifbreq *req = arg;
1567 	struct bridge_iflist *bif = NULL;
1568 	struct ifnet *ifs;
1569 
1570 	ifs = ifunit(req->ifbr_ifsname);
1571 	if (ifs == NULL)
1572 		return (ENOENT);
1573 
1574 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1575 		if (ifs == bif->bif_ifp)
1576 			return (EBUSY);
1577 
1578 	if (ifs->if_bridge != NULL)
1579 		return (EBUSY);
1580 
1581 	switch (ifs->if_type) {
1582 		case IFT_ETHER:
1583 		case IFT_GIF:
1584 		case IFT_L2VLAN:
1585 			break;
1586 		default:
1587 			return (EINVAL);
1588 	}
1589 
1590 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1591 	if (bif == NULL)
1592 		return (ENOMEM);
1593 
1594 	bif->bif_ifp = ifs;
1595 	bif->bif_flags = IFBIF_SPAN;
1596 
1597 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1598 
1599 	return (0);
1600 }
1601 
1602 static int
1603 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1604 {
1605 	struct ifbreq *req = arg;
1606 	struct bridge_iflist *bif;
1607 	struct ifnet *ifs;
1608 
1609 	ifs = ifunit(req->ifbr_ifsname);
1610 	if (ifs == NULL)
1611 		return (ENOENT);
1612 
1613 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1614 		if (ifs == bif->bif_ifp)
1615 			break;
1616 
1617 	if (bif == NULL)
1618 		return (ENOENT);
1619 
1620 	bridge_delete_span(sc, bif);
1621 
1622 	return (0);
1623 }
1624 
1625 static int
1626 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1627 {
1628 	struct ifbropreq *req = arg;
1629 	struct bstp_state *bs = &sc->sc_stp;
1630 	struct bstp_port *root_port;
1631 
1632 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1633 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1634 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1635 
1636 	root_port = bs->bs_root_port;
1637 	if (root_port == NULL)
1638 		req->ifbop_root_port = 0;
1639 	else
1640 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1641 
1642 	req->ifbop_holdcount = bs->bs_txholdcount;
1643 	req->ifbop_priority = bs->bs_bridge_priority;
1644 	req->ifbop_protocol = bs->bs_protover;
1645 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1646 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1647 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1648 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1649 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1650 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1651 
1652 	return (0);
1653 }
1654 
1655 static int
1656 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1657 {
1658 	struct ifbrparam *param = arg;
1659 
1660 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1661 	return (0);
1662 }
1663 
1664 static int
1665 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1666 {
1667 	struct ifbpstpconf *bifstp = arg;
1668 	struct bridge_iflist *bif;
1669 	struct bstp_port *bp;
1670 	struct ifbpstpreq bpreq;
1671 	char *buf, *outbuf;
1672 	int count, buflen, len, error = 0;
1673 
1674 	count = 0;
1675 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1676 		if ((bif->bif_flags & IFBIF_STP) != 0)
1677 			count++;
1678 	}
1679 
1680 	buflen = sizeof(bpreq) * count;
1681 	if (bifstp->ifbpstp_len == 0) {
1682 		bifstp->ifbpstp_len = buflen;
1683 		return (0);
1684 	}
1685 
1686 	BRIDGE_UNLOCK(sc);
1687 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1688 	BRIDGE_LOCK(sc);
1689 
1690 	count = 0;
1691 	buf = outbuf;
1692 	len = min(bifstp->ifbpstp_len, buflen);
1693 	bzero(&bpreq, sizeof(bpreq));
1694 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1695 		if (len < sizeof(bpreq))
1696 			break;
1697 
1698 		if ((bif->bif_flags & IFBIF_STP) == 0)
1699 			continue;
1700 
1701 		bp = &bif->bif_stp;
1702 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1703 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1704 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1705 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1706 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1707 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1708 
1709 		memcpy(buf, &bpreq, sizeof(bpreq));
1710 		count++;
1711 		buf += sizeof(bpreq);
1712 		len -= sizeof(bpreq);
1713 	}
1714 
1715 	BRIDGE_UNLOCK(sc);
1716 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1717 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1718 	BRIDGE_LOCK(sc);
1719 	free(outbuf, M_TEMP);
1720 	return (error);
1721 }
1722 
1723 static int
1724 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1725 {
1726 	struct ifbrparam *param = arg;
1727 
1728 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1729 }
1730 
1731 static int
1732 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1733 {
1734 	struct ifbrparam *param = arg;
1735 
1736 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1737 }
1738 
1739 /*
1740  * bridge_ifdetach:
1741  *
1742  *	Detach an interface from a bridge.  Called when a member
1743  *	interface is detaching.
1744  */
1745 static void
1746 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1747 {
1748 	struct bridge_softc *sc = ifp->if_bridge;
1749 	struct bridge_iflist *bif;
1750 
1751 	if (ifp->if_flags & IFF_RENAMING)
1752 		return;
1753 
1754 	/* Check if the interface is a bridge member */
1755 	if (sc != NULL) {
1756 		BRIDGE_LOCK(sc);
1757 
1758 		bif = bridge_lookup_member_if(sc, ifp);
1759 		if (bif != NULL)
1760 			bridge_delete_member(sc, bif, 1);
1761 
1762 		BRIDGE_UNLOCK(sc);
1763 		return;
1764 	}
1765 
1766 	/* Check if the interface is a span port */
1767 	mtx_lock(&bridge_list_mtx);
1768 	LIST_FOREACH(sc, &bridge_list, sc_list) {
1769 		BRIDGE_LOCK(sc);
1770 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1771 			if (ifp == bif->bif_ifp) {
1772 				bridge_delete_span(sc, bif);
1773 				break;
1774 			}
1775 
1776 		BRIDGE_UNLOCK(sc);
1777 	}
1778 	mtx_unlock(&bridge_list_mtx);
1779 }
1780 
1781 /*
1782  * bridge_init:
1783  *
1784  *	Initialize a bridge interface.
1785  */
1786 static void
1787 bridge_init(void *xsc)
1788 {
1789 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1790 	struct ifnet *ifp = sc->sc_ifp;
1791 
1792 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1793 		return;
1794 
1795 	BRIDGE_LOCK(sc);
1796 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1797 	    bridge_timer, sc);
1798 
1799 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1800 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1801 
1802 	BRIDGE_UNLOCK(sc);
1803 }
1804 
1805 /*
1806  * bridge_stop:
1807  *
1808  *	Stop the bridge interface.
1809  */
1810 static void
1811 bridge_stop(struct ifnet *ifp, int disable)
1812 {
1813 	struct bridge_softc *sc = ifp->if_softc;
1814 
1815 	BRIDGE_LOCK_ASSERT(sc);
1816 
1817 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1818 		return;
1819 
1820 	callout_stop(&sc->sc_brcallout);
1821 	bstp_stop(&sc->sc_stp);
1822 
1823 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1824 
1825 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1826 }
1827 
1828 /*
1829  * bridge_enqueue:
1830  *
1831  *	Enqueue a packet on a bridge member interface.
