xref: /freebsd/sys/net/if_bridge.c (revision 3fc9e2c36555140de248a0b4def91bbfa44d7c2c)
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/mbuf.h>
85 #include <sys/malloc.h>
86 #include <sys/protosw.h>
87 #include <sys/systm.h>
88 #include <sys/jail.h>
89 #include <sys/time.h>
90 #include <sys/socket.h> /* for net/if.h */
91 #include <sys/sockio.h>
92 #include <sys/ctype.h>  /* string functions */
93 #include <sys/kernel.h>
94 #include <sys/random.h>
95 #include <sys/syslog.h>
96 #include <sys/sysctl.h>
97 #include <vm/uma.h>
98 #include <sys/module.h>
99 #include <sys/priv.h>
100 #include <sys/proc.h>
101 #include <sys/lock.h>
102 #include <sys/mutex.h>
103 
104 #include <net/bpf.h>
105 #include <net/if.h>
106 #include <net/if_clone.h>
107 #include <net/if_dl.h>
108 #include <net/if_types.h>
109 #include <net/if_var.h>
110 #include <net/pfil.h>
111 #include <net/vnet.h>
112 
113 #include <netinet/in.h> /* for struct arpcom */
114 #include <netinet/in_systm.h>
115 #include <netinet/in_var.h>
116 #include <netinet/ip.h>
117 #include <netinet/ip_var.h>
118 #ifdef INET6
119 #include <netinet/ip6.h>
120 #include <netinet6/ip6_var.h>
121 #include <netinet6/in6_ifattach.h>
122 #endif
123 #if defined(INET) || defined(INET6)
124 #include <netinet/ip_carp.h>
125 #endif
126 #include <machine/in_cksum.h>
127 #include <netinet/if_ether.h> /* for struct arpcom */
128 #include <net/bridgestp.h>
129 #include <net/if_bridgevar.h>
130 #include <net/if_llc.h>
131 #include <net/if_vlan_var.h>
132 
133 #include <net/route.h>
134 
135 /*
136  * Size of the route hash table.  Must be a power of two.
137  */
138 #ifndef BRIDGE_RTHASH_SIZE
139 #define	BRIDGE_RTHASH_SIZE		1024
140 #endif
141 
142 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
143 
144 /*
145  * Default maximum number of addresses to cache.
146  */
147 #ifndef BRIDGE_RTABLE_MAX
148 #define	BRIDGE_RTABLE_MAX		2000
149 #endif
150 
151 /*
152  * Timeout (in seconds) for entries learned dynamically.
153  */
154 #ifndef BRIDGE_RTABLE_TIMEOUT
155 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
156 #endif
157 
158 /*
159  * Number of seconds between walks of the route list.
160  */
161 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
162 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
163 #endif
164 
165 /*
166  * List of capabilities to possibly mask on the member interface.
167  */
168 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
169 
170 /*
171  * List of capabilities to strip
172  */
173 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
174 
175 /*
176  * Bridge interface list entry.
177  */
178 struct bridge_iflist {
179 	LIST_ENTRY(bridge_iflist) bif_next;
180 	struct ifnet		*bif_ifp;	/* member if */
181 	struct bstp_port	bif_stp;	/* STP state */
182 	uint32_t		bif_flags;	/* member if flags */
183 	int			bif_savedcaps;	/* saved capabilities */
184 	uint32_t		bif_addrmax;	/* max # of addresses */
185 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
186 	uint32_t		bif_addrexceeded;/* # of address violations */
187 };
188 
189 /*
190  * Bridge route node.
191  */
192 struct bridge_rtnode {
193 	LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
194 	LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
195 	struct bridge_iflist	*brt_dst;	/* destination if */
196 	unsigned long		brt_expire;	/* expiration time */
197 	uint8_t			brt_flags;	/* address flags */
198 	uint8_t			brt_addr[ETHER_ADDR_LEN];
199 	uint16_t		brt_vlan;	/* vlan id */
200 };
201 #define	brt_ifp			brt_dst->bif_ifp
202 
203 /*
204  * Software state for each bridge.
205  */
206 struct bridge_softc {
207 	struct ifnet		*sc_ifp;	/* make this an interface */
208 	LIST_ENTRY(bridge_softc) sc_list;
209 	struct mtx		sc_mtx;
210 	struct cv		sc_cv;
211 	uint32_t		sc_brtmax;	/* max # of addresses */
212 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
213 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
214 	struct callout		sc_brcallout;	/* bridge callout */
215 	uint32_t		sc_iflist_ref;	/* refcount for sc_iflist */
216 	uint32_t		sc_iflist_xcnt;	/* refcount for sc_iflist */
217 	LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
218 	LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
219 	LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
220 	uint32_t		sc_rthash_key;	/* key for hash */
221 	LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
222 	struct bstp_state	sc_stp;		/* STP state */
223 	uint32_t		sc_brtexceeded;	/* # of cache drops */
224 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
225 	u_char			sc_defaddr[6];	/* Default MAC address */
226 };
227 
228 static struct mtx 	bridge_list_mtx;
229 eventhandler_tag	bridge_detach_cookie = NULL;
230 
231 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
232 
233 uma_zone_t bridge_rtnode_zone;
234 
235 static int	bridge_clone_create(struct if_clone *, int, caddr_t);
236 static void	bridge_clone_destroy(struct ifnet *);
237 
238 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
239 static void	bridge_mutecaps(struct bridge_softc *);
240 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
241 		    int);
242 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
243 static void	bridge_init(void *);
244 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
245 static void	bridge_stop(struct ifnet *, int);
246 static int	bridge_transmit(struct ifnet *, struct mbuf *);
247 static void	bridge_qflush(struct ifnet *);
248 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
249 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
250 		    struct rtentry *);
251 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
252 		    struct mbuf *);
253 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
254 
255 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
256 		    struct mbuf *m);
257 
258 static void	bridge_timer(void *);
259 
260 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
261 		    struct mbuf *, int);
262 static void	bridge_span(struct bridge_softc *, struct mbuf *);
263 
264 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
265 		    uint16_t, struct bridge_iflist *, int, uint8_t);
266 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
267 		    uint16_t);
268 static void	bridge_rttrim(struct bridge_softc *);
269 static void	bridge_rtage(struct bridge_softc *);
270 static void	bridge_rtflush(struct bridge_softc *, int);
271 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
272 		    uint16_t);
273 
274 static void	bridge_rtable_init(struct bridge_softc *);
275 static void	bridge_rtable_fini(struct bridge_softc *);
276 
277 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
278 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
279 		    const uint8_t *, uint16_t);
280 static int	bridge_rtnode_insert(struct bridge_softc *,
281 		    struct bridge_rtnode *);
282 static void	bridge_rtnode_destroy(struct bridge_softc *,
283 		    struct bridge_rtnode *);
284 static void	bridge_rtable_expire(struct ifnet *, int);
285 static void	bridge_state_change(struct ifnet *, int);
286 
287 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
288 		    const char *name);
289 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
290 		    struct ifnet *ifp);
291 static void	bridge_delete_member(struct bridge_softc *,
292 		    struct bridge_iflist *, int);
293 static void	bridge_delete_span(struct bridge_softc *,
294 		    struct bridge_iflist *);
295 
296 static int	bridge_ioctl_add(struct bridge_softc *, void *);
297 static int	bridge_ioctl_del(struct bridge_softc *, void *);
298 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
299 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
300 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
301 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
302 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
303 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
304 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
305 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
306 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
307 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
308 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
309 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
310 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
311 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
312 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
313 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
314 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
315 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
316 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
317 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
318 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
319 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
320 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
321 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
322 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
323 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
324 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
325 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
326 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
327 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
328 		    int);
329 static int	bridge_ip_checkbasic(struct mbuf **mp);
330 #ifdef INET6
331 static int	bridge_ip6_checkbasic(struct mbuf **mp);
332 #endif /* INET6 */
333 static int	bridge_fragment(struct ifnet *, struct mbuf *,
334 		    struct ether_header *, int, struct llc *);
335 static void	bridge_linkstate(struct ifnet *ifp);
336 static void	bridge_linkcheck(struct bridge_softc *sc);
337 
338 extern void (*bridge_linkstate_p)(struct ifnet *ifp);
339 
340 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
341 #define	VLANTAGOF(_m)	\
342     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
343 
344 static struct bstp_cb_ops bridge_ops = {
345 	.bcb_state = bridge_state_change,
346 	.bcb_rtage = bridge_rtable_expire
347 };
348 
349 SYSCTL_DECL(_net_link);
350 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
351 
352 static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */
353 static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */
354 static int pfil_member = 1; /* run pfil hooks on the member interface */
355 static int pfil_ipfw = 0;   /* layer2 filter with ipfw */
356 static int pfil_ipfw_arp = 0;   /* layer2 filter with ipfw */
357 static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for
358                                    locally destined packets */
359 static int log_stp   = 0;   /* log STP state changes */
360 static int bridge_inherit_mac = 0;   /* share MAC with first bridge member */
361 TUNABLE_INT("net.link.bridge.pfil_onlyip", &pfil_onlyip);
362 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
363     &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
364 TUNABLE_INT("net.link.bridge.ipfw_arp", &pfil_ipfw_arp);
365 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
366     &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
367 TUNABLE_INT("net.link.bridge.pfil_bridge", &pfil_bridge);
368 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
369     &pfil_bridge, 0, "Packet filter on the bridge interface");
370 TUNABLE_INT("net.link.bridge.pfil_member", &pfil_member);
371 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
372     &pfil_member, 0, "Packet filter on the member interface");
373 TUNABLE_INT("net.link.bridge.pfil_local_phys", &pfil_local_phys);
374 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
375     &pfil_local_phys, 0,
376     "Packet filter on the physical interface for locally destined packets");
377 TUNABLE_INT("net.link.bridge.log_stp", &log_stp);
378 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
379     &log_stp, 0, "Log STP state changes");
380 TUNABLE_INT("net.link.bridge.inherit_mac", &bridge_inherit_mac);
381 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, CTLFLAG_RW,
382     &bridge_inherit_mac, 0,
383     "Inherit MAC address from the first bridge member");
384 
385 static VNET_DEFINE(int, allow_llz_overlap) = 0;
386 #define	V_allow_llz_overlap	VNET(allow_llz_overlap)
387 SYSCTL_VNET_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap, CTLFLAG_RW,
388     &VNET_NAME(allow_llz_overlap), 0, "Allow overlap of link-local scope "
389     "zones of a bridge interface and the member interfaces");
390 
391 struct bridge_control {
392 	int	(*bc_func)(struct bridge_softc *, void *);
393 	int	bc_argsize;
394 	int	bc_flags;
395 };
396 
397 #define	BC_F_COPYIN		0x01	/* copy arguments in */
398 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
399 #define	BC_F_SUSER		0x04	/* do super-user check */
400 
401 const struct bridge_control bridge_control_table[] = {
402 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
403 	  BC_F_COPYIN|BC_F_SUSER },
404 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
405 	  BC_F_COPYIN|BC_F_SUSER },
406 
407 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
408 	  BC_F_COPYIN|BC_F_COPYOUT },
409 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
410 	  BC_F_COPYIN|BC_F_SUSER },
411 
412 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
413 	  BC_F_COPYIN|BC_F_SUSER },
414 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
415 	  BC_F_COPYOUT },
416 
417 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
418 	  BC_F_COPYIN|BC_F_COPYOUT },
419 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
420 	  BC_F_COPYIN|BC_F_COPYOUT },
421 
422 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
423 	  BC_F_COPYIN|BC_F_SUSER },
424 
425 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
426 	  BC_F_COPYIN|BC_F_SUSER },
427 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
428 	  BC_F_COPYOUT },
429 
430 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
431 	  BC_F_COPYIN|BC_F_SUSER },
432 
433 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
434 	  BC_F_COPYIN|BC_F_SUSER },
435 
436 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
437 	  BC_F_COPYOUT },
438 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
439 	  BC_F_COPYIN|BC_F_SUSER },
440 
441 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
442 	  BC_F_COPYOUT },
443 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
444 	  BC_F_COPYIN|BC_F_SUSER },
445 
446 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
447 	  BC_F_COPYOUT },
448 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
449 	  BC_F_COPYIN|BC_F_SUSER },
450 
451 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
452 	  BC_F_COPYOUT },
453 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
454 	  BC_F_COPYIN|BC_F_SUSER },
455 
456 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
457 	  BC_F_COPYIN|BC_F_SUSER },
458 
459 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
460 	  BC_F_COPYIN|BC_F_SUSER },
461 
462 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
463 	  BC_F_COPYIN|BC_F_SUSER },
464 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
465 	  BC_F_COPYIN|BC_F_SUSER },
466 
467 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
468 	  BC_F_COPYOUT },
469 
470 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
471 	  BC_F_COPYOUT },
472 
473 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
474 	  BC_F_COPYIN|BC_F_COPYOUT },
475 
476 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
477 	  BC_F_COPYIN|BC_F_SUSER },
478 
479 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
480 	  BC_F_COPYIN|BC_F_SUSER },
481 
482 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
483 	  BC_F_COPYIN|BC_F_SUSER },
484 
485 };
486 const int bridge_control_table_size =
487     sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
488 
489 LIST_HEAD(, bridge_softc) bridge_list;
490 
491 static struct if_clone *bridge_cloner;
492 static const char bridge_name[] = "bridge";
493 
494 static int
495 bridge_modevent(module_t mod, int type, void *data)
496 {
497 
498 	switch (type) {
499 	case MOD_LOAD:
500 		mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
501 		bridge_cloner = if_clone_simple(bridge_name,
502 		    bridge_clone_create, bridge_clone_destroy, 0);
503 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
504 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
505 		    UMA_ALIGN_PTR, 0);
506 		LIST_INIT(&bridge_list);
507 		bridge_input_p = bridge_input;
508 		bridge_output_p = bridge_output;
509 		bridge_dn_p = bridge_dummynet;
510 		bridge_linkstate_p = bridge_linkstate;
511 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
512 		    ifnet_departure_event, bridge_ifdetach, NULL,
513 		    EVENTHANDLER_PRI_ANY);
514 		break;
515 	case MOD_UNLOAD:
516 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
517 		    bridge_detach_cookie);
518 		if_clone_detach(bridge_cloner);
519 		uma_zdestroy(bridge_rtnode_zone);
520 		bridge_input_p = NULL;
521 		bridge_output_p = NULL;
522 		bridge_dn_p = NULL;
523 		bridge_linkstate_p = NULL;
524 		mtx_destroy(&bridge_list_mtx);
525 		break;
526 	default:
527 		return (EOPNOTSUPP);
528 	}
529 	return (0);
530 }
531 
532 static moduledata_t bridge_mod = {
533 	"if_bridge",
534 	bridge_modevent,
535 	0
536 };
537 
538 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
539 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
540 
541 /*
542  * handler for net.link.bridge.ipfw
543  */
544 static int
545 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
546 {
547 	int enable = pfil_ipfw;
548 	int error;
549 
550 	error = sysctl_handle_int(oidp, &enable, 0, req);
551 	enable = (enable) ? 1 : 0;
552 
553 	if (enable != pfil_ipfw) {
554 		pfil_ipfw = enable;
555 
556 		/*
557 		 * Disable pfil so that ipfw doesnt run twice, if the user
558 		 * really wants both then they can re-enable pfil_bridge and/or
559 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
560 		 * layer2 type.
