xref: /freebsd/sys/net/if_bridge.c (revision b1d046441de9053152c7cf03d6b60d9882687e1b)
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 #include <sys/rwlock.h>
104 
105 #include <net/bpf.h>
106 #include <net/if.h>
107 #include <net/if_clone.h>
108 #include <net/if_dl.h>
109 #include <net/if_types.h>
110 #include <net/if_var.h>
111 #include <net/pfil.h>
112 #include <net/vnet.h>
113 
114 #include <netinet/in.h> /* for struct arpcom */
115 #include <netinet/in_systm.h>
116 #include <netinet/in_var.h>
117 #include <netinet/ip.h>
118 #include <netinet/ip_var.h>
119 #ifdef INET6
120 #include <netinet/ip6.h>
121 #include <netinet6/ip6_var.h>
122 #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 #include <netinet/ip_fw.h>
135 #include <netinet/ipfw/ip_fw_private.h>
136 
137 /*
138  * Size of the route hash table.  Must be a power of two.
139  */
140 #ifndef BRIDGE_RTHASH_SIZE
141 #define	BRIDGE_RTHASH_SIZE		1024
142 #endif
143 
144 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
145 
146 /*
147  * Maximum number of addresses to cache.
148  */
149 #ifndef BRIDGE_RTABLE_MAX
150 #define	BRIDGE_RTABLE_MAX		100
151 #endif
152 
153 /*
154  * Timeout (in seconds) for entries learned dynamically.
155  */
156 #ifndef BRIDGE_RTABLE_TIMEOUT
157 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
158 #endif
159 
160 /*
161  * Number of seconds between walks of the route list.
162  */
163 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
164 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
165 #endif
166 
167 /*
168  * List of capabilities to possibly mask on the member interface.
169  */
170 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
171 
172 /*
173  * List of capabilities to strip
174  */
175 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
176 
177 /*
178  * Bridge interface list entry.
179  */
180 struct bridge_iflist {
181 	LIST_ENTRY(bridge_iflist) bif_next;
182 	struct ifnet		*bif_ifp;	/* member if */
183 	struct bstp_port	bif_stp;	/* STP state */
184 	uint32_t		bif_flags;	/* member if flags */
185 	int			bif_savedcaps;	/* saved capabilities */
186 	uint32_t		bif_addrmax;	/* max # of addresses */
187 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
188 	uint32_t		bif_addrexceeded;/* # of address violations */
189 };
190 
191 /*
192  * Bridge route node.
193  */
194 struct bridge_rtnode {
195 	LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
196 	LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
197 	struct bridge_iflist	*brt_dst;	/* destination if */
198 	unsigned long		brt_expire;	/* expiration time */
199 	uint8_t			brt_flags;	/* address flags */
200 	uint8_t			brt_addr[ETHER_ADDR_LEN];
201 	uint16_t		brt_vlan;	/* vlan id */
202 };
203 #define	brt_ifp			brt_dst->bif_ifp
204 
205 /*
206  * Software state for each bridge.
207  */
208 struct bridge_softc {
209 	struct ifnet		*sc_ifp;	/* make this an interface */
210 	LIST_ENTRY(bridge_softc) sc_list;
211 	struct mtx		sc_mtx;
212 	struct cv		sc_cv;
213 	uint32_t		sc_brtmax;	/* max # of addresses */
214 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
215 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
216 	struct callout		sc_brcallout;	/* bridge callout */
217 	uint32_t		sc_iflist_ref;	/* refcount for sc_iflist */
218 	uint32_t		sc_iflist_xcnt;	/* refcount for sc_iflist */
219 	LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
220 	LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
221 	LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
222 	uint32_t		sc_rthash_key;	/* key for hash */
223 	LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
224 	struct bstp_state	sc_stp;		/* STP state */
225 	uint32_t		sc_brtexceeded;	/* # of cache drops */
226 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
227 	u_char			sc_defaddr[6];	/* Default MAC address */
228 };
229 
230 static struct mtx 	bridge_list_mtx;
231 eventhandler_tag	bridge_detach_cookie = NULL;
232 
233 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
234 
235 uma_zone_t bridge_rtnode_zone;
236 
237 static int	bridge_clone_create(struct if_clone *, int, caddr_t);
238 static void	bridge_clone_destroy(struct ifnet *);
239 
240 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
241 static void	bridge_mutecaps(struct bridge_softc *);
242 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
243 		    int);
244 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
245 static void	bridge_init(void *);
246 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
247 static void	bridge_stop(struct ifnet *, int);
248 static void	bridge_start(struct ifnet *);
249 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
250 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
251 		    struct rtentry *);
252 static void	bridge_enqueue(struct bridge_softc *, struct ifnet *,
253 		    struct mbuf *);
254 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
255 
256 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
257 		    struct mbuf *m);
258 
259 static void	bridge_timer(void *);
260 
261 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
262 		    struct mbuf *, int);
263 static void	bridge_span(struct bridge_softc *, struct mbuf *);
264 
265 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
266 		    uint16_t, struct bridge_iflist *, int, uint8_t);
267 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
268 		    uint16_t);
269 static void	bridge_rttrim(struct bridge_softc *);
270 static void	bridge_rtage(struct bridge_softc *);
271 static void	bridge_rtflush(struct bridge_softc *, int);
272 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
273 		    uint16_t);
274 
275 static int	bridge_rtable_init(struct bridge_softc *);
276 static void	bridge_rtable_fini(struct bridge_softc *);
277 
278 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
279 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
280 		    const uint8_t *, uint16_t);
281 static int	bridge_rtnode_insert(struct bridge_softc *,
282 		    struct bridge_rtnode *);
283 static void	bridge_rtnode_destroy(struct bridge_softc *,
284 		    struct bridge_rtnode *);
285 static void	bridge_rtable_expire(struct ifnet *, int);
286 static void	bridge_state_change(struct ifnet *, int);
287 
288 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
289 		    const char *name);
290 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
291 		    struct ifnet *ifp);