1832  *
1833  */
1834 static int
1835 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1836 {
1837 	int len, err = 0;
1838 	short mflags;
1839 	struct mbuf *m0;
1840 
1841 	/* We may be sending a fragment so traverse the mbuf */
1842 	for (; m; m = m0) {
1843 		m0 = m->m_nextpkt;
1844 		m->m_nextpkt = NULL;
1845 		len = m->m_pkthdr.len;
1846 		mflags = m->m_flags;
1847 
1848 		/*
1849 		 * If underlying interface can not do VLAN tag insertion itself
1850 		 * then attach a packet tag that holds it.
1851 		 */
1852 		if ((m->m_flags & M_VLANTAG) &&
1853 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1854 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1855 			if (m == NULL) {
1856 				if_printf(dst_ifp,
1857 				    "unable to prepend VLAN header\n");
1858 				dst_ifp->if_oerrors++;
1859 				continue;
1860 			}
1861 			m->m_flags &= ~M_VLANTAG;
1862 		}
1863 
1864 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1865 			m_freem(m0);
1866 			sc->sc_ifp->if_oerrors++;
1867 			break;
1868 		}
1869 
1870 		sc->sc_ifp->if_opackets++;
1871 		sc->sc_ifp->if_obytes += len;
1872 		if (mflags & M_MCAST)
1873 			sc->sc_ifp->if_omcasts++;
1874 	}
1875 
1876 	return (err);
1877 }
1878 
1879 /*
1880  * bridge_dummynet:
1881  *
1882  * 	Receive a queued packet from dummynet and pass it on to the output
1883  * 	interface.
1884  *
1885  *	The mbuf has the Ethernet header already attached.
1886  */
1887 static void
1888 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1889 {
1890 	struct bridge_softc *sc;
1891 
1892 	sc = ifp->if_bridge;
1893 
1894 	/*
1895 	 * The packet didnt originate from a member interface. This should only
1896 	 * ever happen if a member interface is removed while packets are
1897 	 * queued for it.
1898 	 */
1899 	if (sc == NULL) {
1900 		m_freem(m);
1901 		return;
1902 	}
1903 
1904 	if (PFIL_HOOKED(&V_inet_pfil_hook)
1905 #ifdef INET6
1906 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1907 #endif
1908 	    ) {
1909 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1910 			return;
1911 		if (m == NULL)
1912 			return;
1913 	}
1914 
1915 	bridge_enqueue(sc, ifp, m);
1916 }
1917 
1918 /*
1919  * bridge_output:
1920  *
1921  *	Send output from a bridge member interface.  This
1922  *	performs the bridging function for locally originated
1923  *	packets.
1924  *
1925  *	The mbuf has the Ethernet header already attached.  We must
1926  *	enqueue or free the mbuf before returning.
1927  */
1928 static int
1929 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1930     struct rtentry *rt)
1931 {
1932 	struct ether_header *eh;
1933 	struct ifnet *dst_if;
1934 	struct bridge_softc *sc;
1935 	uint16_t vlan;
1936 
1937 	if (m->m_len < ETHER_HDR_LEN) {
1938 		m = m_pullup(m, ETHER_HDR_LEN);
1939 		if (m == NULL)
1940 			return (0);
1941 	}
1942 
1943 	eh = mtod(m, struct ether_header *);
1944 	sc = ifp->if_bridge;
1945 	vlan = VLANTAGOF(m);
1946 
1947 	BRIDGE_LOCK(sc);
1948 
1949 	/*
1950 	 * If bridge is down, but the original output interface is up,
1951 	 * go ahead and send out that interface.  Otherwise, the packet
1952 	 * is dropped below.
1953 	 */
1954 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1955 		dst_if = ifp;
1956 		goto sendunicast;
1957 	}
1958 
1959 	/*
1960 	 * If the packet is a multicast, or we don't know a better way to
1961 	 * get there, send to all interfaces.
1962 	 */
1963 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
1964 		dst_if = NULL;
1965 	else
1966 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1967 	if (dst_if == NULL) {
1968 		struct bridge_iflist *bif;
1969 		struct mbuf *mc;
1970 		int error = 0, used = 0;
1971 
1972 		bridge_span(sc, m);
1973 
1974 		BRIDGE_LOCK2REF(sc, error);
1975 		if (error) {
1976 			m_freem(m);
1977 			return (0);
1978 		}
1979 
1980 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1981 			dst_if = bif->bif_ifp;
1982 
1983 			if (dst_if->if_type == IFT_GIF)
1984 				continue;
1985 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1986 				continue;
1987 
1988 			/*
1989 			 * If this is not the original output interface,
1990 			 * and the interface is participating in spanning
1991 			 * tree, make sure the port is in a state that
1992 			 * allows forwarding.
1993 			 */
1994 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
1995 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1996 				continue;
1997 
1998 			if (LIST_NEXT(bif, bif_next) == NULL) {
1999 				used = 1;
2000 				mc = m;
2001 			} else {
2002 				mc = m_copypacket(m, M_NOWAIT);
2003 				if (mc == NULL) {
2004 					sc->sc_ifp->if_oerrors++;
2005 					continue;
2006 				}
2007 			}
2008 
2009 			bridge_enqueue(sc, dst_if, mc);
2010 		}
2011 		if (used == 0)
2012 			m_freem(m);
2013 		BRIDGE_UNREF(sc);
2014 		return (0);
2015 	}
2016 
2017 sendunicast:
2018 	/*
2019 	 * XXX Spanning tree consideration here?
2020 	 */
2021 
2022 	bridge_span(sc, m);
2023 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2024 		m_freem(m);
2025 		BRIDGE_UNLOCK(sc);
2026 		return (0);
2027 	}
2028 
2029 	BRIDGE_UNLOCK(sc);
2030 	bridge_enqueue(sc, dst_if, m);
2031 	return (0);
2032 }
2033 
2034 /*
2035  * bridge_transmit:
2036  *
2037  *	Do output on a bridge.
2038  *
2039  */
2040 static int
2041 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2042 {
2043 	struct bridge_softc *sc;
2044 	struct ether_header *eh;
2045 	struct ifnet *dst_if;
2046 	int error = 0;
2047 
2048 	sc = ifp->if_softc;
2049 
2050 	ETHER_BPF_MTAP(ifp, m);
2051 
2052 	eh = mtod(m, struct ether_header *);
2053 
2054 	BRIDGE_LOCK(sc);
2055 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2056 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2057 		BRIDGE_UNLOCK(sc);
2058 		error = bridge_enqueue(sc, dst_if, m);
2059 	} else
2060 		bridge_broadcast(sc, ifp, m, 0);
2061 
2062 	return (error);
2063 }
2064 
2065 /*
2066  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2067  */
2068 static void
2069 bridge_qflush(struct ifnet *ifp __unused)
2070 {
2071 }
2072 
2073 /*
2074  * bridge_forward:
2075  *
2076  *	The forwarding function of the bridge.
2077  *
2078  *	NOTE: Releases the lock on return.