561 		 */
562 		if (pfil_ipfw) {
563 			pfil_onlyip = 0;
564 			pfil_bridge = 0;
565 			pfil_member = 0;
566 		}
567 	}
568 
569 	return (error);
570 }
571 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
572 	    &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
573 
574 /*
575  * bridge_clone_create:
576  *
577  *	Create a new bridge instance.
578  */
579 static int
580 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
581 {
582 	struct bridge_softc *sc, *sc2;
583 	struct ifnet *bifp, *ifp;
584 	int fb, retry;
585 	unsigned long hostid;
586 
587 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
588 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
589 	if (ifp == NULL) {
590 		free(sc, M_DEVBUF);
591 		return (ENOSPC);
592 	}
593 
594 	BRIDGE_LOCK_INIT(sc);
595 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
596 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
597 
598 	/* Initialize our routing table. */
599 	bridge_rtable_init(sc);
600 
601 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
602 
603 	LIST_INIT(&sc->sc_iflist);
604 	LIST_INIT(&sc->sc_spanlist);
605 
606 	ifp->if_softc = sc;
607 	if_initname(ifp, bridge_name, unit);
608 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
609 	ifp->if_ioctl = bridge_ioctl;
610 	ifp->if_transmit = bridge_transmit;
611 	ifp->if_qflush = bridge_qflush;
612 	ifp->if_init = bridge_init;
613 	ifp->if_type = IFT_BRIDGE;
614 
615 	/*
616 	 * Generate an ethernet address with a locally administered address.
617 	 *
618 	 * Since we are using random ethernet addresses for the bridge, it is
619 	 * possible that we might have address collisions, so make sure that
620 	 * this hardware address isn't already in use on another bridge.
621 	 * The first try uses the hostid and falls back to arc4rand().
622 	 */
623 	fb = 0;
624 	getcredhostid(curthread->td_ucred, &hostid);
625 	do {
626 		if (fb || hostid == 0) {
627 			arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
628 			sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
629 			sc->sc_defaddr[0] |= 2;	/* set the LAA bit */
630 		} else {
631 			sc->sc_defaddr[0] = 0x2;
632 			sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
633 			sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
634 			sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
635 			sc->sc_defaddr[4] =  hostid        & 0xff;
636 			sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
637 		}
638 
639 		fb = 1;
640 		retry = 0;
641 		mtx_lock(&bridge_list_mtx);
642 		LIST_FOREACH(sc2, &bridge_list, sc_list) {
643 			bifp = sc2->sc_ifp;
644 			if (memcmp(sc->sc_defaddr,
645 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
646 				retry = 1;
647 				break;
648 			}
649 		}
650 		mtx_unlock(&bridge_list_mtx);
651 	} while (retry == 1);
652 
653 	bstp_attach(&sc->sc_stp, &bridge_ops);
654 	ether_ifattach(ifp, sc->sc_defaddr);
655 	/* Now undo some of the damage... */
656 	ifp->if_baudrate = 0;
657 	ifp->if_type = IFT_BRIDGE;
658 
659 	mtx_lock(&bridge_list_mtx);
660 	LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
661 	mtx_unlock(&bridge_list_mtx);
662 
663 	return (0);
664 }
665 
666 /*
667  * bridge_clone_destroy:
668  *
669  *	Destroy a bridge instance.
670  */
671 static void
672 bridge_clone_destroy(struct ifnet *ifp)
673 {
674 	struct bridge_softc *sc = ifp->if_softc;
675 	struct bridge_iflist *bif;
676 
677 	BRIDGE_LOCK(sc);
678 
679 	bridge_stop(ifp, 1);
680 	ifp->if_flags &= ~IFF_UP;
681 
682 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
683 		bridge_delete_member(sc, bif, 0);
684 
685 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
686 		bridge_delete_span(sc, bif);
687 	}
688 
689 	BRIDGE_UNLOCK(sc);
690 
691 	callout_drain(&sc->sc_brcallout);
692 
693 	mtx_lock(&bridge_list_mtx);
694 	LIST_REMOVE(sc, sc_list);
695 	mtx_unlock(&bridge_list_mtx);
696 
697 	bstp_detach(&sc->sc_stp);
698 	ether_ifdetach(ifp);
699 	if_free(ifp);
700 
701 	/* Tear down the routing table. */
702 	bridge_rtable_fini(sc);
703 
704 	BRIDGE_LOCK_DESTROY(sc);
705 	free(sc, M_DEVBUF);
706 }
707 
708 /*
709  * bridge_ioctl:
710  *
711  *	Handle a control request from the operator.
712  */
713 static int
714 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
715 {
716 	struct bridge_softc *sc = ifp->if_softc;
717 	struct ifreq *ifr = (struct ifreq *)data;
718 	struct bridge_iflist *bif;
719 	struct thread *td = curthread;
720 	union {
721 		struct ifbreq ifbreq;
722 		struct ifbifconf ifbifconf;
723 		struct ifbareq ifbareq;
724 		struct ifbaconf ifbaconf;
725 		struct ifbrparam ifbrparam;
726 		struct ifbropreq ifbropreq;
727 	} args;
728 	struct ifdrv *ifd = (struct ifdrv *) data;
729 	const struct bridge_control *bc;
730 	int error = 0;
731 
732 	switch (cmd) {
733 
734 	case SIOCADDMULTI:
735 	case SIOCDELMULTI:
736 		break;
737 
738 	case SIOCGDRVSPEC:
739 	case SIOCSDRVSPEC:
740 		if (ifd->ifd_cmd >= bridge_control_table_size) {
741 			error = EINVAL;
742 			break;
743 		}
744 		bc = &bridge_control_table[ifd->ifd_cmd];
745 
746 		if (cmd == SIOCGDRVSPEC &&
747 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
748 			error = EINVAL;
749 			break;
750 		}
751 		else if (cmd == SIOCSDRVSPEC &&
752 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
753 			error = EINVAL;
754 			break;
755 		}
756 
757 		if (bc->bc_flags & BC_F_SUSER) {
758 			error = priv_check(td, PRIV_NET_BRIDGE);
759 			if (error)
760 				break;
761 		}
762 
763 		if (ifd->ifd_len != bc->bc_argsize ||
764 		    ifd->ifd_len > sizeof(args)) {
765 			error = EINVAL;
766 			break;
767 		}
768 
769 		bzero(&args, sizeof(args));
770 		if (bc->bc_flags & BC_F_COPYIN) {
771 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
772 			if (error)
773 				break;
774 		}
775 
776 		BRIDGE_LOCK(sc);
777 		error = (*bc->bc_func)(sc, &args);
778 		BRIDGE_UNLOCK(sc);
779 		if (error)
780 			break;
781 
782 		if (bc->bc_flags & BC_F_COPYOUT)
783 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
784 
785 		break;
786 
787 	case SIOCSIFFLAGS:
788 		if (!(ifp->if_flags & IFF_UP) &&
789 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
790 			/*
791 			 * If interface is marked down and it is running,
792 			 * then stop and disable it.
793 			 */
794 			BRIDGE_LOCK(sc);
795 			bridge_stop(ifp, 1);
796 			BRIDGE_UNLOCK(sc);
797 		} else if ((ifp->if_flags & IFF_UP) &&
798 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
799 			/*
800 			 * If interface is marked up and it is stopped, then
801 			 * start it.
802 			 */
803 			(*ifp->if_init)(sc);
804 		}
805 		break;
806 
807 	case SIOCSIFMTU:
808 		if (ifr->ifr_mtu < 576) {
809 			error = EINVAL;
810 			break;
811 		}
812 		if (LIST_EMPTY(&sc->sc_iflist)) {
813 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
814 			break;
815 		}
816 		BRIDGE_LOCK(sc);
817 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
818 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
819 				log(LOG_NOTICE, "%s: invalid MTU: %lu(%s)"
820 				    " != %d\n", sc->sc_ifp->if_xname,
821 				    bif->bif_ifp->if_mtu,
822 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
823 				error = EINVAL;
824 				break;
825 			}
826 		}
827 		if (!error)
828 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
829 		BRIDGE_UNLOCK(sc);
830 		break;
831 	default:
832 		/*
833 		 * drop the lock as ether_ioctl() will call bridge_start() and
834 		 * cause the lock to be recursed.
835 		 */
836 		error = ether_ioctl(ifp, cmd, data);
837 		break;
838 	}
839 
840 	return (error);
841 }
842 
843 /*
844  * bridge_mutecaps:
845  *
846  *	Clear or restore unwanted capabilities on the member interface
847  */
848 static void
849 bridge_mutecaps(struct bridge_softc *sc)
850 {
851 	struct bridge_iflist *bif;
852 	int enabled, mask;
853 
854 	/* Initial bitmask of capabilities to test */
855 	mask = BRIDGE_IFCAPS_MASK;
856 
857 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
858 		/* Every member must support it or its disabled */
859 		mask &= bif->bif_savedcaps;
860 	}
861 
862 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
863 		enabled = bif->bif_ifp->if_capenable;
864 		enabled &= ~BRIDGE_IFCAPS_STRIP;
865 		/* strip off mask bits and enable them again if allowed */
866 		enabled &= ~BRIDGE_IFCAPS_MASK;
867 		enabled |= mask;
868 		bridge_set_ifcap(sc, bif, enabled);
869 	}
870 
871 }
872 
873 static void
874 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
875 {
876 	struct ifnet *ifp = bif->bif_ifp;
877 	struct ifreq ifr;
878 	int error;
879 
880 	bzero(&ifr, sizeof(ifr));
881 	ifr.ifr_reqcap = set;
882 
883 	if (ifp->if_capenable != set) {
884 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
885 		if (error)
886 			if_printf(sc->sc_ifp,
887 			    "error setting interface capabilities on %s\n",
888 			    ifp->if_xname);
889 	}
890 }
891 
892 /*
893  * bridge_lookup_member:
894  *
895  *	Lookup a bridge member interface.