292 static void	bridge_delete_member(struct bridge_softc *,
293 		    struct bridge_iflist *, int);
294 static void	bridge_delete_span(struct bridge_softc *,
295 		    struct bridge_iflist *);
296 
297 static int	bridge_ioctl_add(struct bridge_softc *, void *);
298 static int	bridge_ioctl_del(struct bridge_softc *, void *);
299 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
300 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
301 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
302 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
303 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
304 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
305 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
306 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
307 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
308 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
309 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
310 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
311 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
312 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
313 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
314 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
315 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
316 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
317 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
318 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
319 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
320 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
321 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
322 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
323 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
324 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
325 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
326 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
327 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
328 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
329 		    int);
330 static int	bridge_ip_checkbasic(struct mbuf **mp);
331 #ifdef INET6
332 static int	bridge_ip6_checkbasic(struct mbuf **mp);
333 #endif /* INET6 */
334 static int	bridge_fragment(struct ifnet *, struct mbuf *,
335 		    struct ether_header *, int, struct llc *);
336 
337 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
338 #define	VLANTAGOF(_m)	\
339     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
340 
341 static struct bstp_cb_ops bridge_ops = {
342 	.bcb_state = bridge_state_change,
343 	.bcb_rtage = bridge_rtable_expire
344 };
345 
346 SYSCTL_DECL(_net_link);
347 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
348 
349 static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */
350 static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */
351 static int pfil_member = 1; /* run pfil hooks on the member interface */
352 static int pfil_ipfw = 0;   /* layer2 filter with ipfw */
353 static int pfil_ipfw_arp = 0;   /* layer2 filter with ipfw */
354 static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for
355                                    locally destined packets */
356 static int log_stp   = 0;   /* log STP state changes */
357 static int bridge_inherit_mac = 0;   /* share MAC with first bridge member */
358 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
359     &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
360 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
361     &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
362 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
363     &pfil_bridge, 0, "Packet filter on the bridge interface");
364 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
365     &pfil_member, 0, "Packet filter on the member interface");
366 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
367     &pfil_local_phys, 0,
368     "Packet filter on the physical interface for locally destined packets");
369 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
370     &log_stp, 0, "Log STP state changes");
371 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, CTLFLAG_RW,
372     &bridge_inherit_mac, 0,
373     "Inherit MAC address from the first bridge member");
374 
375 struct bridge_control {
376 	int	(*bc_func)(struct bridge_softc *, void *);
377 	int	bc_argsize;
378 	int	bc_flags;
379 };
380 
381 #define	BC_F_COPYIN		0x01	/* copy arguments in */
382 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
383 #define	BC_F_SUSER		0x04	/* do super-user check */
384 
385 const struct bridge_control bridge_control_table[] = {
386 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
387 	  BC_F_COPYIN|BC_F_SUSER },
388 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
389 	  BC_F_COPYIN|BC_F_SUSER },
390 
391 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
392 	  BC_F_COPYIN|BC_F_COPYOUT },
393 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
394 	  BC_F_COPYIN|BC_F_SUSER },
395 
396 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
397 	  BC_F_COPYIN|BC_F_SUSER },
398 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
399 	  BC_F_COPYOUT },
400 
401 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
402 	  BC_F_COPYIN|BC_F_COPYOUT },
403 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
404 	  BC_F_COPYIN|BC_F_COPYOUT },
405 
406 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
407 	  BC_F_COPYIN|BC_F_SUSER },
408 
409 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
410 	  BC_F_COPYIN|BC_F_SUSER },
411 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
412 	  BC_F_COPYOUT },
413 
414 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
415 	  BC_F_COPYIN|BC_F_SUSER },
416 
417 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
418 	  BC_F_COPYIN|BC_F_SUSER },
419 
420 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
421 	  BC_F_COPYOUT },
422 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
423 	  BC_F_COPYIN|BC_F_SUSER },
424 
425 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
426 	  BC_F_COPYOUT },
427 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
428 	  BC_F_COPYIN|BC_F_SUSER },
429 
430 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
431 	  BC_F_COPYOUT },
432 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
433 	  BC_F_COPYIN|BC_F_SUSER },
434 
435 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