2079  */
2080 static void
2081 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2082     struct mbuf *m)
2083 {
2084 	struct bridge_iflist *dbif;
2085 	struct ifnet *src_if, *dst_if, *ifp;
2086 	struct ether_header *eh;
2087 	uint16_t vlan;
2088 	uint8_t *dst;
2089 	int error;
2090 
2091 	src_if = m->m_pkthdr.rcvif;
2092 	ifp = sc->sc_ifp;
2093 
2094 	ifp->if_ipackets++;
2095 	ifp->if_ibytes += m->m_pkthdr.len;
2096 	vlan = VLANTAGOF(m);
2097 
2098 	if ((sbif->bif_flags & IFBIF_STP) &&
2099 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2100 		goto drop;
2101 
2102 	eh = mtod(m, struct ether_header *);
2103 	dst = eh->ether_dhost;
2104 
2105 	/* If the interface is learning, record the address. */
2106 	if (sbif->bif_flags & IFBIF_LEARNING) {
2107 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2108 		    sbif, 0, IFBAF_DYNAMIC);
2109 		/*
2110 		 * If the interface has addresses limits then deny any source
2111 		 * that is not in the cache.
2112 		 */
2113 		if (error && sbif->bif_addrmax)
2114 			goto drop;
2115 	}
2116 
2117 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2118 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2119 		goto drop;
2120 
2121 	/*
2122 	 * At this point, the port either doesn't participate
2123 	 * in spanning tree or it is in the forwarding state.
2124 	 */
2125 
2126 	/*
2127 	 * If the packet is unicast, destined for someone on
2128 	 * "this" side of the bridge, drop it.
2129 	 */
2130 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2131 		dst_if = bridge_rtlookup(sc, dst, vlan);
2132 		if (src_if == dst_if)
2133 			goto drop;
2134 	} else {
2135 		/*
2136 		 * Check if its a reserved multicast address, any address
2137 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2138 		 * bridge.
2139 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2140 		 */
2141 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2142 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2143 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2144 			goto drop;
2145 
2146 		/* ...forward it to all interfaces. */
2147 		ifp->if_imcasts++;
2148 		dst_if = NULL;
2149 	}
2150 
2151 	/*
2152 	 * If we have a destination interface which is a member of our bridge,
2153 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2154 	 * For broadcast or multicast packets, don't bother because it will
2155 	 * be reinjected into ether_input. We do this before we pass the packets
2156 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2157 	 * drop the packet, or possibly modify it, making it difficult to debug
2158 	 * firewall issues on the bridge.
2159 	 */
2160 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2161 		ETHER_BPF_MTAP(ifp, m);
2162 
2163 	/* run the packet filter */
2164 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2165 #ifdef INET6
2166 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2167 #endif
2168 	    ) {
2169 		BRIDGE_UNLOCK(sc);
2170 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2171 			return;
2172 		if (m == NULL)
2173 			return;
2174 		BRIDGE_LOCK(sc);
2175 	}
2176 
2177 	if (dst_if == NULL) {
2178 		bridge_broadcast(sc, src_if, m, 1);
2179 		return;
2180 	}
2181 
2182 	/*
2183 	 * At this point, we're dealing with a unicast frame
2184 	 * going to a different interface.
2185 	 */
2186 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2187 		goto drop;
2188 
2189 	dbif = bridge_lookup_member_if(sc, dst_if);
2190 	if (dbif == NULL)
2191 		/* Not a member of the bridge (anymore?) */
2192 		goto drop;
2193 
2194 	/* Private segments can not talk to each other */
2195 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2196 		goto drop;
2197 
2198 	if ((dbif->bif_flags & IFBIF_STP) &&
2199 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2200 		goto drop;
2201 
2202 	BRIDGE_UNLOCK(sc);
2203 
2204 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2205 #ifdef INET6
2206 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2207 #endif
2208 	    ) {
2209 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2210 			return;
2211 		if (m == NULL)
2212 			return;
2213 	}
2214 
2215 	bridge_enqueue(sc, dst_if, m);
2216 	return;
2217 
2218 drop:
2219 	BRIDGE_UNLOCK(sc);
2220 	m_freem(m);
2221 }
2222 
2223 /*
2224  * bridge_input:
2225  *
2226  *	Receive input from a member interface.  Queue the packet for
2227  *	bridging if it is not for us.
2228  */
2229 static struct mbuf *
2230 bridge_input(struct ifnet *ifp, struct mbuf *m)
2231 {
2232 	struct bridge_softc *sc = ifp->if_bridge;
2233 	struct bridge_iflist *bif, *bif2;
2234 	struct ifnet *bifp;
2235 	struct ether_header *eh;
2236 	struct mbuf *mc, *mc2;
2237 	uint16_t vlan;
2238 	int error;
2239 
2240 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2241 		return (m);
2242 
2243 	bifp = sc->sc_ifp;
2244 	vlan = VLANTAGOF(m);
2245 
2246 	/*
2247 	 * Implement support for bridge monitoring. If this flag has been
2248 	 * set on this interface, discard the packet once we push it through
2249 	 * the bpf(4) machinery, but before we do, increment the byte and
2250 	 * packet counters associated with this interface.
2251 	 */
2252 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2253 		m->m_pkthdr.rcvif  = bifp;
2254 		ETHER_BPF_MTAP(bifp, m);
2255 		bifp->if_ipackets++;
2256 		bifp->if_ibytes += m->m_pkthdr.len;
2257 		m_freem(m);
2258 		return (NULL);
2259 	}
2260 	BRIDGE_LOCK(sc);
2261 	bif = bridge_lookup_member_if(sc, ifp);
2262 	if (bif == NULL) {
2263 		BRIDGE_UNLOCK(sc);
2264 		return (m);
2265 	}
2266 
2267 	eh = mtod(m, struct ether_header *);
2268 
2269 	bridge_span(sc, m);
2270 
2271 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2272 		/* Tap off 802.1D packets; they do not get forwarded. */
2273 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2274 		    ETHER_ADDR_LEN) == 0) {
2275 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2276 			BRIDGE_UNLOCK(sc);
2277 			return (NULL);
2278 		}
2279 
2280 		if ((bif->bif_flags & IFBIF_STP) &&
2281 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2282 			BRIDGE_UNLOCK(sc);
2283 			return (m);
2284 		}
2285 
2286 		/*
2287 		 * Make a deep copy of the packet and enqueue the copy
2288 		 * for bridge processing; return the original packet for
2289 		 * local processing.
2290 		 */
2291 		mc = m_dup(m, M_NOWAIT);
2292 		if (mc == NULL) {
2293 			BRIDGE_UNLOCK(sc);
2294 			return (m);
2295 		}
2296 
2297 		/* Perform the bridge forwarding function with the copy. */
2298 		bridge_forward(sc, bif, mc);
2299 
2300 		/*
2301 		 * Reinject the mbuf as arriving on the bridge so we have a
2302 		 * chance at claiming multicast packets. We can not loop back
2303 		 * here from ether_input as a bridge is never a member of a
2304 		 * bridge.