896  */
897 static struct bridge_iflist *
898 bridge_lookup_member(struct bridge_softc *sc, const char *name)
899 {
900 	struct bridge_iflist *bif;
901 	struct ifnet *ifp;
902 
903 	BRIDGE_LOCK_ASSERT(sc);
904 
905 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
906 		ifp = bif->bif_ifp;
907 		if (strcmp(ifp->if_xname, name) == 0)
908 			return (bif);
909 	}
910 
911 	return (NULL);
912 }
913 
914 /*
915  * bridge_lookup_member_if:
916  *
917  *	Lookup a bridge member interface by ifnet*.
918  */
919 static struct bridge_iflist *
920 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
921 {
922 	struct bridge_iflist *bif;
923 
924 	BRIDGE_LOCK_ASSERT(sc);
925 
926 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
927 		if (bif->bif_ifp == member_ifp)
928 			return (bif);
929 	}
930 
931 	return (NULL);
932 }
933 
934 /*
935  * bridge_delete_member:
936  *
937  *	Delete the specified member interface.
938  */
939 static void
940 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
941     int gone)
942 {
943 	struct ifnet *ifs = bif->bif_ifp;
944 	struct ifnet *fif = NULL;
945 
946 	BRIDGE_LOCK_ASSERT(sc);
947 
948 	if (bif->bif_flags & IFBIF_STP)
949 		bstp_disable(&bif->bif_stp);
950 
951 	ifs->if_bridge = NULL;
952 	BRIDGE_XLOCK(sc);
953 	LIST_REMOVE(bif, bif_next);
954 	BRIDGE_XDROP(sc);
955 
956 	/*
957 	 * If removing the interface that gave the bridge its mac address, set
958 	 * the mac address of the bridge to the address of the next member, or
959 	 * to its default address if no members are left.
960 	 */
961 	if (bridge_inherit_mac && sc->sc_ifaddr == ifs) {
962 		if (LIST_EMPTY(&sc->sc_iflist)) {
963 			bcopy(sc->sc_defaddr,
964 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
965 			sc->sc_ifaddr = NULL;
966 		} else {
967 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
968 			bcopy(IF_LLADDR(fif),
969 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
970 			sc->sc_ifaddr = fif;
971 		}
972 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
973 	}
974 
975 	bridge_linkcheck(sc);
976 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
977 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
978 	KASSERT(bif->bif_addrcnt == 0,
979 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
980 
981 	BRIDGE_UNLOCK(sc);
982 	if (!gone) {
983 		switch (ifs->if_type) {
984 		case IFT_ETHER:
985 		case IFT_L2VLAN:
986 			/*
987 			 * Take the interface out of promiscuous mode.
988 			 */
989 			(void) ifpromisc(ifs, 0);
990 			break;
991 
992 		case IFT_GIF:
993 			break;
994 
995 		default:
996 #ifdef DIAGNOSTIC
997 			panic("bridge_delete_member: impossible");
998 #endif
999 			break;
1000 		}
1001 		/* reneable any interface capabilities */
1002 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1003 	}
1004 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1005 	BRIDGE_LOCK(sc);
1006 	free(bif, M_DEVBUF);
1007 }
1008 
1009 /*
1010  * bridge_delete_span:
1011  *
1012  *	Delete the specified span interface.
1013  */
1014 static void
1015 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1016 {
1017 	BRIDGE_LOCK_ASSERT(sc);
1018 
1019 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1020 	    ("%s: not a span interface", __func__));
1021 
1022 	LIST_REMOVE(bif, bif_next);
1023 	free(bif, M_DEVBUF);
1024 }
1025 
1026 static int
1027 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1028 {
1029 	struct ifbreq *req = arg;
1030 	struct bridge_iflist *bif = NULL;
1031 	struct ifnet *ifs;
1032 	int error = 0;
1033 
1034 	ifs = ifunit(req->ifbr_ifsname);
1035 	if (ifs == NULL)
1036 		return (ENOENT);
1037 	if (ifs->if_ioctl == NULL)	/* must be supported */
1038 		return (EINVAL);
1039 
1040 	/* If it's in the span list, it can't be a member. */
1041 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1042 		if (ifs == bif->bif_ifp)
1043 			return (EBUSY);
1044 
1045 	if (ifs->if_bridge == sc)
1046 		return (EEXIST);
1047 
1048 	if (ifs->if_bridge != NULL)
1049 		return (EBUSY);
1050 
1051 	switch (ifs->if_type) {
1052 	case IFT_ETHER:
1053 	case IFT_L2VLAN:
1054 	case IFT_GIF:
1055 		/* permitted interface types */
1056 		break;
1057 	default:
1058 		return (EINVAL);
1059 	}
1060 
1061 #ifdef INET6
1062 	/*
1063 	 * Two valid inet6 addresses with link-local scope must not be
1064 	 * on the parent interface and the member interfaces at the
1065 	 * same time.  This restriction is needed to prevent violation
1066 	 * of link-local scope zone.  Attempts to add a member
1067 	 * interface which has inet6 addresses when the parent has
1068 	 * inet6 triggers removal of all inet6 addresses on the member
1069 	 * interface.
1070 	 */
1071 
1072 	/* Check if the parent interface has a link-local scope addr. */
1073 	if (V_allow_llz_overlap == 0 &&
1074 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1075 		/*
1076 		 * If any, remove all inet6 addresses from the member
1077 		 * interfaces.
1078 		 */
1079 		BRIDGE_XLOCK(sc);
1080 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1081  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1082 				BRIDGE_UNLOCK(sc);
1083 				in6_ifdetach(bif->bif_ifp);
1084 				BRIDGE_LOCK(sc);
1085 				if_printf(sc->sc_ifp,
1086 				    "IPv6 addresses on %s have been removed "
1087 				    "before adding it as a member to prevent "
1088 				    "IPv6 address scope violation.\n",
1089 				    bif->bif_ifp->if_xname);
1090 			}
1091 		}
1092 		BRIDGE_XDROP(sc);
1093 		if (in6ifa_llaonifp(ifs)) {
1094 			BRIDGE_UNLOCK(sc);
1095 			in6_ifdetach(ifs);
1096 			BRIDGE_LOCK(sc);
1097 			if_printf(sc->sc_ifp,
1098 			    "IPv6 addresses on %s have been removed "
1099 			    "before adding it as a member to prevent "
1100 			    "IPv6 address scope violation.\n",
1101 			    ifs->if_xname);
1102 		}
1103 	}
1104 #endif
1105 	/* Allow the first Ethernet member to define the MTU */
1106 	if (LIST_EMPTY(&sc->sc_iflist))
1107 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1108 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1109 		if_printf(sc->sc_ifp, "invalid MTU: %lu(%s) != %lu\n",
1110 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1111 		return (EINVAL);
1112 	}
1113 
1114 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1115 	if (bif == NULL)
1116 		return (ENOMEM);
1117 
1118 	bif->bif_ifp = ifs;
1119 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1120 	bif->bif_savedcaps = ifs->if_capenable;
1121 
1122 	/*
1123 	 * Assign the interface's MAC address to the bridge if it's the first
1124 	 * member and the MAC address of the bridge has not been changed from
1125 	 * the default randomly generated one.
1126 	 */
1127 	if (bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1128 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1129 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1130 		sc->sc_ifaddr = ifs;
1131 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1132 	}
1133 
1134 	ifs->if_bridge = sc;
1135 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1136 	/*
1137 	 * XXX: XLOCK HERE!?!
1138 	 *
1139 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1140 	 */
1141 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1142 
1143 	/* Set interface capabilities to the intersection set of all members */
1144 	bridge_mutecaps(sc);
1145 	bridge_linkcheck(sc);
1146 
1147 	/* Place the interface into promiscuous mode */
1148 	switch (ifs->if_type) {
1149 		case IFT_ETHER:
1150 		case IFT_L2VLAN:
1151 			BRIDGE_UNLOCK(sc);
1152 			error = ifpromisc(ifs, 1);
1153 			BRIDGE_LOCK(sc);
1154 			break;
1155 	}
1156 
1157 	if (error) {
1158 		bridge_delete_member(sc, bif, 0);
1159 		free(bif, M_DEVBUF);
1160 	}
1161 	return (error);
1162 }
1163 
1164 static int
1165 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1166 {
1167 	struct ifbreq *req = arg;
1168 	struct bridge_iflist *bif;
1169 
1170 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1171 	if (bif == NULL)
1172 		return (ENOENT);
1173 
1174 	bridge_delete_member(sc, bif, 0);
1175 
1176 	return (0);
1177 }
1178 
1179 static int
1180 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1181 {
1182 	struct ifbreq *req = arg;
1183 	struct bridge_iflist *bif;
1184 	struct bstp_port *bp;
1185 
1186 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1187 	if (bif == NULL)
1188 		return (ENOENT);
1189 
1190 	bp = &bif->bif_stp;
1191 	req->ifbr_ifsflags = bif->bif_flags;
1192 	req->ifbr_state = bp->bp_state;
1193 	req->ifbr_priority = bp->bp_priority;
1194 	req->ifbr_path_cost = bp->bp_path_cost;
1195 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1196 	req->ifbr_proto = bp->bp_protover;
1197 	req->ifbr_role = bp->bp_role;
1198 	req->ifbr_stpflags = bp->bp_flags;
1199 	req->ifbr_addrcnt = bif->bif_addrcnt;
1200 	req->ifbr_addrmax = bif->bif_addrmax;
1201 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1202 
1203 	/* Copy STP state options as flags */
1204 	if (bp->bp_operedge)
1205 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1206 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1207 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1208 	if (bp->bp_ptp_link)
1209 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1210 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1211 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1212 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1213 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1214 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1215 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1216 	return (0);
1217 }
1218 
1219 static int
1220 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1221 {
1222 	struct ifbreq *req = arg;
1223 	struct bridge_iflist *bif;
1224 	struct bstp_port *bp;
1225 	int error;
1226 
1227 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1228 	if (bif == NULL)
1229 		return (ENOENT);
1230 	bp = &bif->bif_stp;
1231 
1232 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1233 		/* SPAN is readonly */
1234 		return (EINVAL);
1235 
1236 	if (req->ifbr_ifsflags & IFBIF_STP) {
1237 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1238 			error = bstp_enable(&bif->bif_stp);
1239 			if (error)
1240 				return (error);
1241 		}
1242 	} else {
1243 		if ((bif->bif_flags & IFBIF_STP) != 0)
1244 			bstp_disable(&bif->bif_stp);
1245 	}
1246 
1247 	/* Pass on STP flags */
1248 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1249 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1250 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1251 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1252 
1253 	/* Save the bits relating to the bridge */
1254 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1255 
1256 	return (0);
1257 }
1258 
1259 static int
1260 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1261 {
1262 	struct ifbrparam *param = arg;
1263 
1264 	sc->sc_brtmax = param->ifbrp_csize;
1265 	bridge_rttrim(sc);
1266 
1267 	return (0);
1268 }
1269 
1270 static int
1271 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1272 {
1273 	struct ifbrparam *param = arg;
1274 
1275 	param->ifbrp_csize = sc->sc_brtmax;
1276 
1277 	return (0);
1278 }
1279 
1280 static int
1281 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1282 {
1283 	struct ifbifconf *bifc = arg;
1284 	struct bridge_iflist *bif;
1285 	struct ifbreq breq;
1286 	char *buf, *outbuf;
1287 	int count, buflen, len, error = 0;
1288 
1289 	count = 0;
1290 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1291 		count++;
1292 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1293 		count++;
1294 
1295 	buflen = sizeof(breq) * count;
1296 	if (bifc->ifbic_len == 0) {
1297 		bifc->ifbic_len = buflen;
1298 		return (0);
1299 	}
1300 	BRIDGE_UNLOCK(sc);
1301 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1302 	BRIDGE_LOCK(sc);
1303 
1304 	count = 0;
1305 	buf = outbuf;
1306 	len = min(bifc->ifbic_len, buflen);
1307 	bzero(&breq, sizeof(breq));
1308 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1309 		if (len < sizeof(breq))
1310 			break;
1311 
1312 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1313 		    sizeof(breq.ifbr_ifsname));
1314 		/* Fill in the ifbreq structure */
1315 		error = bridge_ioctl_gifflags(sc, &breq);
1316 		if (error)
1317 			break;
1318 		memcpy(buf, &breq, sizeof(breq));
1319 		count++;
1320 		buf += sizeof(breq);
1321 		len -= sizeof(breq);
1322 	}
1323 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1324 		if (len < sizeof(breq))
1325 			break;
1326 
1327 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1328 		    sizeof(breq.ifbr_ifsname));