436 	  BC_F_COPYOUT },
437 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
438 	  BC_F_COPYIN|BC_F_SUSER },
439 
440 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
441 	  BC_F_COPYIN|BC_F_SUSER },
442 
443 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
444 	  BC_F_COPYIN|BC_F_SUSER },
445 
446 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
447 	  BC_F_COPYIN|BC_F_SUSER },
448 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
449 	  BC_F_COPYIN|BC_F_SUSER },
450 
451 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
452 	  BC_F_COPYOUT },
453 
454 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
455 	  BC_F_COPYOUT },
456 
457 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
458 	  BC_F_COPYIN|BC_F_COPYOUT },
459 
460 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
461 	  BC_F_COPYIN|BC_F_SUSER },
462 
463 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
464 	  BC_F_COPYIN|BC_F_SUSER },
465 
466 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
467 	  BC_F_COPYIN|BC_F_SUSER },
468 
469 };
470 const int bridge_control_table_size =
471     sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
472 
473 LIST_HEAD(, bridge_softc) bridge_list;
474 
475 IFC_SIMPLE_DECLARE(bridge, 0);
476 
477 static int
478 bridge_modevent(module_t mod, int type, void *data)
479 {
480 
481 	switch (type) {
482 	case MOD_LOAD:
483 		mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
484 		if_clone_attach(&bridge_cloner);
485 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
486 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
487 		    UMA_ALIGN_PTR, 0);
488 		LIST_INIT(&bridge_list);
489 		bridge_input_p = bridge_input;
490 		bridge_output_p = bridge_output;
491 		bridge_dn_p = bridge_dummynet;
492 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
493 		    ifnet_departure_event, bridge_ifdetach, NULL,
494 		    EVENTHANDLER_PRI_ANY);
495 		break;
496 	case MOD_UNLOAD:
497 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
498 		    bridge_detach_cookie);
499 		if_clone_detach(&bridge_cloner);
500 		uma_zdestroy(bridge_rtnode_zone);
501 		bridge_input_p = NULL;
502 		bridge_output_p = NULL;
503 		bridge_dn_p = NULL;
504 		mtx_destroy(&bridge_list_mtx);
505 		break;
506 	default:
507 		return (EOPNOTSUPP);
508 	}
509 	return (0);
510 }
511 
512 static moduledata_t bridge_mod = {
513 	"if_bridge",
514 	bridge_modevent,
515 	0
516 };
517 
518 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
519 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
520 
521 /*
522  * handler for net.link.bridge.pfil_ipfw
523  */
524 static int
525 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
526 {
527 	int enable = pfil_ipfw;
528 	int error;
529 
530 	error = sysctl_handle_int(oidp, &enable, 0, req);
531 	enable = (enable) ? 1 : 0;
532 
533 	if (enable != pfil_ipfw) {
534 		pfil_ipfw = enable;
535 
536 		/*
537 		 * Disable pfil so that ipfw doesnt run twice, if the user
538 		 * really wants both then they can re-enable pfil_bridge and/or
539 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
540 		 * layer2 type.
541 		 */
542 		if (pfil_ipfw) {
543 			pfil_onlyip = 0;
544 			pfil_bridge = 0;
545 			pfil_member = 0;
546 		}
547 	}
548 
549 	return (error);
550 }
551 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
552 	    &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
553 
554 /*
555  * bridge_clone_create:
556  *
557  *	Create a new bridge instance.
558  */
559 static int
560 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
561 {
562 	struct bridge_softc *sc, *sc2;
563 	struct ifnet *bifp, *ifp;
564 	int fb, retry;
565 	unsigned long hostid;
566 
567 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
568 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
569 	if (ifp == NULL) {
570 		free(sc, M_DEVBUF);
571 		return (ENOSPC);
572 	}
573 
574 	BRIDGE_LOCK_INIT(sc);
575 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
576 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
577 
578 	/* Initialize our routing table. */
579 	bridge_rtable_init(sc);
580 
581 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
582 
583 	LIST_INIT(&sc->sc_iflist);
584 	LIST_INIT(&sc->sc_spanlist);
585 
586 	ifp->if_softc = sc;
587 	if_initname(ifp, ifc->ifc_name, unit);
588 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
589 	ifp->if_ioctl = bridge_ioctl;
590 	ifp->if_start = bridge_start;
591 	ifp->if_init = bridge_init;
592 	ifp->if_type = IFT_BRIDGE;
593 	IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
594 	ifp->if_snd.ifq_drv_maxlen = ifqmaxlen;
595 	IFQ_SET_READY(&ifp->if_snd);
596 
597 	/*
598 	 * Generate an ethernet address with a locally administered address.
599 	 *
600 	 * Since we are using random ethernet addresses for the bridge, it is
601 	 * possible that we might have address collisions, so make sure that
602 	 * this hardware address isn't already in use on another bridge.
603 	 * The first try uses the hostid and falls back to arc4rand().
604 	 */
605 	fb = 0;
606 	getcredhostid(curthread->td_ucred, &hostid);
607 	for (retry = 1; retry != 0;) {
608 		if (fb || hostid == 0) {
609 			arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
610 			sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
611 			sc->sc_defaddr[0] |= 2;	/* set the LAA bit */
612 		} else {
613 			sc->sc_defaddr[0] = 0x2;
614 			sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
615 			sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
616 			sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
617 			sc->sc_defaddr[4] =  hostid        & 0xff;
618 			sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
619 		}
620 
621 		fb = 1;
622 		retry = 0;
623 		mtx_lock(&bridge_list_mtx);
624 		LIST_FOREACH(sc2, &bridge_list, sc_list) {
625 			bifp = sc2->sc_ifp;
626 			if (memcmp(sc->sc_defaddr,
627 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0)
628 				retry = 1;
629 		}
630 		mtx_unlock(&bridge_list_mtx);
631 	}
632 
633 	bstp_attach(&sc->sc_stp, &bridge_ops);
634 	ether_ifattach(ifp, sc->sc_defaddr);
635 	/* Now undo some of the damage... */
636 	ifp->if_baudrate = 0;
637 	ifp->if_type = IFT_BRIDGE;
638 
639 	mtx_lock(&bridge_list_mtx);
640 	LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
641 	mtx_unlock(&bridge_list_mtx);
642 
643 	return (0);
644 }
645 
646 /*
647  * bridge_clone_destroy:
648  *
649  *	Destroy a bridge instance.