2305 		 */
2306 		KASSERT(bifp->if_bridge == NULL,
2307 		    ("loop created in bridge_input"));
2308 		mc2 = m_dup(m, M_NOWAIT);
2309 		if (mc2 != NULL) {
2310 			/* Keep the layer3 header aligned */
2311 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2312 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2313 		}
2314 		if (mc2 != NULL) {
2315 			mc2->m_pkthdr.rcvif = bifp;
2316 			(*bifp->if_input)(bifp, mc2);
2317 		}
2318 
2319 		/* Return the original packet for local processing. */
2320 		return (m);
2321 	}
2322 
2323 	if ((bif->bif_flags & IFBIF_STP) &&
2324 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2325 		BRIDGE_UNLOCK(sc);
2326 		return (m);
2327 	}
2328 
2329 #if (defined(INET) || defined(INET6))
2330 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2331 	|| ((iface)->if_carp \
2332 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2333 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2334 	|| ((iface)->if_carp \
2335 	    && (*carp_forus_p)((iface), eh->ether_shost))
2336 #else
2337 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2338 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2339 #endif
2340 
2341 #ifdef INET6
2342 #   define OR_PFIL_HOOKED_INET6 \
2343 	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2344 #else
2345 #   define OR_PFIL_HOOKED_INET6
2346 #endif
2347 
2348 #define GRAB_OUR_PACKETS(iface) \
2349 	if ((iface)->if_type == IFT_GIF) \
2350 		continue; \
2351 	/* It is destined for us. */ \
2352 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2353 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2354 	    ) {								\
2355 		if ((iface)->if_type == IFT_BRIDGE) {			\
2356 			ETHER_BPF_MTAP(iface, m);			\
2357 			iface->if_ipackets++;				\
2358 			iface->if_ibytes += m->m_pkthdr.len;		\
2359 			/* Filter on the physical interface. */		\
2360 			if (pfil_local_phys &&				\
2361 			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2362 			     OR_PFIL_HOOKED_INET6)) {			\
2363 				if (bridge_pfil(&m, NULL, ifp,		\
2364 				    PFIL_IN) != 0 || m == NULL) {	\
2365 					BRIDGE_UNLOCK(sc);		\
2366 					return (NULL);			\
2367 				}					\
2368 				eh = mtod(m, struct ether_header *);	\
2369 			}						\
2370 		}							\
2371 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2372 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2373 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2374 			if (error && bif->bif_addrmax) {		\
2375 				BRIDGE_UNLOCK(sc);			\
2376 				m_freem(m);				\
2377 				return (NULL);				\
2378 			}						\
2379 		}							\
2380 		m->m_pkthdr.rcvif = iface;				\
2381 		BRIDGE_UNLOCK(sc);					\
2382 		return (m);						\
2383 	}								\
2384 									\
2385 	/* We just received a packet that we sent out. */		\
2386 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2387 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2388 	    ) {								\
2389 		BRIDGE_UNLOCK(sc);					\
2390 		m_freem(m);						\
2391 		return (NULL);						\
2392 	}
2393 
2394 	/*
2395 	 * Unicast.  Make sure it's not for the bridge.
2396 	 */
2397 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2398 
2399 	/*
2400 	 * Give a chance for ifp at first priority. This will help when	the
2401 	 * packet comes through the interface like VLAN's with the same MACs
2402 	 * on several interfaces from the same bridge. This also will save
2403 	 * some CPU cycles in case the destination interface and the input
2404 	 * interface (eq ifp) are the same.
2405 	 */
2406 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2407 
2408 	/* Now check the all bridge members. */
2409 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2410 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2411 	}
2412 
2413 #undef OR_CARP_CHECK_WE_ARE_DST
2414 #undef OR_CARP_CHECK_WE_ARE_SRC
2415 #undef OR_PFIL_HOOKED_INET6
2416 #undef GRAB_OUR_PACKETS
2417 
2418 	/* Perform the bridge forwarding function. */
2419 	bridge_forward(sc, bif, m);
2420 
2421 	return (NULL);
2422 }
2423 
2424 /*
2425  * bridge_broadcast:
2426  *
2427  *	Send a frame to all interfaces that are members of
2428  *	the bridge, except for the one on which the packet
2429  *	arrived.
2430  *
2431  *	NOTE: Releases the lock on return.
2432  */
2433 static void
2434 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2435     struct mbuf *m, int runfilt)
2436 {
2437 	struct bridge_iflist *dbif, *sbif;
2438 	struct mbuf *mc;
2439 	struct ifnet *dst_if;
2440 	int error = 0, used = 0, i;
2441 
2442 	sbif = bridge_lookup_member_if(sc, src_if);
2443 
2444 	BRIDGE_LOCK2REF(sc, error);
2445 	if (error) {
2446 		m_freem(m);
2447 		return;
2448 	}
2449 
2450 	/* Filter on the bridge interface before broadcasting */
2451 	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2452 #ifdef INET6
2453 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2454 #endif
2455 	    )) {
2456 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2457 			goto out;
2458 		if (m == NULL)
2459 			goto out;
2460 	}
2461 
2462 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2463 		dst_if = dbif->bif_ifp;
2464 		if (dst_if == src_if)
2465 			continue;
2466 
2467 		/* Private segments can not talk to each other */
2468 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2469 			continue;
2470 
2471 		if ((dbif->bif_flags & IFBIF_STP) &&
2472 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2473 			continue;
2474 
2475 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2476 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2477 			continue;
2478 
2479 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2480 			continue;
2481 
2482 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2483 			mc = m;
2484 			used = 1;
2485 		} else {
2486 			mc = m_dup(m, M_NOWAIT);
2487 			if (mc == NULL) {
2488 				sc->sc_ifp->if_oerrors++;
2489 				continue;
2490 			}
2491 		}
2492 
2493 		/*
2494 		 * Filter on the output interface. Pass a NULL bridge interface
2495 		 * pointer so we do not redundantly filter on the bridge for
2496 		 * each interface we broadcast on.
2497 		 */
2498 		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2499 #ifdef INET6
2500 		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2501 #endif
2502 		    )) {
2503 			if (used == 0) {
2504 				/* Keep the layer3 header aligned */
2505 				i = min(mc->m_pkthdr.len, max_protohdr);
2506 				mc = m_copyup(mc, i, ETHER_ALIGN);
2507 				if (mc == NULL) {
2508 					sc->sc_ifp->if_oerrors++;
2509 					continue;
2510 				}
2511 			}
2512 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2513 				continue;
2514 			if (mc == NULL)
2515 				continue;
2516 		}
2517 
2518 		bridge_enqueue(sc, dst_if, mc);
2519 	}
2520 	if (used == 0)
2521 		m_freem(m);
2522 
2523 out:
2524 	BRIDGE_UNREF(sc);
2525 }
2526 
2527 /*
2528  * bridge_span:
2529  *
2530  *	Duplicate a packet out one or more interfaces that are in span mode,
2531  *	the original mbuf is unmodified.