1329 		breq.ifbr_ifsflags = bif->bif_flags;
1330 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1331 		memcpy(buf, &breq, sizeof(breq));
1332 		count++;
1333 		buf += sizeof(breq);
1334 		len -= sizeof(breq);
1335 	}
1336 
1337 	BRIDGE_UNLOCK(sc);
1338 	bifc->ifbic_len = sizeof(breq) * count;
1339 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1340 	BRIDGE_LOCK(sc);
1341 	free(outbuf, M_TEMP);
1342 	return (error);
1343 }
1344 
1345 static int
1346 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1347 {
1348 	struct ifbaconf *bac = arg;
1349 	struct bridge_rtnode *brt;
1350 	struct ifbareq bareq;
1351 	char *buf, *outbuf;
1352 	int count, buflen, len, error = 0;
1353 
1354 	if (bac->ifbac_len == 0)
1355 		return (0);
1356 
1357 	count = 0;
1358 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1359 		count++;
1360 	buflen = sizeof(bareq) * count;
1361 
1362 	BRIDGE_UNLOCK(sc);
1363 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1364 	BRIDGE_LOCK(sc);
1365 
1366 	count = 0;
1367 	buf = outbuf;
1368 	len = min(bac->ifbac_len, buflen);
1369 	bzero(&bareq, sizeof(bareq));
1370 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1371 		if (len < sizeof(bareq))
1372 			goto out;
1373 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1374 		    sizeof(bareq.ifba_ifsname));
1375 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1376 		bareq.ifba_vlan = brt->brt_vlan;
1377 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1378 				time_uptime < brt->brt_expire)
1379 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1380 		else
1381 			bareq.ifba_expire = 0;
1382 		bareq.ifba_flags = brt->brt_flags;
1383 
1384 		memcpy(buf, &bareq, sizeof(bareq));
1385 		count++;
1386 		buf += sizeof(bareq);
1387 		len -= sizeof(bareq);
1388 	}
1389 out:
1390 	BRIDGE_UNLOCK(sc);
1391 	bac->ifbac_len = sizeof(bareq) * count;
1392 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1393 	BRIDGE_LOCK(sc);
1394 	free(outbuf, M_TEMP);
1395 	return (error);
1396 }
1397 
1398 static int
1399 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1400 {
1401 	struct ifbareq *req = arg;
1402 	struct bridge_iflist *bif;
1403 	int error;
1404 
1405 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1406 	if (bif == NULL)
1407 		return (ENOENT);
1408 
1409 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1410 	    req->ifba_flags);
1411 
1412 	return (error);
1413 }
1414 
1415 static int
1416 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1417 {
1418 	struct ifbrparam *param = arg;
1419 
1420 	sc->sc_brttimeout = param->ifbrp_ctime;
1421 	return (0);
1422 }
1423 
1424 static int
1425 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1426 {
1427 	struct ifbrparam *param = arg;
1428 
1429 	param->ifbrp_ctime = sc->sc_brttimeout;
1430 	return (0);
1431 }
1432 
1433 static int
1434 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1435 {
1436 	struct ifbareq *req = arg;
1437 
1438 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1439 }
1440 
1441 static int
1442 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1443 {
1444 	struct ifbreq *req = arg;
1445 
1446 	bridge_rtflush(sc, req->ifbr_ifsflags);
1447 	return (0);
1448 }
1449 
1450 static int
1451 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1452 {
1453 	struct ifbrparam *param = arg;
1454 	struct bstp_state *bs = &sc->sc_stp;
1455 
1456 	param->ifbrp_prio = bs->bs_bridge_priority;
1457 	return (0);
1458 }
1459 
1460 static int
1461 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1462 {
1463 	struct ifbrparam *param = arg;
1464 
1465 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1466 }
1467 
1468 static int
1469 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1470 {
1471 	struct ifbrparam *param = arg;
1472 	struct bstp_state *bs = &sc->sc_stp;
1473 
1474 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1475 	return (0);
1476 }
1477 
1478 static int
1479 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1480 {
1481 	struct ifbrparam *param = arg;
1482 
1483 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1484 }
1485 
1486 static int
1487 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1488 {
1489 	struct ifbrparam *param = arg;
1490 	struct bstp_state *bs = &sc->sc_stp;
1491 
1492 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1493 	return (0);
1494 }
1495 
1496 static int
1497 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1498 {
1499 	struct ifbrparam *param = arg;
1500 
1501 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1502 }
1503 
1504 static int
1505 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1506 {
1507 	struct ifbrparam *param = arg;
1508 	struct bstp_state *bs = &sc->sc_stp;
1509 
1510 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1511 	return (0);
1512 }
1513 
1514 static int
1515 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1516 {
1517 	struct ifbrparam *param = arg;
1518 
1519 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1520 }
1521 
1522 static int
1523 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1524 {
1525 	struct ifbreq *req = arg;
1526 	struct bridge_iflist *bif;
1527 
1528 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1529 	if (bif == NULL)
1530 		return (ENOENT);
1531 
1532 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1533 }
1534 
1535 static int
1536 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1537 {
1538 	struct ifbreq *req = arg;
1539 	struct bridge_iflist *bif;
1540 
1541 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1542 	if (bif == NULL)
1543 		return (ENOENT);
1544 
1545 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1546 }
1547 
1548 static int
1549 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1550 {
1551 	struct ifbreq *req = arg;
1552 	struct bridge_iflist *bif;
1553 
1554 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1555 	if (bif == NULL)
1556 		return (ENOENT);
1557 
1558 	bif->bif_addrmax = req->ifbr_addrmax;
1559 	return (0);
1560 }
1561 
1562 static int
1563 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1564 {
1565 	struct ifbreq *req = arg;
1566 	struct bridge_iflist *bif = NULL;
1567 	struct ifnet *ifs;
1568 
1569 	ifs = ifunit(req->ifbr_ifsname);
1570 	if (ifs == NULL)
1571 		return (ENOENT);
1572 
1573 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1574 		if (ifs == bif->bif_ifp)
1575 			return (EBUSY);
1576 
1577 	if (ifs->if_bridge != NULL)
1578 		return (EBUSY);
1579 
1580 	switch (ifs->if_type) {
1581 		case IFT_ETHER:
1582 		case IFT_GIF:
1583 		case IFT_L2VLAN:
1584 			break;
1585 		default:
1586 			return (EINVAL);
1587 	}
1588 
1589 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1590 	if (bif == NULL)
1591 		return (ENOMEM);
1592 
1593 	bif->bif_ifp = ifs;
1594 	bif->bif_flags = IFBIF_SPAN;
1595 
1596 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1597 
1598 	return (0);
1599 }
1600 
1601 static int
1602 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1603 {
1604 	struct ifbreq *req = arg;
1605 	struct bridge_iflist *bif;
1606 	struct ifnet *ifs;
1607 
1608 	ifs = ifunit(req->ifbr_ifsname);
1609 	if (ifs == NULL)
1610 		return (ENOENT);
1611 
1612 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1613 		if (ifs == bif->bif_ifp)
1614 			break;
1615 
1616 	if (bif == NULL)
1617 		return (ENOENT);
1618 
1619 	bridge_delete_span(sc, bif);
1620 
1621 	return (0);
1622 }
1623 
1624 static int
1625 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1626 {
1627 	struct ifbropreq *req = arg;
1628 	struct bstp_state *bs = &sc->sc_stp;
1629 	struct bstp_port *root_port;
1630 
1631 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1632 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1633 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1634 
1635 	root_port = bs->bs_root_port;
1636 	if (root_port == NULL)
1637 		req->ifbop_root_port = 0;
1638 	else
1639 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1640 
1641 	req->ifbop_holdcount = bs->bs_txholdcount;
1642 	req->ifbop_priority = bs->bs_bridge_priority;
1643 	req->ifbop_protocol = bs->bs_protover;
1644 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1645 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1646 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1647 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1648 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1649 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1650 
1651 	return (0);
1652 }
1653 
1654 static int
1655 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1656 {
1657 	struct ifbrparam *param = arg;
1658 
1659 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1660 	return (0);
1661 }
1662 
1663 static int
1664 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1665 {
1666 	struct ifbpstpconf *bifstp = arg;
1667 	struct bridge_iflist *bif;
1668 	struct bstp_port *bp;
1669 	struct ifbpstpreq bpreq;
1670 	char *buf, *outbuf;
1671 	int count, buflen, len, error = 0;
1672 
1673 	count = 0;
1674 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1675 		if ((bif->bif_flags & IFBIF_STP) != 0)
1676 			count++;
1677 	}
1678 
1679 	buflen = sizeof(bpreq) * count;
1680 	if (bifstp->ifbpstp_len == 0) {
1681 		bifstp->ifbpstp_len = buflen;
1682 		return (0);
1683 	}
1684 
1685 	BRIDGE_UNLOCK(sc);
1686 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1687 	BRIDGE_LOCK(sc);
1688 
1689 	count = 0;
1690 	buf = outbuf;
1691 	len = min(bifstp->ifbpstp_len, buflen);
1692 	bzero(&bpreq, sizeof(bpreq));
1693 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1694 		if (len < sizeof(bpreq))
1695 			break;
1696 
1697 		if ((bif->bif_flags & IFBIF_STP) == 0)
1698 			continue;
1699 
1700 		bp = &bif->bif_stp;
1701 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1702 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1703 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1704 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1705 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1706 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1707 
1708 		memcpy(buf, &bpreq, sizeof(bpreq));
1709 		count++;
1710 		buf += sizeof(bpreq);
1711 		len -= sizeof(bpreq);
1712 	}
1713 
1714 	BRIDGE_UNLOCK(sc);
1715 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1716 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1717 	BRIDGE_LOCK(sc);
1718 	free(outbuf, M_TEMP);
1719 	return (error);
1720 }
1721 
1722 static int
1723 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1724 {
1725 	struct ifbrparam *param = arg;
1726 
1727 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1728 }
1729 
1730 static int
1731 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1732 {
1733 	struct ifbrparam *param = arg;
1734 
1735 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1736 }
1737 
1738 /*
1739  * bridge_ifdetach:
1740  *
1741  *	Detach an interface from a bridge.  Called when a member
1742  *	interface is detaching.
1743  */
1744 static void
1745 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1746 {
1747 	struct bridge_softc *sc = ifp->if_bridge;
1748 	struct bridge_iflist *bif;
1749 
1750 	if (ifp->if_flags & IFF_RENAMING)
1751 		return;
1752 
1753 	/* Check if the interface is a bridge member */
1754 	if (sc != NULL) {
1755 		BRIDGE_LOCK(sc);
1756 
1757 		bif = bridge_lookup_member_if(sc, ifp);
1758 		if (bif != NULL)
1759 			bridge_delete_member(sc, bif, 1);
1760 
1761 		BRIDGE_UNLOCK(sc);
1762 		return;
1763 	}
1764 
1765 	/* Check if the interface is a span port */
1766 	mtx_lock(&bridge_list_mtx);
1767 	LIST_FOREACH(sc, &bridge_list, sc_list) {
1768 		BRIDGE_LOCK(sc);
1769 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1770 			if (ifp == bif->bif_ifp) {
1771 				bridge_delete_span(sc, bif);
1772 				break;
1773 			}
1774 
1775 		BRIDGE_UNLOCK(sc);
1776 	}
1777 	mtx_unlock(&bridge_list_mtx);
1778 }
1779 
1780 /*
1781  * bridge_init:
1782  *
1783  *	Initialize a bridge interface.
1784  */
1785 static void
1786 bridge_init(void *xsc)
1787 {
1788 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1789 	struct ifnet *ifp = sc->sc_ifp;
1790 
1791 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1792 		return;
1793 
1794 	BRIDGE_LOCK(sc);
1795 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1796 	    bridge_timer, sc);
1797 
1798 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1799 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1800 
1801 	BRIDGE_UNLOCK(sc);
1802 }
1803 
1804 /*
1805  * bridge_stop:
1806  *
1807  *	Stop the bridge interface.