650  */
651 static void
652 bridge_clone_destroy(struct ifnet *ifp)
653 {
654 	struct bridge_softc *sc = ifp->if_softc;
655 	struct bridge_iflist *bif;
656 
657 	BRIDGE_LOCK(sc);
658 
659 	bridge_stop(ifp, 1);
660 	ifp->if_flags &= ~IFF_UP;
661 
662 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
663 		bridge_delete_member(sc, bif, 0);
664 
665 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
666 		bridge_delete_span(sc, bif);
667 	}
668 
669 	BRIDGE_UNLOCK(sc);
670 
671 	callout_drain(&sc->sc_brcallout);
672 
673 	mtx_lock(&bridge_list_mtx);
674 	LIST_REMOVE(sc, sc_list);
675 	mtx_unlock(&bridge_list_mtx);
676 
677 	bstp_detach(&sc->sc_stp);
678 	ether_ifdetach(ifp);
679 	if_free(ifp);
680 
681 	/* Tear down the routing table. */
682 	bridge_rtable_fini(sc);
683 
684 	BRIDGE_LOCK_DESTROY(sc);
685 	free(sc, M_DEVBUF);
686 }
687 
688 /*
689  * bridge_ioctl:
690  *
691  *	Handle a control request from the operator.
692  */
693 static int
694 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
695 {
696 	struct bridge_softc *sc = ifp->if_softc;
697 	struct ifreq *ifr = (struct ifreq *)data;
698 	struct bridge_iflist *bif;
699 	struct thread *td = curthread;
700 	union {
701 		struct ifbreq ifbreq;
702 		struct ifbifconf ifbifconf;
703 		struct ifbareq ifbareq;
704 		struct ifbaconf ifbaconf;
705 		struct ifbrparam ifbrparam;
706 		struct ifbropreq ifbropreq;
707 	} args;
708 	struct ifdrv *ifd = (struct ifdrv *) data;
709 	const struct bridge_control *bc;
710 	int error = 0;
711 
712 	switch (cmd) {
713 
714 	case SIOCADDMULTI:
715 	case SIOCDELMULTI:
716 		break;
717 
718 	case SIOCGDRVSPEC:
719 	case SIOCSDRVSPEC:
720 		if (ifd->ifd_cmd >= bridge_control_table_size) {
721 			error = EINVAL;
722 			break;
723 		}
724 		bc = &bridge_control_table[ifd->ifd_cmd];
725 
726 		if (cmd == SIOCGDRVSPEC &&
727 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
728 			error = EINVAL;
729 			break;
730 		}
731 		else if (cmd == SIOCSDRVSPEC &&
732 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
733 			error = EINVAL;
734 			break;
735 		}
736 
737 		if (bc->bc_flags & BC_F_SUSER) {
738 			error = priv_check(td, PRIV_NET_BRIDGE);
739 			if (error)
740 				break;
741 		}
742 
743 		if (ifd->ifd_len != bc->bc_argsize ||
744 		    ifd->ifd_len > sizeof(args)) {
745 			error = EINVAL;
746 			break;
747 		}
748 
749 		bzero(&args, sizeof(args));
750 		if (bc->bc_flags & BC_F_COPYIN) {
751 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
752 			if (error)
753 				break;
754 		}
755 
756 		BRIDGE_LOCK(sc);
757 		error = (*bc->bc_func)(sc, &args);
758 		BRIDGE_UNLOCK(sc);
759 		if (error)
760 			break;
761 
762 		if (bc->bc_flags & BC_F_COPYOUT)
763 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
764 
765 		break;
766 
767 	case SIOCSIFFLAGS:
768 		if (!(ifp->if_flags & IFF_UP) &&
769 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
770 			/*
771 			 * If interface is marked down and it is running,
772 			 * then stop and disable it.
773 			 */
774 			BRIDGE_LOCK(sc);
775 			bridge_stop(ifp, 1);
776 			BRIDGE_UNLOCK(sc);
777 		} else if ((ifp->if_flags & IFF_UP) &&
778 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
779 			/*
780 			 * If interface is marked up and it is stopped, then
781 			 * start it.
782 			 */
783 			(*ifp->if_init)(sc);
784 		}
785 		break;
786 
787 	case SIOCSIFMTU:
788 		if (ifr->ifr_mtu < 576) {
789 			error = EINVAL;
790 			break;
791 		}
792 		if (LIST_EMPTY(&sc->sc_iflist)) {
793 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
794 			break;
795 		}
796 		BRIDGE_LOCK(sc);
797 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
798 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
799 				log(LOG_NOTICE, "%s: invalid MTU: %lu(%s)"
800 				    " != %d\n", sc->sc_ifp->if_xname,
801 				    bif->bif_ifp->if_mtu,
802 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
803 				error = EINVAL;
804 				break;
805 			}
806 		}
807 		if (!error)
808 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
809 		BRIDGE_UNLOCK(sc);
810 		break;
811 	default:
812 		/*
813 		 * drop the lock as ether_ioctl() will call bridge_start() and
814 		 * cause the lock to be recursed.