2532  */
2533 static void
2534 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2535 {
2536 	struct bridge_iflist *bif;
2537 	struct ifnet *dst_if;
2538 	struct mbuf *mc;
2539 
2540 	if (LIST_EMPTY(&sc->sc_spanlist))
2541 		return;
2542 
2543 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2544 		dst_if = bif->bif_ifp;
2545 
2546 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2547 			continue;
2548 
2549 		mc = m_copypacket(m, M_NOWAIT);
2550 		if (mc == NULL) {
2551 			sc->sc_ifp->if_oerrors++;
2552 			continue;
2553 		}
2554 
2555 		bridge_enqueue(sc, dst_if, mc);
2556 	}
2557 }
2558 
2559 /*
2560  * bridge_rtupdate:
2561  *
2562  *	Add a bridge routing entry.
2563  */
2564 static int
2565 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2566     struct bridge_iflist *bif, int setflags, uint8_t flags)
2567 {
2568 	struct bridge_rtnode *brt;
2569 	int error;
2570 
2571 	BRIDGE_LOCK_ASSERT(sc);
2572 
2573 	/* Check the source address is valid and not multicast. */
2574 	if (ETHER_IS_MULTICAST(dst) ||
2575 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2576 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2577 		return (EINVAL);
2578 
2579 	/* 802.1p frames map to vlan 1 */
2580 	if (vlan == 0)
2581 		vlan = 1;
2582 
2583 	/*
2584 	 * A route for this destination might already exist.  If so,
2585 	 * update it, otherwise create a new one.
2586 	 */
2587 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2588 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2589 			sc->sc_brtexceeded++;
2590 			return (ENOSPC);
2591 		}
2592 		/* Check per interface address limits (if enabled) */
2593 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2594 			bif->bif_addrexceeded++;
2595 			return (ENOSPC);
2596 		}
2597 
2598 		/*
2599 		 * Allocate a new bridge forwarding node, and
2600 		 * initialize the expiration time and Ethernet
2601 		 * address.
2602 		 */
2603 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2604 		if (brt == NULL)
2605 			return (ENOMEM);
2606 
2607 		if (bif->bif_flags & IFBIF_STICKY)
2608 			brt->brt_flags = IFBAF_STICKY;
2609 		else
2610 			brt->brt_flags = IFBAF_DYNAMIC;
2611 
2612 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2613 		brt->brt_vlan = vlan;
2614 
2615 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2616 			uma_zfree(bridge_rtnode_zone, brt);
2617 			return (error);
2618 		}
2619 		brt->brt_dst = bif;
2620 		bif->bif_addrcnt++;
2621 	}
2622 
2623 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2624 	    brt->brt_dst != bif) {
2625 		brt->brt_dst->bif_addrcnt--;
2626 		brt->brt_dst = bif;
2627 		brt->brt_dst->bif_addrcnt++;
2628 	}
2629 
2630 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2631 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2632 	if (setflags)
2633 		brt->brt_flags = flags;
2634 
2635 	return (0);
2636 }
2637 
2638 /*
2639  * bridge_rtlookup:
2640  *
2641  *	Lookup the destination interface for an address.
2642  */
2643 static struct ifnet *
2644 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2645 {
2646 	struct bridge_rtnode *brt;
2647 
2648 	BRIDGE_LOCK_ASSERT(sc);
2649 
2650 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2651 		return (NULL);
2652 
2653 	return (brt->brt_ifp);
2654 }
2655 
2656 /*
2657  * bridge_rttrim:
2658  *
2659  *	Trim the routine table so that we have a number
2660  *	of routing entries less than or equal to the
2661  *	maximum number.
2662  */
2663 static void
2664 bridge_rttrim(struct bridge_softc *sc)
2665 {
2666 	struct bridge_rtnode *brt, *nbrt;
2667 
2668 	BRIDGE_LOCK_ASSERT(sc);
2669 
2670 	/* Make sure we actually need to do this. */
2671 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2672 		return;
2673 
2674 	/* Force an aging cycle; this might trim enough addresses. */
2675 	bridge_rtage(sc);
2676 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2677 		return;
2678 
2679 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2680 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2681 			bridge_rtnode_destroy(sc, brt);
2682 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2683 				return;
2684 		}
2685 	}
2686 }
2687 
2688 /*
2689  * bridge_timer:
2690  *
2691  *	Aging timer for the bridge.
2692  */
2693 static void
2694 bridge_timer(void *arg)
2695 {
2696 	struct bridge_softc *sc = arg;
2697 
2698 	BRIDGE_LOCK_ASSERT(sc);
2699 
2700 	bridge_rtage(sc);
2701 
2702 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2703 		callout_reset(&sc->sc_brcallout,
2704 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2705 }
2706 
2707 /*
2708  * bridge_rtage:
2709  *
2710  *	Perform an aging cycle.
2711  */
2712 static void
2713 bridge_rtage(struct bridge_softc *sc)
2714 {
2715 	struct bridge_rtnode *brt, *nbrt;
2716 
2717 	BRIDGE_LOCK_ASSERT(sc);
2718 
2719 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2720 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2721 			if (time_uptime >= brt->brt_expire)
2722 				bridge_rtnode_destroy(sc, brt);
2723 		}
2724 	}
2725 }
2726 
2727 /*
2728  * bridge_rtflush:
2729  *
2730  *	Remove all dynamic addresses from the bridge.
2731  */
2732 static void
2733 bridge_rtflush(struct bridge_softc *sc, int full)
2734 {
2735 	struct bridge_rtnode *brt, *nbrt;
2736 
2737 	BRIDGE_LOCK_ASSERT(sc);
2738 
2739 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2740 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2741 			bridge_rtnode_destroy(sc, brt);
2742 	}
2743 }
2744 
2745 /*
2746  * bridge_rtdaddr:
2747  *
2748  *	Remove an address from the table.
2749  */
2750 static int
2751 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2752 {
2753 	struct bridge_rtnode *brt;
2754 	int found = 0;
2755 
2756 	BRIDGE_LOCK_ASSERT(sc);
2757 
2758 	/*
2759 	 * If vlan is zero then we want to delete for all vlans so the lookup
2760 	 * may return more than one.
2761 	 */
2762 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2763 		bridge_rtnode_destroy(sc, brt);
2764 		found = 1;
2765 	}
2766 
2767 	return (found ? 0 : ENOENT);
2768 }
2769 
2770 /*
2771  * bridge_rtdelete:
2772  *
2773  *	Delete routes to a speicifc member interface.
2774  */
2775 static void
2776 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2777 {
2778 	struct bridge_rtnode *brt, *nbrt;
2779 
2780 	BRIDGE_LOCK_ASSERT(sc);
2781 
2782 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2783 		if (brt->brt_ifp == ifp && (full ||
2784 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2785 			bridge_rtnode_destroy(sc, brt);
2786 	}
2787 }
2788 
2789 /*
2790  * bridge_rtable_init:
2791  *
2792  *	Initialize the route table for this bridge.