1808  */
1809 static void
1810 bridge_stop(struct ifnet *ifp, int disable)
1811 {
1812 	struct bridge_softc *sc = ifp->if_softc;
1813 
1814 	BRIDGE_LOCK_ASSERT(sc);
1815 
1816 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1817 		return;
1818 
1819 	callout_stop(&sc->sc_brcallout);
1820 	bstp_stop(&sc->sc_stp);
1821 
1822 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1823 
1824 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1825 }
1826 
1827 /*
1828  * bridge_enqueue:
1829  *
1830  *	Enqueue a packet on a bridge member interface.
1831  *
1832  */
1833 static int
1834 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1835 {
1836 	int len, err = 0;
1837 	short mflags;
1838 	struct mbuf *m0;
1839 
1840 	/* We may be sending a fragment so traverse the mbuf */
1841 	for (; m; m = m0) {
1842 		m0 = m->m_nextpkt;
1843 		m->m_nextpkt = NULL;
1844 		len = m->m_pkthdr.len;
1845 		mflags = m->m_flags;
1846 
1847 		/*
1848 		 * If underlying interface can not do VLAN tag insertion itself
1849 		 * then attach a packet tag that holds it.
1850 		 */
1851 		if ((m->m_flags & M_VLANTAG) &&
1852 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1853 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1854 			if (m == NULL) {
1855 				if_printf(dst_ifp,
1856 				    "unable to prepend VLAN header\n");
1857 				dst_ifp->if_oerrors++;
1858 				continue;
1859 			}
1860 			m->m_flags &= ~M_VLANTAG;
1861 		}
1862 
1863 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1864 			m_freem(m0);
1865 			sc->sc_ifp->if_oerrors++;
1866 			break;
1867 		}
1868 
1869 		sc->sc_ifp->if_opackets++;
1870 		sc->sc_ifp->if_obytes += len;
1871 		if (mflags & M_MCAST)
1872 			sc->sc_ifp->if_omcasts++;
1873 	}
1874 
1875 	return (err);
1876 }
1877 
1878 /*
1879  * bridge_dummynet:
1880  *
1881  * 	Receive a queued packet from dummynet and pass it on to the output
1882  * 	interface.
1883  *
1884  *	The mbuf has the Ethernet header already attached.
1885  */
1886 static void
1887 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1888 {
1889 	struct bridge_softc *sc;
1890 
1891 	sc = ifp->if_bridge;
1892 
1893 	/*
1894 	 * The packet didnt originate from a member interface. This should only
1895 	 * ever happen if a member interface is removed while packets are
1896 	 * queued for it.
1897 	 */
1898 	if (sc == NULL) {
1899 		m_freem(m);
1900 		return;
1901 	}
1902 
1903 	if (PFIL_HOOKED(&V_inet_pfil_hook)
1904 #ifdef INET6
1905 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1906 #endif
1907 	    ) {
1908 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1909 			return;
1910 		if (m == NULL)
1911 			return;
1912 	}
1913 
1914 	bridge_enqueue(sc, ifp, m);
1915 }
1916 
1917 /*
1918  * bridge_output:
1919  *
1920  *	Send output from a bridge member interface.  This
1921  *	performs the bridging function for locally originated
1922  *	packets.
1923  *
1924  *	The mbuf has the Ethernet header already attached.  We must
1925  *	enqueue or free the mbuf before returning.
1926  */
1927 static int
1928 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1929     struct rtentry *rt)
1930 {
1931 	struct ether_header *eh;
1932 	struct ifnet *dst_if;
1933 	struct bridge_softc *sc;
1934 	uint16_t vlan;
1935 
1936 	if (m->m_len < ETHER_HDR_LEN) {
1937 		m = m_pullup(m, ETHER_HDR_LEN);
1938 		if (m == NULL)
1939 			return (0);
1940 	}
1941 
1942 	eh = mtod(m, struct ether_header *);
1943 	sc = ifp->if_bridge;
1944 	vlan = VLANTAGOF(m);
1945 
1946 	BRIDGE_LOCK(sc);
1947 
1948 	/*
1949 	 * If bridge is down, but the original output interface is up,
1950 	 * go ahead and send out that interface.  Otherwise, the packet
1951 	 * is dropped below.
1952 	 */
1953 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1954 		dst_if = ifp;
1955 		goto sendunicast;
1956 	}
1957 
1958 	/*
1959 	 * If the packet is a multicast, or we don't know a better way to
1960 	 * get there, send to all interfaces.
1961 	 */
1962 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
1963 		dst_if = NULL;
1964 	else
1965 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1966 	if (dst_if == NULL) {
1967 		struct bridge_iflist *bif;
1968 		struct mbuf *mc;
1969 		int error = 0, used = 0;
1970 
1971 		bridge_span(sc, m);
1972 
1973 		BRIDGE_LOCK2REF(sc, error);
1974 		if (error) {
1975 			m_freem(m);
1976 			return (0);
1977 		}
1978 
1979 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1980 			dst_if = bif->bif_ifp;
1981 
1982 			if (dst_if->if_type == IFT_GIF)
1983 				continue;
1984 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1985 				continue;
1986 
1987 			/*
1988 			 * If this is not the original output interface,
1989 			 * and the interface is participating in spanning
1990 			 * tree, make sure the port is in a state that
1991 			 * allows forwarding.
1992 			 */
1993 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
1994 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1995 				continue;
1996 
1997 			if (LIST_NEXT(bif, bif_next) == NULL) {
1998 				used = 1;
1999 				mc = m;
2000 			} else {
2001 				mc = m_copypacket(m, M_NOWAIT);
2002 				if (mc == NULL) {
2003 					sc->sc_ifp->if_oerrors++;
2004 					continue;
2005 				}
2006 			}
2007 
2008 			bridge_enqueue(sc, dst_if, mc);
2009 		}
2010 		if (used == 0)
2011 			m_freem(m);
2012 		BRIDGE_UNREF(sc);
2013 		return (0);
2014 	}
2015 
2016 sendunicast:
2017 	/*
2018 	 * XXX Spanning tree consideration here?
2019 	 */
2020 
2021 	bridge_span(sc, m);
2022 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2023 		m_freem(m);
2024 		BRIDGE_UNLOCK(sc);
2025 		return (0);
2026 	}
2027 
2028 	BRIDGE_UNLOCK(sc);
2029 	bridge_enqueue(sc, dst_if, m);
2030 	return (0);
2031 }
2032 
2033 /*
2034  * bridge_transmit:
2035  *
2036  *	Do output on a bridge.
2037  *
2038  */
2039 static int
2040 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2041 {
2042 	struct bridge_softc *sc;
2043 	struct ether_header *eh;
2044 	struct ifnet *dst_if;
2045 	int error = 0;
2046 
2047 	sc = ifp->if_softc;
2048 
2049 	ETHER_BPF_MTAP(ifp, m);
2050 
2051 	eh = mtod(m, struct ether_header *);
2052 
2053 	BRIDGE_LOCK(sc);
2054 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2055 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2056 		BRIDGE_UNLOCK(sc);
2057 		error = bridge_enqueue(sc, dst_if, m);
2058 	} else
2059 		bridge_broadcast(sc, ifp, m, 0);
2060 
2061 	return (error);
2062 }
2063 
2064 /*
2065  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2066  */
2067 static void
2068 bridge_qflush(struct ifnet *ifp __unused)
2069 {
2070 }
2071 
2072 /*
2073  * bridge_forward:
2074  *
2075  *	The forwarding function of the bridge.
2076  *
2077  *	NOTE: Releases the lock on return.
2078  */
2079 static void
2080 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2081     struct mbuf *m)
2082 {
2083 	struct bridge_iflist *dbif;
2084 	struct ifnet *src_if, *dst_if, *ifp;
2085 	struct ether_header *eh;
2086 	uint16_t vlan;
2087 	uint8_t *dst;
2088 	int error;
2089 
2090 	src_if = m->m_pkthdr.rcvif;
2091 	ifp = sc->sc_ifp;
2092 
2093 	ifp->if_ipackets++;
2094 	ifp->if_ibytes += m->m_pkthdr.len;
2095 	vlan = VLANTAGOF(m);
2096 
2097 	if ((sbif->bif_flags & IFBIF_STP) &&
2098 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2099 		goto drop;
2100 
2101 	eh = mtod(m, struct ether_header *);
2102 	dst = eh->ether_dhost;
2103 
2104 	/* If the interface is learning, record the address. */
2105 	if (sbif->bif_flags & IFBIF_LEARNING) {
2106 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2107 		    sbif, 0, IFBAF_DYNAMIC);
2108 		/*
2109 		 * If the interface has addresses limits then deny any source
2110 		 * that is not in the cache.
2111 		 */
2112 		if (error && sbif->bif_addrmax)
2113 			goto drop;
2114 	}
2115 
2116 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2117 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2118 		goto drop;
2119 
2120 	/*
2121 	 * At this point, the port either doesn't participate
2122 	 * in spanning tree or it is in the forwarding state.
2123 	 */
2124 
2125 	/*
2126 	 * If the packet is unicast, destined for someone on
2127 	 * "this" side of the bridge, drop it.
2128 	 */
2129 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2130 		dst_if = bridge_rtlookup(sc, dst, vlan);
2131 		if (src_if == dst_if)
2132 			goto drop;
2133 	} else {
2134 		/*
2135 		 * Check if its a reserved multicast address, any address
2136 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2137 		 * bridge.
2138 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2139 		 */
2140 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2141 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2142 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2143 			goto drop;
2144 
2145 		/* ...forward it to all interfaces. */
2146 		ifp->if_imcasts++;
2147 		dst_if = NULL;
2148 	}
2149 
2150 	/*
2151 	 * If we have a destination interface which is a member of our bridge,
2152 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2153 	 * For broadcast or multicast packets, don't bother because it will
2154 	 * be reinjected into ether_input. We do this before we pass the packets
2155 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2156 	 * drop the packet, or possibly modify it, making it difficult to debug
2157 	 * firewall issues on the bridge.
2158 	 */
2159 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2160 		ETHER_BPF_MTAP(ifp, m);
2161 
2162 	/* run the packet filter */
2163 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2164 #ifdef INET6
2165 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2166 #endif
2167 	    ) {
2168 		BRIDGE_UNLOCK(sc);
2169 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2170 			return;
2171 		if (m == NULL)
2172 			return;
2173 		BRIDGE_LOCK(sc);
2174 	}
2175 
2176 	if (dst_if == NULL) {
2177 		bridge_broadcast(sc, src_if, m, 1);
2178 		return;
2179 	}
2180 
2181 	/*
2182 	 * At this point, we're dealing with a unicast frame
2183 	 * going to a different interface.
2184 	 */
2185 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2186 		goto drop;
2187 
2188 	dbif = bridge_lookup_member_if(sc, dst_if);
2189 	if (dbif == NULL)
2190 		/* Not a member of the bridge (anymore?) */
2191 		goto drop;
2192 
2193 	/* Private segments can not talk to each other */
2194 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2195 		goto drop;
2196 
2197 	if ((dbif->bif_flags & IFBIF_STP) &&
2198 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2199 		goto drop;
2200 
2201 	BRIDGE_UNLOCK(sc);
2202 
2203 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2204 #ifdef INET6
2205 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2206 #endif
2207 	    ) {
2208 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2209 			return;
2210 		if (m == NULL)
2211 			return;
2212 	}
2213 
2214 	bridge_enqueue(sc, dst_if, m);
2215 	return;
2216 
2217 drop:
2218 	BRIDGE_UNLOCK(sc);
2219 	m_freem(m);
2220 }
2221 
2222 /*
2223  * bridge_input:
2224  *
2225  *	Receive input from a member interface.  Queue the packet for
2226  *	bridging if it is not for us.
2227  */
2228 static struct mbuf *
2229 bridge_input(struct ifnet *ifp, struct mbuf *m)
2230 {
2231 	struct bridge_softc *sc = ifp->if_bridge;
2232 	struct bridge_iflist *bif, *bif2;
2233 	struct ifnet *bifp;
2234 	struct ether_header *eh;
2235 	struct mbuf *mc, *mc2;
2236 	uint16_t vlan;
2237 	int error;
2238 
2239 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2240 		return (m);
2241 
2242 	bifp = sc->sc_ifp;
2243 	vlan = VLANTAGOF(m);
2244 
2245 	/*
2246 	 * Implement support for bridge monitoring. If this flag has been
2247 	 * set on this interface, discard the packet once we push it through
2248 	 * the bpf(4) machinery, but before we do, increment the byte and
2249 	 * packet counters associated with this interface.