815 		 */
816 		error = ether_ioctl(ifp, cmd, data);
817 		break;
818 	}
819 
820 	return (error);
821 }
822 
823 /*
824  * bridge_mutecaps:
825  *
826  *	Clear or restore unwanted capabilities on the member interface
827  */
828 static void
829 bridge_mutecaps(struct bridge_softc *sc)
830 {
831 	struct bridge_iflist *bif;
832 	int enabled, mask;
833 
834 	/* Initial bitmask of capabilities to test */
835 	mask = BRIDGE_IFCAPS_MASK;
836 
837 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
838 		/* Every member must support it or its disabled */
839 		mask &= bif->bif_savedcaps;
840 	}
841 
842 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
843 		enabled = bif->bif_ifp->if_capenable;
844 		enabled &= ~BRIDGE_IFCAPS_STRIP;
845 		/* strip off mask bits and enable them again if allowed */
846 		enabled &= ~BRIDGE_IFCAPS_MASK;
847 		enabled |= mask;
848 		bridge_set_ifcap(sc, bif, enabled);
849 	}
850 
851 }
852 
853 static void
854 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
855 {
856 	struct ifnet *ifp = bif->bif_ifp;
857 	struct ifreq ifr;
858 	int error;
859 
860 	bzero(&ifr, sizeof(ifr));
861 	ifr.ifr_reqcap = set;
862 
863 	if (ifp->if_capenable != set) {
864 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
865 		if (error)
866 			if_printf(sc->sc_ifp,
867 			    "error setting interface capabilities on %s\n",
868 			    ifp->if_xname);
869 	}
870 }
871 
872 /*
873  * bridge_lookup_member:
874  *
875  *	Lookup a bridge member interface.
876  */
877 static struct bridge_iflist *
878 bridge_lookup_member(struct bridge_softc *sc, const char *name)
879 {
880 	struct bridge_iflist *bif;
881 	struct ifnet *ifp;
882 
883 	BRIDGE_LOCK_ASSERT(sc);
884 
885 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
886 		ifp = bif->bif_ifp;
887 		if (strcmp(ifp->if_xname, name) == 0)
888 			return (bif);
889 	}
890 
891 	return (NULL);
892 }
893 
894 /*
895  * bridge_lookup_member_if:
896  *
897  *	Lookup a bridge member interface by ifnet*.
898  */
899 static struct bridge_iflist *
900 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
901 {
902 	struct bridge_iflist *bif;
903 
904 	BRIDGE_LOCK_ASSERT(sc);
905 
906 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
907 		if (bif->bif_ifp == member_ifp)
908 			return (bif);
909 	}
910 
911 	return (NULL);
912 }
913 
914 /*
915  * bridge_delete_member:
916  *
917  *	Delete the specified member interface.
918  */
919 static void
920 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
921     int gone)
922 {
923 	struct ifnet *ifs = bif->bif_ifp;
924 	struct ifnet *fif = NULL;
925 
926 	BRIDGE_LOCK_ASSERT(sc);
927 
928 	if (bif->bif_flags & IFBIF_STP)
929 		bstp_disable(&bif->bif_stp);
930 
931 	ifs->if_bridge = NULL;
932 	BRIDGE_XLOCK(sc);
933 	LIST_REMOVE(bif, bif_next);
934 	BRIDGE_XDROP(sc);
935 
936 	/*
937 	 * If removing the interface that gave the bridge its mac address, set
938 	 * the mac address of the bridge to the address of the next member, or
939 	 * to its default address if no members are left.
940 	 */
941 	if (bridge_inherit_mac && sc->sc_ifaddr == ifs) {
942 		if (LIST_EMPTY(&sc->sc_iflist)) {
943 			bcopy(sc->sc_defaddr,
944 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
945 			sc->sc_ifaddr = NULL;
946 		} else {
947 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
948 			bcopy(IF_LLADDR(fif),
949 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
950 			sc->sc_ifaddr = fif;
951 		}
952 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
953 	}
954 
955 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
956 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
957 	KASSERT(bif->bif_addrcnt == 0,
958 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
959 
960 	BRIDGE_UNLOCK(sc);
961 	if (!gone) {
962 		switch (ifs->if_type) {
963 		case IFT_ETHER:
964 		case IFT_L2VLAN:
965 			/*
966 			 * Take the interface out of promiscuous mode.
967 			 */
968 			(void) ifpromisc(ifs, 0);
969 			break;
970 
971 		case IFT_GIF:
972 			break;
973 
974 		default:
975 #ifdef DIAGNOSTIC
976 			panic("bridge_delete_member: impossible");
977 #endif
978 			break;
979 		}
980 		/* reneable any interface capabilities */
981 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
982 	}
983 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
984 	BRIDGE_LOCK(sc);
985 	free(bif, M_DEVBUF);
986 }
987 
988 /*
989  * bridge_delete_span:
990  *
991  *	Delete the specified span interface.