2793  */
2794 static void
2795 bridge_rtable_init(struct bridge_softc *sc)
2796 {
2797 	int i;
2798 
2799 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2800 	    M_DEVBUF, M_WAITOK);
2801 
2802 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2803 		LIST_INIT(&sc->sc_rthash[i]);
2804 
2805 	sc->sc_rthash_key = arc4random();
2806 	LIST_INIT(&sc->sc_rtlist);
2807 }
2808 
2809 /*
2810  * bridge_rtable_fini:
2811  *
2812  *	Deconstruct the route table for this bridge.
2813  */
2814 static void
2815 bridge_rtable_fini(struct bridge_softc *sc)
2816 {
2817 
2818 	KASSERT(sc->sc_brtcnt == 0,
2819 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2820 	free(sc->sc_rthash, M_DEVBUF);
2821 }
2822 
2823 /*
2824  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2825  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2826  */
2827 #define	mix(a, b, c)							\
2828 do {									\
2829 	a -= b; a -= c; a ^= (c >> 13);					\
2830 	b -= c; b -= a; b ^= (a << 8);					\
2831 	c -= a; c -= b; c ^= (b >> 13);					\
2832 	a -= b; a -= c; a ^= (c >> 12);					\
2833 	b -= c; b -= a; b ^= (a << 16);					\
2834 	c -= a; c -= b; c ^= (b >> 5);					\
2835 	a -= b; a -= c; a ^= (c >> 3);					\
2836 	b -= c; b -= a; b ^= (a << 10);					\
2837 	c -= a; c -= b; c ^= (b >> 15);					\
2838 } while (/*CONSTCOND*/0)
2839 
2840 static __inline uint32_t
2841 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2842 {
2843 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2844 
2845 	b += addr[5] << 8;
2846 	b += addr[4];
2847 	a += addr[3] << 24;
2848 	a += addr[2] << 16;
2849 	a += addr[1] << 8;
2850 	a += addr[0];
2851 
2852 	mix(a, b, c);
2853 
2854 	return (c & BRIDGE_RTHASH_MASK);
2855 }
2856 
2857 #undef mix
2858 
2859 static int
2860 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2861 {
2862 	int i, d;
2863 
2864 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2865 		d = ((int)a[i]) - ((int)b[i]);
2866 	}
2867 
2868 	return (d);
2869 }
2870 
2871 /*
2872  * bridge_rtnode_lookup:
2873  *
2874  *	Look up a bridge route node for the specified destination. Compare the
2875  *	vlan id or if zero then just return the first match.
2876  */
2877 static struct bridge_rtnode *
2878 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2879 {
2880 	struct bridge_rtnode *brt;
2881 	uint32_t hash;
2882 	int dir;
2883 
2884 	BRIDGE_LOCK_ASSERT(sc);
2885 
2886 	hash = bridge_rthash(sc, addr);
2887 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2888 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2889 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2890 			return (brt);
2891 		if (dir > 0)
2892 			return (NULL);
2893 	}
2894 
2895 	return (NULL);
2896 }
2897 
2898 /*
2899  * bridge_rtnode_insert:
2900  *
2901  *	Insert the specified bridge node into the route table.  We
2902  *	assume the entry is not already in the table.
2903  */
2904 static int
2905 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2906 {
2907 	struct bridge_rtnode *lbrt;
2908 	uint32_t hash;
2909 	int dir;
2910 
2911 	BRIDGE_LOCK_ASSERT(sc);
2912 
2913 	hash = bridge_rthash(sc, brt->brt_addr);
2914 
2915 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2916 	if (lbrt == NULL) {
2917 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2918 		goto out;
2919 	}
2920 
2921 	do {
2922 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2923 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2924 			return (EEXIST);
2925 		if (dir > 0) {
2926 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2927 			goto out;
2928 		}
2929 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2930 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2931 			goto out;
2932 		}
2933 		lbrt = LIST_NEXT(lbrt, brt_hash);
2934 	} while (lbrt != NULL);
2935 
2936 #ifdef DIAGNOSTIC
2937 	panic("bridge_rtnode_insert: impossible");
2938 #endif
2939 
2940 out:
2941 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2942 	sc->sc_brtcnt++;
2943 
2944 	return (0);
2945 }
2946 
2947 /*
2948  * bridge_rtnode_destroy:
2949  *
2950  *	Destroy a bridge rtnode.
2951  */
2952 static void
2953 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2954 {
2955 	BRIDGE_LOCK_ASSERT(sc);
2956 
2957 	LIST_REMOVE(brt, brt_hash);
2958 
2959 	LIST_REMOVE(brt, brt_list);
2960 	sc->sc_brtcnt--;
2961 	brt->brt_dst->bif_addrcnt--;
2962 	uma_zfree(bridge_rtnode_zone, brt);
2963 }
2964 
2965 /*
2966  * bridge_rtable_expire:
2967  *
2968  *	Set the expiry time for all routes on an interface.
2969  */
2970 static void
2971 bridge_rtable_expire(struct ifnet *ifp, int age)
2972 {
2973 	struct bridge_softc *sc = ifp->if_bridge;
2974 	struct bridge_rtnode *brt;
2975 
2976 	BRIDGE_LOCK(sc);
2977 
2978 	/*
2979 	 * If the age is zero then flush, otherwise set all the expiry times to
2980 	 * age for the interface
2981 	 */
2982 	if (age == 0)
2983 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2984 	else {
2985 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2986 			/* Cap the expiry time to 'age' */
2987 			if (brt->brt_ifp == ifp &&
2988 			    brt->brt_expire > time_uptime + age &&
2989 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2990 				brt->brt_expire = time_uptime + age;
2991 		}
2992 	}
2993 	BRIDGE_UNLOCK(sc);
2994 }
2995 
2996 /*
2997  * bridge_state_change:
2998  *
2999  *	Callback from the bridgestp code when a port changes states.
3000  */
3001 static void
3002 bridge_state_change(struct ifnet *ifp, int state)
3003 {
3004 	struct bridge_softc *sc = ifp->if_bridge;
3005 	static const char *stpstates[] = {
3006 		"disabled",
3007 		"listening",
3008 		"learning",
3009 		"forwarding",
3010 		"blocking",
3011 		"discarding"
3012 	};
3013 
3014 	if (log_stp)
3015 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3016 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3017 }
3018 
3019 /*
3020  * Send bridge packets through pfil if they are one of the types pfil can deal
3021  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3022  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3023  * that interface.
3024  */
3025 static int
3026 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3027 {
3028 	int snap, error, i, hlen;
3029 	struct ether_header *eh1, eh2;
3030 	struct ip *ip;
3031 	struct llc llc1;
3032 	u_int16_t ether_type;
3033 
3034 	snap = 0;
3035 	error = -1;	/* Default error if not error == 0 */
3036 
3037 #if 0
3038 	/* we may return with the IP fields swapped, ensure its not shared */
3039 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3040 #endif
3041 
3042 	if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
3043 		return (0); /* filtering is disabled */
3044 
3045 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3046 	if ((*mp)->m_len < i) {
3047 	    *mp = m_pullup(*mp, i);
3048 	    if (*mp == NULL) {
3049 		printf("%s: m_pullup failed\n", __func__);
3050 		return (-1);
3051 	    }
3052 	}
3053 
3054 	eh1 = mtod(*mp, struct ether_header *);
3055 	ether_type = ntohs(eh1->ether_type);
3056 
3057 	/*
3058 	 * Check for SNAP/LLC.