2250 	 */
2251 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2252 		m->m_pkthdr.rcvif  = bifp;
2253 		ETHER_BPF_MTAP(bifp, m);
2254 		bifp->if_ipackets++;
2255 		bifp->if_ibytes += m->m_pkthdr.len;
2256 		m_freem(m);
2257 		return (NULL);
2258 	}
2259 	BRIDGE_LOCK(sc);
2260 	bif = bridge_lookup_member_if(sc, ifp);
2261 	if (bif == NULL) {
2262 		BRIDGE_UNLOCK(sc);
2263 		return (m);
2264 	}
2265 
2266 	eh = mtod(m, struct ether_header *);
2267 
2268 	bridge_span(sc, m);
2269 
2270 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2271 		/* Tap off 802.1D packets; they do not get forwarded. */
2272 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2273 		    ETHER_ADDR_LEN) == 0) {
2274 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2275 			BRIDGE_UNLOCK(sc);
2276 			return (NULL);
2277 		}
2278 
2279 		if ((bif->bif_flags & IFBIF_STP) &&
2280 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2281 			BRIDGE_UNLOCK(sc);
2282 			return (m);
2283 		}
2284 
2285 		/*
2286 		 * Make a deep copy of the packet and enqueue the copy
2287 		 * for bridge processing; return the original packet for
2288 		 * local processing.
2289 		 */
2290 		mc = m_dup(m, M_NOWAIT);
2291 		if (mc == NULL) {
2292 			BRIDGE_UNLOCK(sc);
2293 			return (m);
2294 		}
2295 
2296 		/* Perform the bridge forwarding function with the copy. */
2297 		bridge_forward(sc, bif, mc);
2298 
2299 		/*
2300 		 * Reinject the mbuf as arriving on the bridge so we have a
2301 		 * chance at claiming multicast packets. We can not loop back
2302 		 * here from ether_input as a bridge is never a member of a
2303 		 * bridge.
2304 		 */
2305 		KASSERT(bifp->if_bridge == NULL,
2306 		    ("loop created in bridge_input"));
2307 		mc2 = m_dup(m, M_NOWAIT);
2308 		if (mc2 != NULL) {
2309 			/* Keep the layer3 header aligned */
2310 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2311 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2312 		}
2313 		if (mc2 != NULL) {
2314 			mc2->m_pkthdr.rcvif = bifp;
2315 			(*bifp->if_input)(bifp, mc2);
2316 		}
2317 
2318 		/* Return the original packet for local processing. */
2319 		return (m);
2320 	}
2321 
2322 	if ((bif->bif_flags & IFBIF_STP) &&
2323 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2324 		BRIDGE_UNLOCK(sc);
2325 		return (m);
2326 	}
2327 
2328 #if (defined(INET) || defined(INET6))
2329 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2330 	|| ((iface)->if_carp \
2331 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2332 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2333 	|| ((iface)->if_carp \
2334 	    && (*carp_forus_p)((iface), eh->ether_shost))
2335 #else
2336 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2337 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2338 #endif
2339 
2340 #ifdef INET6
2341 #   define OR_PFIL_HOOKED_INET6 \
2342 	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2343 #else
2344 #   define OR_PFIL_HOOKED_INET6
2345 #endif
2346 
2347 #define GRAB_OUR_PACKETS(iface) \
2348 	if ((iface)->if_type == IFT_GIF) \
2349 		continue; \
2350 	/* It is destined for us. */ \
2351 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2352 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2353 	    ) {								\
2354 		if ((iface)->if_type == IFT_BRIDGE) {			\
2355 			ETHER_BPF_MTAP(iface, m);			\
2356 			iface->if_ipackets++;				\
2357 			iface->if_ibytes += m->m_pkthdr.len;		\
2358 			/* Filter on the physical interface. */		\
2359 			if (pfil_local_phys &&				\
2360 			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2361 			     OR_PFIL_HOOKED_INET6)) {			\
2362 				if (bridge_pfil(&m, NULL, ifp,		\
2363 				    PFIL_IN) != 0 || m == NULL) {	\
2364 					BRIDGE_UNLOCK(sc);		\
2365 					return (NULL);			\
2366 				}					\
2367 				eh = mtod(m, struct ether_header *);	\
2368 			}						\
2369 		}							\
2370 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2371 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2372 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2373 			if (error && bif->bif_addrmax) {		\
2374 				BRIDGE_UNLOCK(sc);			\
2375 				m_freem(m);				\
2376 				return (NULL);				\
2377 			}						\
2378 		}							\
2379 		m->m_pkthdr.rcvif = iface;				\
2380 		BRIDGE_UNLOCK(sc);					\
2381 		return (m);						\
2382 	}								\
2383 									\
2384 	/* We just received a packet that we sent out. */		\
2385 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2386 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2387 	    ) {								\
2388 		BRIDGE_UNLOCK(sc);					\
2389 		m_freem(m);						\
2390 		return (NULL);						\
2391 	}
2392 
2393 	/*
2394 	 * Unicast.  Make sure it's not for the bridge.
2395 	 */
2396 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2397 
2398 	/*
2399 	 * Give a chance for ifp at first priority. This will help when	the
2400 	 * packet comes through the interface like VLAN's with the same MACs
2401 	 * on several interfaces from the same bridge. This also will save
2402 	 * some CPU cycles in case the destination interface and the input
2403 	 * interface (eq ifp) are the same.
2404 	 */
2405 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2406 
2407 	/* Now check the all bridge members. */
2408 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2409 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2410 	}
2411 
2412 #undef OR_CARP_CHECK_WE_ARE_DST
2413 #undef OR_CARP_CHECK_WE_ARE_SRC
2414 #undef OR_PFIL_HOOKED_INET6
2415 #undef GRAB_OUR_PACKETS
2416 
2417 	/* Perform the bridge forwarding function. */
2418 	bridge_forward(sc, bif, m);
2419 
2420 	return (NULL);
2421 }
2422 
2423 /*
2424  * bridge_broadcast:
2425  *
2426  *	Send a frame to all interfaces that are members of
2427  *	the bridge, except for the one on which the packet
2428  *	arrived.
2429  *
2430  *	NOTE: Releases the lock on return.
2431  */
2432 static void
2433 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2434     struct mbuf *m, int runfilt)
2435 {
2436 	struct bridge_iflist *dbif, *sbif;
2437 	struct mbuf *mc;
2438 	struct ifnet *dst_if;
2439 	int error = 0, used = 0, i;
2440 
2441 	sbif = bridge_lookup_member_if(sc, src_if);
2442 
2443 	BRIDGE_LOCK2REF(sc, error);
2444 	if (error) {
2445 		m_freem(m);
2446 		return;
2447 	}
2448 
2449 	/* Filter on the bridge interface before broadcasting */
2450 	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2451 #ifdef INET6
2452 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2453 #endif
2454 	    )) {
2455 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2456 			goto out;
2457 		if (m == NULL)
2458 			goto out;
2459 	}
2460 
2461 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2462 		dst_if = dbif->bif_ifp;
2463 		if (dst_if == src_if)
2464 			continue;
2465 
2466 		/* Private segments can not talk to each other */
2467 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2468 			continue;
2469 
2470 		if ((dbif->bif_flags & IFBIF_STP) &&
2471 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2472 			continue;
2473 
2474 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2475 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2476 			continue;
2477 
2478 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2479 			continue;
2480 
2481 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2482 			mc = m;
2483 			used = 1;
2484 		} else {
2485 			mc = m_dup(m, M_NOWAIT);
2486 			if (mc == NULL) {
2487 				sc->sc_ifp->if_oerrors++;
2488 				continue;
2489 			}
2490 		}
2491 
2492 		/*
2493 		 * Filter on the output interface. Pass a NULL bridge interface
2494 		 * pointer so we do not redundantly filter on the bridge for
2495 		 * each interface we broadcast on.
2496 		 */
2497 		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2498 #ifdef INET6
2499 		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2500 #endif
2501 		    )) {
2502 			if (used == 0) {
2503 				/* Keep the layer3 header aligned */
2504 				i = min(mc->m_pkthdr.len, max_protohdr);
2505 				mc = m_copyup(mc, i, ETHER_ALIGN);
2506 				if (mc == NULL) {
2507 					sc->sc_ifp->if_oerrors++;
2508 					continue;
2509 				}
2510 			}
2511 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2512 				continue;
2513 			if (mc == NULL)
2514 				continue;
2515 		}
2516 
2517 		bridge_enqueue(sc, dst_if, mc);
2518 	}
2519 	if (used == 0)
2520 		m_freem(m);
2521 
2522 out:
2523 	BRIDGE_UNREF(sc);
2524 }
2525 
2526 /*
2527  * bridge_span:
2528  *
2529  *	Duplicate a packet out one or more interfaces that are in span mode,
2530  *	the original mbuf is unmodified.
2531  */
2532 static void
2533 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2534 {
2535 	struct bridge_iflist *bif;
2536 	struct ifnet *dst_if;
2537 	struct mbuf *mc;
2538 
2539 	if (LIST_EMPTY(&sc->sc_spanlist))
2540 		return;
2541 
2542 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2543 		dst_if = bif->bif_ifp;
2544 
2545 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2546 			continue;
2547 
2548 		mc = m_copypacket(m, M_NOWAIT);
2549 		if (mc == NULL) {
2550 			sc->sc_ifp->if_oerrors++;
2551 			continue;
2552 		}
2553 
2554 		bridge_enqueue(sc, dst_if, mc);
2555 	}
2556 }
2557 
2558 /*
2559  * bridge_rtupdate:
2560  *
2561  *	Add a bridge routing entry.
2562  */
2563 static int
2564 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2565     struct bridge_iflist *bif, int setflags, uint8_t flags)
2566 {
2567 	struct bridge_rtnode *brt;
2568 	int error;
2569 
2570 	BRIDGE_LOCK_ASSERT(sc);
2571 
2572 	/* Check the source address is valid and not multicast. */
2573 	if (ETHER_IS_MULTICAST(dst) ||
2574 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2575 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2576 		return (EINVAL);
2577 
2578 	/* 802.1p frames map to vlan 1 */
2579 	if (vlan == 0)
2580 		vlan = 1;
2581 
2582 	/*
2583 	 * A route for this destination might already exist.  If so,
2584 	 * update it, otherwise create a new one.
2585 	 */
2586 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2587 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2588 			sc->sc_brtexceeded++;
2589 			return (ENOSPC);
2590 		}
2591 		/* Check per interface address limits (if enabled) */
2592 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2593 			bif->bif_addrexceeded++;
2594 			return (ENOSPC);
2595 		}
2596 
2597 		/*
2598 		 * Allocate a new bridge forwarding node, and
2599 		 * initialize the expiration time and Ethernet
2600 		 * address.
2601 		 */
2602 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2603 		if (brt == NULL)
2604 			return (ENOMEM);
2605 
2606 		if (bif->bif_flags & IFBIF_STICKY)
2607 			brt->brt_flags = IFBAF_STICKY;
2608 		else
2609 			brt->brt_flags = IFBAF_DYNAMIC;
2610 
2611 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2612 		brt->brt_vlan = vlan;
2613 
2614 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2615 			uma_zfree(bridge_rtnode_zone, brt);
2616 			return (error);
2617 		}
2618 		brt->brt_dst = bif;
2619 		bif->bif_addrcnt++;
2620 	}
2621 
2622 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2623 	    brt->brt_dst != bif) {
2624 		brt->brt_dst->bif_addrcnt--;
2625 		brt->brt_dst = bif;
2626 		brt->brt_dst->bif_addrcnt++;
2627 	}
2628 
2629 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2630 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2631 	if (setflags)
2632 		brt->brt_flags = flags;
2633 
2634 	return (0);
2635 }
2636 
2637 /*
2638  * bridge_rtlookup:
2639  *
2640  *	Lookup the destination interface for an address.