992  */
993 static void
994 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
995 {
996 	BRIDGE_LOCK_ASSERT(sc);
997 
998 	KASSERT(bif->bif_ifp->if_bridge == NULL,
999 	    ("%s: not a span interface", __func__));
1000 
1001 	LIST_REMOVE(bif, bif_next);
1002 	free(bif, M_DEVBUF);
1003 }
1004 
1005 static int
1006 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1007 {
1008 	struct ifbreq *req = arg;
1009 	struct bridge_iflist *bif = NULL;
1010 	struct ifnet *ifs;
1011 	int error = 0;
1012 
1013 	ifs = ifunit(req->ifbr_ifsname);
1014 	if (ifs == NULL)
1015 		return (ENOENT);
1016 	if (ifs->if_ioctl == NULL)	/* must be supported */
1017 		return (EINVAL);
1018 
1019 	/* If it's in the span list, it can't be a member. */
1020 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1021 		if (ifs == bif->bif_ifp)
1022 			return (EBUSY);
1023 
1024 	if (ifs->if_bridge == sc)
1025 		return (EEXIST);
1026 
1027 	if (ifs->if_bridge != NULL)
1028 		return (EBUSY);
1029 
1030 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1031 	if (bif == NULL)
1032 		return (ENOMEM);
1033 
1034 	bif->bif_ifp = ifs;
1035 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1036 	bif->bif_savedcaps = ifs->if_capenable;
1037 
1038 	switch (ifs->if_type) {
1039 	case IFT_ETHER:
1040 	case IFT_L2VLAN:
1041 	case IFT_GIF:
1042 		/* permitted interface types */
1043 		break;
1044 	default:
1045 		error = EINVAL;
1046 		goto out;
1047 	}
1048 
1049 	/* Allow the first Ethernet member to define the MTU */
1050 	if (LIST_EMPTY(&sc->sc_iflist))
1051 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1052 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1053 		if_printf(sc->sc_ifp, "invalid MTU: %lu(%s) != %lu\n",
1054 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1055 		error = EINVAL;
1056 		goto out;
1057 	}
1058 
1059 	/*
1060 	 * Assign the interface's MAC address to the bridge if it's the first
1061 	 * member and the MAC address of the bridge has not been changed from
1062 	 * the default randomly generated one.
1063 	 */
1064 	if (bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1065 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1066 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1067 		sc->sc_ifaddr = ifs;
1068 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1069 	}
1070 
1071 	ifs->if_bridge = sc;
1072 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1073 	/*
1074 	 * XXX: XLOCK HERE!?!
1075 	 *
1076 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1077 	 */
1078 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1079 
1080 	/* Set interface capabilities to the intersection set of all members */
1081 	bridge_mutecaps(sc);
1082 
1083 	switch (ifs->if_type) {
1084 	case IFT_ETHER:
1085 	case IFT_L2VLAN:
1086 		/*
1087 		 * Place the interface into promiscuous mode.
1088 		 */
1089 		BRIDGE_UNLOCK(sc);
1090 		error = ifpromisc(ifs, 1);
1091 		BRIDGE_LOCK(sc);
1092 		break;
1093 	}
1094 	if (error)
1095 		bridge_delete_member(sc, bif, 0);
1096 out:
1097 	if (error) {
1098 		if (bif != NULL)
1099 			free(bif, M_DEVBUF);
1100 	}
1101 	return (error);
1102 }
1103 
1104 static int
1105 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1106 {
1107 	struct ifbreq *req = arg;
1108 	struct bridge_iflist *bif;
1109 
1110 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1111 	if (bif == NULL)
1112 		return (ENOENT);
1113 
1114 	bridge_delete_member(sc, bif, 0);
1115 
1116 	return (0);
1117 }
1118 
1119 static int
1120 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1121 {
1122 	struct ifbreq *req = arg;
1123 	struct bridge_iflist *bif;
1124 	struct bstp_port *bp;
1125 
1126 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1127 	if (bif == NULL)
1128 		return (ENOENT);
1129 
1130 	bp = &bif->bif_stp;
1131 	req->ifbr_ifsflags = bif->bif_flags;
1132 	req->ifbr_state = bp->bp_state;
1133 	req->ifbr_priority = bp->bp_priority;
1134 	req->ifbr_path_cost = bp->bp_path_cost;
1135 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1136 	req->ifbr_proto = bp->bp_protover;
1137 	req->ifbr_role = bp->bp_role;
1138 	req->ifbr_stpflags = bp->bp_flags;
1139 	req->ifbr_addrcnt = bif->bif_addrcnt;
1140 	req->ifbr_addrmax = bif->bif_addrmax;
1141 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1142 
1143 	/* Copy STP state options as flags */
1144 	if (bp->bp_operedge)
1145 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1146 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1147 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1148 	if (bp->bp_ptp_link)
1149 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1150 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1151 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1152 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1153 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1154 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1155 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1156 	return (0);
1157 }
1158 
1159 static int
1160 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1161 {
1162 	struct ifbreq *req = arg;
1163 	struct bridge_iflist *bif;
1164 	struct bstp_port *bp;
1165 	int error;
1166 
1167 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1168 	if (bif == NULL)
1169 		return (ENOENT);
1170 	bp = &bif->bif_stp;
1171 
1172 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1173 		/* SPAN is readonly */
1174 		return (EINVAL);
1175 
1176 	if (req->ifbr_ifsflags & IFBIF_STP) {
1177 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1178 			error = bstp_enable(&bif->bif_stp);
1179 			if (error)
1180 				return (error);
1181 		}
1182 	} else {
1183 		if ((bif->bif_flags & IFBIF_STP) != 0)