3059 	 */
3060 	if (ether_type < ETHERMTU) {
3061 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3062 
3063 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3064 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3065 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3066 		    llc2->llc_control == LLC_UI) {
3067 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3068 			snap = 1;
3069 		}
3070 	}
3071 
3072 	/*
3073 	 * If we're trying to filter bridge traffic, don't look at anything
3074 	 * other than IP and ARP traffic.  If the filter doesn't understand
3075 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3076 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3077 	 * but of course we don't have an AppleTalk filter to begin with.
3078 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3079 	 * ARP traffic.)
3080 	 */
3081 	switch (ether_type) {
3082 		case ETHERTYPE_ARP:
3083 		case ETHERTYPE_REVARP:
3084 			if (pfil_ipfw_arp == 0)
3085 				return (0); /* Automatically pass */
3086 			break;
3087 
3088 		case ETHERTYPE_IP:
3089 #ifdef INET6
3090 		case ETHERTYPE_IPV6:
3091 #endif /* INET6 */
3092 			break;
3093 		default:
3094 			/*
3095 			 * Check to see if the user wants to pass non-ip
3096 			 * packets, these will not be checked by pfil(9) and
3097 			 * passed unconditionally so the default is to drop.
3098 			 */
3099 			if (pfil_onlyip)
3100 				goto bad;
3101 	}
3102 
3103 	/* Run the packet through pfil before stripping link headers */
3104 	if (PFIL_HOOKED(&V_link_pfil_hook) && pfil_ipfw != 0 &&
3105 			dir == PFIL_OUT && ifp != NULL) {
3106 
3107 		error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, NULL);
3108 
3109 		if (*mp == NULL || error != 0) /* packet consumed by filter */
3110 			return (error);
3111 	}
3112 
3113 	/* Strip off the Ethernet header and keep a copy. */
3114 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3115 	m_adj(*mp, ETHER_HDR_LEN);
3116 
3117 	/* Strip off snap header, if present */
3118 	if (snap) {
3119 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3120 		m_adj(*mp, sizeof(struct llc));
3121 	}
3122 
3123 	/*
3124 	 * Check the IP header for alignment and errors
3125 	 */
3126 	if (dir == PFIL_IN) {
3127 		switch (ether_type) {
3128 			case ETHERTYPE_IP:
3129 				error = bridge_ip_checkbasic(mp);
3130 				break;
3131 #ifdef INET6
3132 			case ETHERTYPE_IPV6:
3133 				error = bridge_ip6_checkbasic(mp);
3134 				break;
3135 #endif /* INET6 */
3136 			default:
3137 				error = 0;
3138 		}
3139 		if (error)
3140 			goto bad;
3141 	}
3142 
3143 	error = 0;
3144 
3145 	/*
3146 	 * Run the packet through pfil
3147 	 */
3148 	switch (ether_type) {
3149 	case ETHERTYPE_IP:
3150 		/*
3151 		 * Run pfil on the member interface and the bridge, both can
3152 		 * be skipped by clearing pfil_member or pfil_bridge.
3153 		 *
3154 		 * Keep the order:
3155 		 *   in_if -> bridge_if -> out_if
3156 		 */
3157 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3158 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3159 					dir, NULL);
3160 
3161 		if (*mp == NULL || error != 0) /* filter may consume */
3162 			break;
3163 
3164 		if (pfil_member && ifp != NULL)
3165 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3166 					dir, NULL);
3167 
3168 		if (*mp == NULL || error != 0) /* filter may consume */
3169 			break;
3170 
3171 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3172 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3173 					dir, NULL);
3174 
3175 		if (*mp == NULL || error != 0) /* filter may consume */
3176 			break;
3177 
3178 		/* check if we need to fragment the packet */
3179 		if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3180 			i = (*mp)->m_pkthdr.len;
3181 			if (i > ifp->if_mtu) {
3182 				error = bridge_fragment(ifp, *mp, &eh2, snap,
3183 					    &llc1);
3184 				return (error);
3185 			}
3186 		}
3187 
3188 		/* Recalculate the ip checksum. */
3189 		ip = mtod(*mp, struct ip *);
3190 		hlen = ip->ip_hl << 2;
3191 		if (hlen < sizeof(struct ip))
3192 			goto bad;
3193 		if (hlen > (*mp)->m_len) {
3194 			if ((*mp = m_pullup(*mp, hlen)) == 0)
3195 				goto bad;
3196 			ip = mtod(*mp, struct ip *);
3197 			if (ip == NULL)
3198 				goto bad;
3199 		}
3200 		ip->ip_sum = 0;
3201 		if (hlen == sizeof(struct ip))
3202 			ip->ip_sum = in_cksum_hdr(ip);
3203 		else
3204 			ip->ip_sum = in_cksum(*mp, hlen);
3205 
3206 		break;
3207 #ifdef INET6
3208 	case ETHERTYPE_IPV6:
3209 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3210 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3211 					dir, NULL);
3212 
3213 		if (*mp == NULL || error != 0) /* filter may consume */
3214 			break;
3215 
3216 		if (pfil_member && ifp != NULL)
3217 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3218 					dir, NULL);
3219 
3220 		if (*mp == NULL || error != 0) /* filter may consume */
3221 			break;
3222 
3223 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3224 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3225 					dir, NULL);
3226 		break;
3227 #endif
3228 	default:
3229 		error = 0;
3230 		break;
3231 	}
3232 
3233 	if (*mp == NULL)
3234 		return (error);
3235 	if (error != 0)
3236 		goto bad;
3237 
3238 	error = -1;
3239 
3240 	/*
3241 	 * Finally, put everything back the way it was and return
3242 	 */
3243 	if (snap) {
3244 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3245 		if (*mp == NULL)
3246 			return (error);
3247 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3248 	}
3249 
3250 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3251 	if (*mp == NULL)
3252 		return (error);
3253 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3254 
3255 	return (0);
3256 
3257 bad:
3258 	m_freem(*mp);
3259 	*mp = NULL;
3260 	return (error);
3261 }
3262 
3263 /*
3264  * Perform basic checks on header size since
3265  * pfil assumes ip_input has already processed
3266  * it for it.  Cut-and-pasted from ip_input.c.
3267  * Given how simple the IPv6 version is,
3268  * does the IPv4 version really need to be
3269  * this complicated?
3270  *
3271  * XXX Should we update ipstat here, or not?
3272  * XXX Right now we update ipstat but not
3273  * XXX csum_counter.