2641  */
2642 static struct ifnet *
2643 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2644 {
2645 	struct bridge_rtnode *brt;
2646 
2647 	BRIDGE_LOCK_ASSERT(sc);
2648 
2649 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2650 		return (NULL);
2651 
2652 	return (brt->brt_ifp);
2653 }
2654 
2655 /*
2656  * bridge_rttrim:
2657  *
2658  *	Trim the routine table so that we have a number
2659  *	of routing entries less than or equal to the
2660  *	maximum number.
2661  */
2662 static void
2663 bridge_rttrim(struct bridge_softc *sc)
2664 {
2665 	struct bridge_rtnode *brt, *nbrt;
2666 
2667 	BRIDGE_LOCK_ASSERT(sc);
2668 
2669 	/* Make sure we actually need to do this. */
2670 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2671 		return;
2672 
2673 	/* Force an aging cycle; this might trim enough addresses. */
2674 	bridge_rtage(sc);
2675 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2676 		return;
2677 
2678 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2679 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2680 			bridge_rtnode_destroy(sc, brt);
2681 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2682 				return;
2683 		}
2684 	}
2685 }
2686 
2687 /*
2688  * bridge_timer:
2689  *
2690  *	Aging timer for the bridge.
2691  */
2692 static void
2693 bridge_timer(void *arg)
2694 {
2695 	struct bridge_softc *sc = arg;
2696 
2697 	BRIDGE_LOCK_ASSERT(sc);
2698 
2699 	bridge_rtage(sc);
2700 
2701 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2702 		callout_reset(&sc->sc_brcallout,
2703 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2704 }
2705 
2706 /*
2707  * bridge_rtage:
2708  *
2709  *	Perform an aging cycle.
2710  */
2711 static void
2712 bridge_rtage(struct bridge_softc *sc)
2713 {
2714 	struct bridge_rtnode *brt, *nbrt;
2715 
2716 	BRIDGE_LOCK_ASSERT(sc);
2717 
2718 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2719 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2720 			if (time_uptime >= brt->brt_expire)
2721 				bridge_rtnode_destroy(sc, brt);
2722 		}
2723 	}
2724 }
2725 
2726 /*
2727  * bridge_rtflush:
2728  *
2729  *	Remove all dynamic addresses from the bridge.
2730  */
2731 static void
2732 bridge_rtflush(struct bridge_softc *sc, int full)
2733 {
2734 	struct bridge_rtnode *brt, *nbrt;
2735 
2736 	BRIDGE_LOCK_ASSERT(sc);
2737 
2738 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2739 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2740 			bridge_rtnode_destroy(sc, brt);
2741 	}
2742 }
2743 
2744 /*
2745  * bridge_rtdaddr:
2746  *
2747  *	Remove an address from the table.
2748  */
2749 static int
2750 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2751 {
2752 	struct bridge_rtnode *brt;
2753 	int found = 0;
2754 
2755 	BRIDGE_LOCK_ASSERT(sc);
2756 
2757 	/*
2758 	 * If vlan is zero then we want to delete for all vlans so the lookup
2759 	 * may return more than one.
2760 	 */
2761 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2762 		bridge_rtnode_destroy(sc, brt);
2763 		found = 1;
2764 	}
2765 
2766 	return (found ? 0 : ENOENT);
2767 }
2768 
2769 /*
2770  * bridge_rtdelete:
2771  *
2772  *	Delete routes to a speicifc member interface.
2773  */
2774 static void
2775 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2776 {
2777 	struct bridge_rtnode *brt, *nbrt;
2778 
2779 	BRIDGE_LOCK_ASSERT(sc);
2780 
2781 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2782 		if (brt->brt_ifp == ifp && (full ||
2783 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2784 			bridge_rtnode_destroy(sc, brt);
2785 	}
2786 }
2787 
2788 /*
2789  * bridge_rtable_init:
2790  *
2791  *	Initialize the route table for this bridge.
2792  */
2793 static void
2794 bridge_rtable_init(struct bridge_softc *sc)
2795 {
2796 	int i;
2797 
2798 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2799 	    M_DEVBUF, M_WAITOK);
2800 
2801 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2802 		LIST_INIT(&sc->sc_rthash[i]);
2803 
2804 	sc->sc_rthash_key = arc4random();
2805 	LIST_INIT(&sc->sc_rtlist);
2806 }
2807 
2808 /*
2809  * bridge_rtable_fini:
2810  *
2811  *	Deconstruct the route table for this bridge.
2812  */
2813 static void
2814 bridge_rtable_fini(struct bridge_softc *sc)
2815 {
2816 
2817 	KASSERT(sc->sc_brtcnt == 0,
2818 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2819 	free(sc->sc_rthash, M_DEVBUF);
2820 }
2821 
2822 /*
2823  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2824  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2825  */
2826 #define	mix(a, b, c)							\
2827 do {									\
2828 	a -= b; a -= c; a ^= (c >> 13);					\
2829 	b -= c; b -= a; b ^= (a << 8);					\
2830 	c -= a; c -= b; c ^= (b >> 13);					\
2831 	a -= b; a -= c; a ^= (c >> 12);					\
2832 	b -= c; b -= a; b ^= (a << 16);					\
2833 	c -= a; c -= b; c ^= (b >> 5);					\
2834 	a -= b; a -= c; a ^= (c >> 3);					\
2835 	b -= c; b -= a; b ^= (a << 10);					\
2836 	c -= a; c -= b; c ^= (b >> 15);					\
2837 } while (/*CONSTCOND*/0)
2838 
2839 static __inline uint32_t
2840 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2841 {
2842 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2843 
2844 	b += addr[5] << 8;
2845 	b += addr[4];
2846 	a += addr[3] << 24;
2847 	a += addr[2] << 16;
2848 	a += addr[1] << 8;
2849 	a += addr[0];
2850 
2851 	mix(a, b, c);
2852 
2853 	return (c & BRIDGE_RTHASH_MASK);
2854 }
2855 
2856 #undef mix
2857 
2858 static int
2859 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2860 {
2861 	int i, d;
2862 
2863 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2864 		d = ((int)a[i]) - ((int)b[i]);
2865 	}
2866 
2867 	return (d);
2868 }
2869 
2870 /*
2871  * bridge_rtnode_lookup:
2872  *
2873  *	Look up a bridge route node for the specified destination. Compare the
2874  *	vlan id or if zero then just return the first match.
2875  */
2876 static struct bridge_rtnode *
2877 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2878 {
2879 	struct bridge_rtnode *brt;
2880 	uint32_t hash;
2881 	int dir;
2882 
2883 	BRIDGE_LOCK_ASSERT(sc);
2884 
2885 	hash = bridge_rthash(sc, addr);
2886 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2887 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2888 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2889 			return (brt);
2890 		if (dir > 0)
2891 			return (NULL);
2892 	}
2893 
2894 	return (NULL);
2895 }
2896 
2897 /*
2898  * bridge_rtnode_insert:
2899  *
2900  *	Insert the specified bridge node into the route table.  We
2901  *	assume the entry is not already in the table.
2902  */
2903 static int
2904 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2905 {
2906 	struct bridge_rtnode *lbrt;
2907 	uint32_t hash;
2908 	int dir;
2909 
2910 	BRIDGE_LOCK_ASSERT(sc);
2911 
2912 	hash = bridge_rthash(sc, brt->brt_addr);
2913 
2914 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2915 	if (lbrt == NULL) {
2916 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2917 		goto out;
2918 	}
2919 
2920 	do {
2921 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2922 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2923 			return (EEXIST);
2924 		if (dir > 0) {
2925 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2926 			goto out;
2927 		}
2928 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2929 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2930 			goto out;
2931 		}
2932 		lbrt = LIST_NEXT(lbrt, brt_hash);
2933 	} while (lbrt != NULL);
2934 
2935 #ifdef DIAGNOSTIC
2936 	panic("bridge_rtnode_insert: impossible");
2937 #endif
2938 
2939 out:
2940 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2941 	sc->sc_brtcnt++;
2942 
2943 	return (0);
2944 }
2945 
2946 /*
2947  * bridge_rtnode_destroy:
2948  *
2949  *	Destroy a bridge rtnode.
2950  */
2951 static void
2952 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2953 {
2954 	BRIDGE_LOCK_ASSERT(sc);
2955 
2956 	LIST_REMOVE(brt, brt_hash);
2957 
2958 	LIST_REMOVE(brt, brt_list);
2959 	sc->sc_brtcnt--;
2960 	brt->brt_dst->bif_addrcnt--;
2961 	uma_zfree(bridge_rtnode_zone, brt);
2962 }
2963 
2964 /*
2965  * bridge_rtable_expire:
2966  *
2967  *	Set the expiry time for all routes on an interface.
2968  */
2969 static void
2970 bridge_rtable_expire(struct ifnet *ifp, int age)
2971 {
2972 	struct bridge_softc *sc = ifp->if_bridge;
2973 	struct bridge_rtnode *brt;
2974 
2975 	BRIDGE_LOCK(sc);
2976 
2977 	/*
2978 	 * If the age is zero then flush, otherwise set all the expiry times to
2979 	 * age for the interface
2980 	 */
2981 	if (age == 0)
2982 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2983 	else {
2984 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2985 			/* Cap the expiry time to 'age' */
2986 			if (brt->brt_ifp == ifp &&
2987 			    brt->brt_expire > time_uptime + age &&
2988 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2989 				brt->brt_expire = time_uptime + age;
2990 		}
2991 	}
2992 	BRIDGE_UNLOCK(sc);
2993 }
2994 
2995 /*
2996  * bridge_state_change:
2997  *
2998  *	Callback from the bridgestp code when a port changes states.
2999  */
3000 static void
3001 bridge_state_change(struct ifnet *ifp, int state)
3002 {
3003 	struct bridge_softc *sc = ifp->if_bridge;
3004 	static const char *stpstates[] = {
3005 		"disabled",
3006 		"listening",
3007 		"learning",
3008 		"forwarding",
3009 		"blocking",
3010 		"discarding"
3011 	};
3012 
3013 	if (log_stp)
3014 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3015 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3016 }
3017 
3018 /*
3019  * Send bridge packets through pfil if they are one of the types pfil can deal
3020  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3021  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3022  * that interface.
3023  */
3024 static int
3025 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3026 {
3027 	int snap, error, i, hlen;
3028 	struct ether_header *eh1, eh2;
3029 	struct ip *ip;
3030 	struct llc llc1;
3031 	u_int16_t ether_type;
3032 
3033 	snap = 0;
3034 	error = -1;	/* Default error if not error == 0 */
3035 
3036 #if 0
3037 	/* we may return with the IP fields swapped, ensure its not shared */
3038 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3039 #endif
3040 
3041 	if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
3042 		return (0); /* filtering is disabled */
3043 
3044 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3045 	if ((*mp)->m_len < i) {
3046 	    *mp = m_pullup(*mp, i);
3047 	    if (*mp == NULL) {
3048 		printf("%s: m_pullup failed\n", __func__);
3049 		return (-1);
3050 	    }
3051 	}
3052 
3053 	eh1 = mtod(*mp, struct ether_header *);
3054 	ether_type = ntohs(eh1->ether_type);
3055 
3056 	/*
3057 	 * Check for SNAP/LLC.
3058 	 */
3059 	if (ether_type < ETHERMTU) {
3060 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3061 
3062 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3063 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3064 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3065 		    llc2->llc_control == LLC_UI) {
3066 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3067 			snap = 1;
3068 		}
3069 	}
3070 
3071 	/*
3072 	 * If we're trying to filter bridge traffic, don't look at anything
3073 	 * other than IP and ARP traffic.  If the filter doesn't understand
3074 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3075 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3076 	 * but of course we don't have an AppleTalk filter to begin with.
3077 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3078 	 * ARP traffic.)
3079 	 */
3080 	switch (ether_type) {
3081 		case ETHERTYPE_ARP:
3082 		case ETHERTYPE_REVARP:
3083 			if (pfil_ipfw_arp == 0)
3084 				return (0); /* Automatically pass */
3085 			break;
3086 
3087 		case ETHERTYPE_IP:
3088 #ifdef INET6
3089 		case ETHERTYPE_IPV6:
3090 #endif /* INET6 */
3091 			break;
3092 		default:
3093 			/*
3094 			 * Check to see if the user wants to pass non-ip
3095 			 * packets, these will not be checked by pfil(9) and
3096 			 * passed unconditionally so the default is to drop.