1184 			bstp_disable(&bif->bif_stp);
1185 	}
1186 
1187 	/* Pass on STP flags */
1188 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1189 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1190 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1191 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1192 
1193 	/* Save the bits relating to the bridge */
1194 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1195 
1196 	return (0);
1197 }
1198 
1199 static int
1200 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1201 {
1202 	struct ifbrparam *param = arg;
1203 
1204 	sc->sc_brtmax = param->ifbrp_csize;
1205 	bridge_rttrim(sc);
1206 
1207 	return (0);
1208 }
1209 
1210 static int
1211 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1212 {
1213 	struct ifbrparam *param = arg;
1214 
1215 	param->ifbrp_csize = sc->sc_brtmax;
1216 
1217 	return (0);
1218 }
1219 
1220 static int
1221 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1222 {
1223 	struct ifbifconf *bifc = arg;
1224 	struct bridge_iflist *bif;
1225 	struct ifbreq breq;
1226 	char *buf, *outbuf;
1227 	int count, buflen, len, error = 0;
1228 
1229 	count = 0;
1230 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1231 		count++;
1232 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1233 		count++;
1234 
1235 	buflen = sizeof(breq) * count;
1236 	if (bifc->ifbic_len == 0) {
1237 		bifc->ifbic_len = buflen;
1238 		return (0);
1239 	}
1240 	BRIDGE_UNLOCK(sc);
1241 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1242 	BRIDGE_LOCK(sc);
1243 
1244 	count = 0;
1245 	buf = outbuf;
1246 	len = min(bifc->ifbic_len, buflen);
1247 	bzero(&breq, sizeof(breq));
1248 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1249 		if (len < sizeof(breq))
1250 			break;
1251 
1252 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1253 		    sizeof(breq.ifbr_ifsname));
1254 		/* Fill in the ifbreq structure */
1255 		error = bridge_ioctl_gifflags(sc, &breq);
1256 		if (error)
1257 			break;
1258 		memcpy(buf, &breq, sizeof(breq));
1259 		count++;
1260 		buf += sizeof(breq);
1261 		len -= sizeof(breq);
1262 	}
1263 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1264 		if (len < sizeof(breq))
1265 			break;
1266 
1267 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1268 		    sizeof(breq.ifbr_ifsname));
1269 		breq.ifbr_ifsflags = bif->bif_flags;
1270 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1271 		memcpy(buf, &breq, sizeof(breq));
1272 		count++;
1273 		buf += sizeof(breq);
1274 		len -= sizeof(breq);
1275 	}
1276 
1277 	BRIDGE_UNLOCK(sc);
1278 	bifc->ifbic_len = sizeof(breq) * count;
1279 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1280 	BRIDGE_LOCK(sc);
1281 	free(outbuf, M_TEMP);
1282 	return (error);
1283 }
1284 
1285 static int
1286 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1287 {
1288 	struct ifbaconf *bac = arg;
1289 	struct bridge_rtnode *brt;
1290 	struct ifbareq bareq;
1291 	char *buf, *outbuf;
1292 	int count, buflen, len, error = 0;
1293 
1294 	if (bac->ifbac_len == 0)
1295 		return (0);
1296 
1297 	count = 0;
1298 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1299 		count++;
1300 	buflen = sizeof(bareq) * count;
1301 
1302 	BRIDGE_UNLOCK(sc);
1303 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1304 	BRIDGE_LOCK(sc);
1305 
1306 	count = 0;
1307 	buf = outbuf;
1308 	len = min(bac->ifbac_len, buflen);
1309 	bzero(&bareq, sizeof(bareq));
1310 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1311 		if (len < sizeof(bareq))
1312 			goto out;
1313 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1314 		    sizeof(bareq.ifba_ifsname));
1315 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1316 		bareq.ifba_vlan = brt->brt_vlan;
1317 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1318 				time_uptime < brt->brt_expire)
1319 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1320 		else
1321 			bareq.ifba_expire = 0;
1322 		bareq.ifba_flags = brt->brt_flags;
1323 
1324 		memcpy(buf, &bareq, sizeof(bareq));
1325 		count++;
1326 		buf += sizeof(bareq);
1327 		len -= sizeof(bareq);
1328 	}
1329 out:
1330 	BRIDGE_UNLOCK(sc);
1331 	bac->ifbac_len = sizeof(bareq) * count;
1332 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1333 	BRIDGE_LOCK(sc);
1334 	free(outbuf, M_TEMP);
1335 	return (error);
1336 }
1337 
1338 static int
1339 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1340 {
1341 	struct ifbareq *req = arg;
1342 	struct bridge_iflist *bif;
1343 	int error;
1344 
1345 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1346 	if (bif == NULL)
1347 		return (ENOENT);
1348 
1349 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1350 	    req->ifba_flags);
1351 
1352 	return (error);
1353 }
1354 
1355 static int
1356 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1357 {
1358 	struct ifbrparam *param = arg;
1359 
1360 	sc->sc_brttimeout = param->ifbrp_ctime;
1361 	return (0);
1362 }
1363 
1364 static int
1365 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1366 {
1367 	struct ifbrparam *param = arg;
1368 
1369 	param->ifbrp_ctime = sc->sc_brttimeout;
1370 	return (0);
1371 }
1372 
1373 static int
1374 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1375 {
1376 	struct ifbareq *req = arg;
1377 
1378 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1379 }
1380 
1381 static int
1382 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1383 {
1384 	struct ifbreq *req = arg;
1385 
1386 	bridge_rtflush(sc, req->ifbr_ifsflags);
1387 	return (0);
1388 }
1389 
1390 static int
1391 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1392 {
1393 	struct ifbrparam *param = arg;
1394 	struct bstp_state *bs = &sc->sc_stp;