3274  */
3275 static int
3276 bridge_ip_checkbasic(struct mbuf **mp)
3277 {
3278 	struct mbuf *m = *mp;
3279 	struct ip *ip;
3280 	int len, hlen;
3281 	u_short sum;
3282 
3283 	if (*mp == NULL)
3284 		return (-1);
3285 
3286 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3287 		if ((m = m_copyup(m, sizeof(struct ip),
3288 			(max_linkhdr + 3) & ~3)) == NULL) {
3289 			/* XXXJRT new stat, please */
3290 			KMOD_IPSTAT_INC(ips_toosmall);
3291 			goto bad;
3292 		}
3293 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3294 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3295 			KMOD_IPSTAT_INC(ips_toosmall);
3296 			goto bad;
3297 		}
3298 	}
3299 	ip = mtod(m, struct ip *);
3300 	if (ip == NULL) goto bad;
3301 
3302 	if (ip->ip_v != IPVERSION) {
3303 		KMOD_IPSTAT_INC(ips_badvers);
3304 		goto bad;
3305 	}
3306 	hlen = ip->ip_hl << 2;
3307 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3308 		KMOD_IPSTAT_INC(ips_badhlen);
3309 		goto bad;
3310 	}
3311 	if (hlen > m->m_len) {
3312 		if ((m = m_pullup(m, hlen)) == 0) {
3313 			KMOD_IPSTAT_INC(ips_badhlen);
3314 			goto bad;
3315 		}
3316 		ip = mtod(m, struct ip *);
3317 		if (ip == NULL) goto bad;
3318 	}
3319 
3320 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3321 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3322 	} else {
3323 		if (hlen == sizeof(struct ip)) {
3324 			sum = in_cksum_hdr(ip);
3325 		} else {
3326 			sum = in_cksum(m, hlen);
3327 		}
3328 	}
3329 	if (sum) {
3330 		KMOD_IPSTAT_INC(ips_badsum);
3331 		goto bad;
3332 	}
3333 
3334 	/* Retrieve the packet length. */
3335 	len = ntohs(ip->ip_len);
3336 
3337 	/*
3338 	 * Check for additional length bogosity
3339 	 */
3340 	if (len < hlen) {
3341 		KMOD_IPSTAT_INC(ips_badlen);
3342 		goto bad;
3343 	}
3344 
3345 	/*
3346 	 * Check that the amount of data in the buffers
3347 	 * is as at least much as the IP header would have us expect.
3348 	 * Drop packet if shorter than we expect.
3349 	 */
3350 	if (m->m_pkthdr.len < len) {
3351 		KMOD_IPSTAT_INC(ips_tooshort);
3352 		goto bad;
3353 	}
3354 
3355 	/* Checks out, proceed */
3356 	*mp = m;
3357 	return (0);
3358 
3359 bad:
3360 	*mp = m;
3361 	return (-1);
3362 }
3363 
3364 #ifdef INET6
3365 /*
3366  * Same as above, but for IPv6.
3367  * Cut-and-pasted from ip6_input.c.
3368  * XXX Should we update ip6stat, or not?
3369  */
3370 static int
3371 bridge_ip6_checkbasic(struct mbuf **mp)
3372 {
3373 	struct mbuf *m = *mp;
3374 	struct ip6_hdr *ip6;
3375 
3376 	/*
3377 	 * If the IPv6 header is not aligned, slurp it up into a new
3378 	 * mbuf with space for link headers, in the event we forward
3379 	 * it.  Otherwise, if it is aligned, make sure the entire base
3380 	 * IPv6 header is in the first mbuf of the chain.
3381 	 */
3382 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3383 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3384 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3385 			    (max_linkhdr + 3) & ~3)) == NULL) {
3386 			/* XXXJRT new stat, please */
3387 			IP6STAT_INC(ip6s_toosmall);
3388 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3389 			goto bad;
3390 		}
3391 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3392 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3393 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3394 			IP6STAT_INC(ip6s_toosmall);
3395 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3396 			goto bad;
3397 		}
3398 	}
3399 
3400 	ip6 = mtod(m, struct ip6_hdr *);
3401 
3402 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3403 		IP6STAT_INC(ip6s_badvers);
3404 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3405 		goto bad;
3406 	}
3407 
3408 	/* Checks out, proceed */
3409 	*mp = m;
3410 	return (0);
3411 
3412 bad:
3413 	*mp = m;
3414 	return (-1);
3415 }
3416 #endif /* INET6 */
3417 
3418 /*
3419  * bridge_fragment:
3420  *
3421  *	Return a fragmented mbuf chain.
3422  */
3423 static int
3424 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3425     int snap, struct llc *llc)
3426 {
3427 	struct mbuf *m0;
3428 	struct ip *ip;
3429 	int error = -1;
3430 
3431 	if (m->m_len < sizeof(struct ip) &&
3432 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3433 		goto out;
3434 	ip = mtod(m, struct ip *);
3435 
3436 	m->m_pkthdr.csum_flags |= CSUM_IP;
3437 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3438 	if (error)
3439 		goto out;
3440 
3441 	/* walk the chain and re-add the Ethernet header */
3442 	for (m0 = m; m0; m0 = m0->m_nextpkt) {
3443 		if (error == 0) {
3444 			if (snap) {
3445 				M_PREPEND(m0, sizeof(struct llc), M_NOWAIT);
3446 				if (m0 == NULL) {
3447 					error = ENOBUFS;
3448 					continue;
3449 				}
3450 				bcopy(llc, mtod(m0, caddr_t),
3451 				    sizeof(struct llc));
3452 			}
3453 			M_PREPEND(m0, ETHER_HDR_LEN, M_NOWAIT);
3454 			if (m0 == NULL) {
3455 				error = ENOBUFS;
3456 				continue;
3457 			}
3458 			bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3459 		} else
3460 			m_freem(m);
3461 	}
3462 
3463 	if (error == 0)
3464 		KMOD_IPSTAT_INC(ips_fragmented);
3465 
3466 	return (error);
3467 
3468 out:
3469 	if (m != NULL)
3470 		m_freem(m);
3471 	return (error);
3472 }
3473 
3474 static void
3475 bridge_linkstate(struct ifnet *ifp)
3476 {
3477 	struct bridge_softc *sc = ifp->if_bridge;
3478 	struct bridge_iflist *bif;
3479 
3480 	BRIDGE_LOCK(sc);
3481 	bif = bridge_lookup_member_if(sc, ifp);
3482 	if (bif == NULL) {
3483 		BRIDGE_UNLOCK(sc);
3484 		return;
3485 	}
3486 	bridge_linkcheck(sc);
3487 	BRIDGE_UNLOCK(sc);
3488 
3489 	bstp_linkstate(&bif->bif_stp);
3490 }
3491 
3492 static void
3493 bridge_linkcheck(struct bridge_softc *sc)
3494 {
3495 	struct bridge_iflist *bif;
3496 	int new_link, hasls;
3497 
3498 	BRIDGE_LOCK_ASSERT(sc);
3499 	new_link = LINK_STATE_DOWN;
3500 	hasls = 0;
3501 	/* Our link is considered up if at least one of our ports is active */
3502 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3503 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3504 			hasls++;
3505 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3506 			new_link = LINK_STATE_UP;
3507 			break;
3508 		}
3509 	}
3510 	if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3511 		/* If no interfaces support link-state then we default to up */
3512 		new_link = LINK_STATE_UP;
3513 	}
3514 	if_link_state_change(sc->sc_ifp, new_link);
3515 }
3516