3097 			 */
3098 			if (pfil_onlyip)
3099 				goto bad;
3100 	}
3101 
3102 	/* Run the packet through pfil before stripping link headers */
3103 	if (PFIL_HOOKED(&V_link_pfil_hook) && pfil_ipfw != 0 &&
3104 			dir == PFIL_OUT && ifp != NULL) {
3105 
3106 		error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, NULL);
3107 
3108 		if (*mp == NULL || error != 0) /* packet consumed by filter */
3109 			return (error);
3110 	}
3111 
3112 	/* Strip off the Ethernet header and keep a copy. */
3113 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3114 	m_adj(*mp, ETHER_HDR_LEN);
3115 
3116 	/* Strip off snap header, if present */
3117 	if (snap) {
3118 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3119 		m_adj(*mp, sizeof(struct llc));
3120 	}
3121 
3122 	/*
3123 	 * Check the IP header for alignment and errors
3124 	 */
3125 	if (dir == PFIL_IN) {
3126 		switch (ether_type) {
3127 			case ETHERTYPE_IP:
3128 				error = bridge_ip_checkbasic(mp);
3129 				break;
3130 #ifdef INET6
3131 			case ETHERTYPE_IPV6:
3132 				error = bridge_ip6_checkbasic(mp);
3133 				break;
3134 #endif /* INET6 */
3135 			default:
3136 				error = 0;
3137 		}
3138 		if (error)
3139 			goto bad;
3140 	}
3141 
3142 	error = 0;
3143 
3144 	/*
3145 	 * Run the packet through pfil
3146 	 */
3147 	switch (ether_type) {
3148 	case ETHERTYPE_IP:
3149 		/*
3150 		 * Run pfil on the member interface and the bridge, both can
3151 		 * be skipped by clearing pfil_member or pfil_bridge.
3152 		 *
3153 		 * Keep the order:
3154 		 *   in_if -> bridge_if -> out_if
3155 		 */
3156 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3157 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3158 					dir, NULL);
3159 
3160 		if (*mp == NULL || error != 0) /* filter may consume */
3161 			break;
3162 
3163 		if (pfil_member && ifp != NULL)
3164 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3165 					dir, NULL);
3166 
3167 		if (*mp == NULL || error != 0) /* filter may consume */
3168 			break;
3169 
3170 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3171 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3172 					dir, NULL);
3173 
3174 		if (*mp == NULL || error != 0) /* filter may consume */
3175 			break;
3176 
3177 		/* check if we need to fragment the packet */
3178 		if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3179 			i = (*mp)->m_pkthdr.len;
3180 			if (i > ifp->if_mtu) {
3181 				error = bridge_fragment(ifp, *mp, &eh2, snap,
3182 					    &llc1);
3183 				return (error);
3184 			}
3185 		}
3186 
3187 		/* Recalculate the ip checksum. */
3188 		ip = mtod(*mp, struct ip *);
3189 		hlen = ip->ip_hl << 2;
3190 		if (hlen < sizeof(struct ip))
3191 			goto bad;
3192 		if (hlen > (*mp)->m_len) {
3193 			if ((*mp = m_pullup(*mp, hlen)) == 0)
3194 				goto bad;
3195 			ip = mtod(*mp, struct ip *);
3196 			if (ip == NULL)
3197 				goto bad;
3198 		}
3199 		ip->ip_sum = 0;
3200 		if (hlen == sizeof(struct ip))
3201 			ip->ip_sum = in_cksum_hdr(ip);
3202 		else
3203 			ip->ip_sum = in_cksum(*mp, hlen);
3204 
3205 		break;
3206 #ifdef INET6
3207 	case ETHERTYPE_IPV6:
3208 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3209 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3210 					dir, NULL);
3211 
3212 		if (*mp == NULL || error != 0) /* filter may consume */
3213 			break;
3214 
3215 		if (pfil_member && ifp != NULL)
3216 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3217 					dir, NULL);
3218 
3219 		if (*mp == NULL || error != 0) /* filter may consume */
3220 			break;
3221 
3222 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3223 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3224 					dir, NULL);
3225 		break;
3226 #endif
3227 	default:
3228 		error = 0;
3229 		break;
3230 	}
3231 
3232 	if (*mp == NULL)
3233 		return (error);
3234 	if (error != 0)
3235 		goto bad;
3236 
3237 	error = -1;
3238 
3239 	/*
3240 	 * Finally, put everything back the way it was and return
3241 	 */
3242 	if (snap) {
3243 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3244 		if (*mp == NULL)
3245 			return (error);
3246 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3247 	}
3248 
3249 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3250 	if (*mp == NULL)
3251 		return (error);
3252 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3253 
3254 	return (0);
3255 
3256 bad:
3257 	m_freem(*mp);
3258 	*mp = NULL;
3259 	return (error);
3260 }
3261 
3262 /*
3263  * Perform basic checks on header size since
3264  * pfil assumes ip_input has already processed
3265  * it for it.  Cut-and-pasted from ip_input.c.
3266  * Given how simple the IPv6 version is,
3267  * does the IPv4 version really need to be
3268  * this complicated?
3269  *
3270  * XXX Should we update ipstat here, or not?
3271  * XXX Right now we update ipstat but not
3272  * XXX csum_counter.
3273  */
3274 static int
3275 bridge_ip_checkbasic(struct mbuf **mp)
3276 {
3277 	struct mbuf *m = *mp;
3278 	struct ip *ip;
3279 	int len, hlen;
3280 	u_short sum;
3281 
3282 	if (*mp == NULL)
3283 		return (-1);
3284 
3285 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3286 		if ((m = m_copyup(m, sizeof(struct ip),
3287 			(max_linkhdr + 3) & ~3)) == NULL) {
3288 			/* XXXJRT new stat, please */
3289 			KMOD_IPSTAT_INC(ips_toosmall);
3290 			goto bad;
3291 		}
3292 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3293 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3294 			KMOD_IPSTAT_INC(ips_toosmall);
3295 			goto bad;
3296 		}
3297 	}
3298 	ip = mtod(m, struct ip *);
3299 	if (ip == NULL) goto bad;
3300 
3301 	if (ip->ip_v != IPVERSION) {
3302 		KMOD_IPSTAT_INC(ips_badvers);
3303 		goto bad;
3304 	}
3305 	hlen = ip->ip_hl << 2;
3306 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3307 		KMOD_IPSTAT_INC(ips_badhlen);
3308 		goto bad;
3309 	}
3310 	if (hlen > m->m_len) {
3311 		if ((m = m_pullup(m, hlen)) == 0) {
3312 			KMOD_IPSTAT_INC(ips_badhlen);
3313 			goto bad;
3314 		}
3315 		ip = mtod(m, struct ip *);
3316 		if (ip == NULL) goto bad;
3317 	}
3318 
3319 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3320 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3321 	} else {
3322 		if (hlen == sizeof(struct ip)) {
3323 			sum = in_cksum_hdr(ip);
3324 		} else {
3325 			sum = in_cksum(m, hlen);
3326 		}
3327 	}
3328 	if (sum) {
3329 		KMOD_IPSTAT_INC(ips_badsum);
3330 		goto bad;
3331 	}
3332 
3333 	/* Retrieve the packet length. */
3334 	len = ntohs(ip->ip_len);
3335 
3336 	/*
3337 	 * Check for additional length bogosity
3338 	 */
3339 	if (len < hlen) {
3340 		KMOD_IPSTAT_INC(ips_badlen);
3341 		goto bad;
3342 	}
3343 
3344 	/*
3345 	 * Check that the amount of data in the buffers
3346 	 * is as at least much as the IP header would have us expect.
3347 	 * Drop packet if shorter than we expect.
3348 	 */
3349 	if (m->m_pkthdr.len < len) {
3350 		KMOD_IPSTAT_INC(ips_tooshort);
3351 		goto bad;
3352 	}
3353 
3354 	/* Checks out, proceed */
3355 	*mp = m;
3356 	return (0);
3357 
3358 bad:
3359 	*mp = m;
3360 	return (-1);
3361 }
3362 
3363 #ifdef INET6
3364 /*
3365  * Same as above, but for IPv6.
3366  * Cut-and-pasted from ip6_input.c.
3367  * XXX Should we update ip6stat, or not?
3368  */
3369 static int
3370 bridge_ip6_checkbasic(struct mbuf **mp)
3371 {
3372 	struct mbuf *m = *mp;
3373 	struct ip6_hdr *ip6;
3374 
3375 	/*
3376 	 * If the IPv6 header is not aligned, slurp it up into a new
3377 	 * mbuf with space for link headers, in the event we forward
3378 	 * it.  Otherwise, if it is aligned, make sure the entire base
3379 	 * IPv6 header is in the first mbuf of the chain.
3380 	 */
3381 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3382 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3383 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3384 			    (max_linkhdr + 3) & ~3)) == NULL) {
3385 			/* XXXJRT new stat, please */
3386 			IP6STAT_INC(ip6s_toosmall);
3387 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3388 			goto bad;
3389 		}
3390 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3391 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3392 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3393 			IP6STAT_INC(ip6s_toosmall);
3394 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3395 			goto bad;
3396 		}
3397 	}
3398 
3399 	ip6 = mtod(m, struct ip6_hdr *);
3400 
3401 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3402 		IP6STAT_INC(ip6s_badvers);
3403 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3404 		goto bad;
3405 	}
3406 
3407 	/* Checks out, proceed */
3408 	*mp = m;
3409 	return (0);
3410 
3411 bad:
3412 	*mp = m;
3413 	return (-1);
3414 }
3415 #endif /* INET6 */
3416 
3417 /*
3418  * bridge_fragment:
3419  *
3420  *	Return a fragmented mbuf chain.
3421  */
3422 static int
3423 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3424     int snap, struct llc *llc)
3425 {
3426 	struct mbuf *m0;
3427 	struct ip *ip;
3428 	int error = -1;
3429 
3430 	if (m->m_len < sizeof(struct ip) &&
3431 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3432 		goto out;
3433 	ip = mtod(m, struct ip *);
3434 
3435 	m->m_pkthdr.csum_flags |= CSUM_IP;
3436 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3437 	if (error)
3438 		goto out;
3439 
3440 	/* walk the chain and re-add the Ethernet header */
3441 	for (m0 = m; m0; m0 = m0->m_nextpkt) {
3442 		if (error == 0) {
3443 			if (snap) {
3444 				M_PREPEND(m0, sizeof(struct llc), M_NOWAIT);
3445 				if (m0 == NULL) {
3446 					error = ENOBUFS;
3447 					continue;
3448 				}
3449 				bcopy(llc, mtod(m0, caddr_t),
3450 				    sizeof(struct llc));
3451 			}
3452 			M_PREPEND(m0, ETHER_HDR_LEN, M_NOWAIT);
3453 			if (m0 == NULL) {
3454 				error = ENOBUFS;
3455 				continue;
3456 			}
3457 			bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3458 		} else
3459 			m_freem(m);
3460 	}
3461 
3462 	if (error == 0)
3463 		KMOD_IPSTAT_INC(ips_fragmented);
3464 
3465 	return (error);
3466 
3467 out:
3468 	if (m != NULL)
3469 		m_freem(m);
3470 	return (error);
3471 }
3472 
3473 static void
3474 bridge_linkstate(struct ifnet *ifp)
3475 {
3476 	struct bridge_softc *sc = ifp->if_bridge;
3477 	struct bridge_iflist *bif;
3478 
3479 	BRIDGE_LOCK(sc);
3480 	bif = bridge_lookup_member_if(sc, ifp);
3481 	if (bif == NULL) {
3482 		BRIDGE_UNLOCK(sc);
3483 		return;
3484 	}
3485 	bridge_linkcheck(sc);
3486 	BRIDGE_UNLOCK(sc);
3487 
3488 	bstp_linkstate(&bif->bif_stp);
3489 }
3490 
3491 static void
3492 bridge_linkcheck(struct bridge_softc *sc)
3493 {
3494 	struct bridge_iflist *bif;
3495 	int new_link, hasls;
3496 
3497 	BRIDGE_LOCK_ASSERT(sc);
3498 	new_link = LINK_STATE_DOWN;
3499 	hasls = 0;
3500 	/* Our link is considered up if at least one of our ports is active */
3501 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3502 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3503 			hasls++;
3504 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3505 			new_link = LINK_STATE_UP;
3506 			break;
3507 		}
3508 	}
3509 	if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3510 		/* If no interfaces support link-state then we default to up */
3511 		new_link = LINK_STATE_UP;
3512 	}
3513 	if_link_state_change(sc->sc_ifp, new_link);
3514 }
3515