1395 
1396 	param->ifbrp_prio = bs->bs_bridge_priority;
1397 	return (0);
1398 }
1399 
1400 static int
1401 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1402 {
1403 	struct ifbrparam *param = arg;
1404 
1405 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1406 }
1407 
1408 static int
1409 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1410 {
1411 	struct ifbrparam *param = arg;
1412 	struct bstp_state *bs = &sc->sc_stp;
1413 
1414 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1415 	return (0);
1416 }
1417 
1418 static int
1419 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1420 {
1421 	struct ifbrparam *param = arg;
1422 
1423 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1424 }
1425 
1426 static int
1427 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1428 {
1429 	struct ifbrparam *param = arg;
1430 	struct bstp_state *bs = &sc->sc_stp;
1431 
1432 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1433 	return (0);
1434 }
1435 
1436 static int
1437 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1438 {
1439 	struct ifbrparam *param = arg;
1440 
1441 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1442 }
1443 
1444 static int
1445 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1446 {
1447 	struct ifbrparam *param = arg;
1448 	struct bstp_state *bs = &sc->sc_stp;
1449 
1450 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1451 	return (0);
1452 }
1453 
1454 static int
1455 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1456 {
1457 	struct ifbrparam *param = arg;
1458 
1459 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1460 }
1461 
1462 static int
1463 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1464 {
1465 	struct ifbreq *req = arg;
1466 	struct bridge_iflist *bif;
1467 
1468 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1469 	if (bif == NULL)
1470 		return (ENOENT);
1471 
1472 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1473 }
1474 
1475 static int
1476 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1477 {
1478 	struct ifbreq *req = arg;
1479 	struct bridge_iflist *bif;
1480 
1481 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1482 	if (bif == NULL)
1483 		return (ENOENT);
1484 
1485 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1486 }
1487 
1488 static int
1489 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1490 {
1491 	struct ifbreq *req = arg;
1492 	struct bridge_iflist *bif;
1493 
1494 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1495 	if (bif == NULL)
1496 		return (ENOENT);
1497 
1498 	bif->bif_addrmax = req->ifbr_addrmax;
1499 	return (0);
1500 }
1501 
1502 static int
1503 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1504 {
1505 	struct ifbreq *req = arg;
1506 	struct bridge_iflist *bif = NULL;
1507 	struct ifnet *ifs;
1508 
1509 	ifs = ifunit(req->ifbr_ifsname);
1510 	if (ifs == NULL)
1511 		return (ENOENT);
1512 
1513 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1514 		if (ifs == bif->bif_ifp)
1515 			return (EBUSY);
1516 
1517 	if (ifs->if_bridge != NULL)
1518 		return (EBUSY);
1519 
1520 	switch (ifs->if_type) {
1521 		case IFT_ETHER:
1522 		case IFT_GIF:
1523 		case IFT_L2VLAN:
1524 			break;
1525 		default:
1526 			return (EINVAL);
1527 	}
1528 
1529 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1530 	if (bif == NULL)
1531 		return (ENOMEM);
1532 
1533 	bif->bif_ifp = ifs;
1534 	bif->bif_flags = IFBIF_SPAN;
1535 
1536 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1537 
1538 	return (0);
1539 }
1540 
1541 static int
1542 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1543 {
1544 	struct ifbreq *req = arg;
1545 	struct bridge_iflist *bif;
1546 	struct ifnet *ifs;
1547 
1548 	ifs = ifunit(req->ifbr_ifsname);
1549 	if (ifs == NULL)
1550 		return (ENOENT);
1551 
1552 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1553 		if (ifs == bif->bif_ifp)
1554 			break;
1555 
1556 	if (bif == NULL)
1557 		return (ENOENT);
1558 
1559 	bridge_delete_span(sc, bif);
1560 
1561 	return (0);
1562 }
1563 
1564 static int
1565 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1566 {
1567 	struct ifbropreq *req = arg;
1568 	struct bstp_state *bs = &sc->sc_stp;
1569 	struct bstp_port *root_port;
1570 
1571 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1572 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1573 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1574 
1575 	root_port = bs->bs_root_port;
1576 	if (root_port == NULL)
1577 		req->ifbop_root_port = 0;
1578 	else
1579 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1580 
1581 	req->ifbop_holdcount = bs->bs_txholdcount;
1582 	req->ifbop_priority = bs->bs_bridge_priority;
1583 	req->ifbop_protocol = bs->bs_protover;
1584 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1585 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1586 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1587 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1588 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1589 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1590 
1591 	return (0);
1592 }
1593 
1594 static int
1595 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1596 {
1597 	struct ifbrparam *param = arg;
1598 
1599 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1600 	return (0);
1601 }
1602 
1603 static int
1604 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1605 {
1606 	struct ifbpstpconf *bifstp = arg;
1607 	struct bridge_iflist *bif;
1608 	struct bstp_port *bp;
1609 	struct ifbpstpreq bpreq;
1610 	char *buf, *outbuf;
1611 	int count, buflen, len, error = 0;
1612 
1613 	count = 0;
1614 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1615 		if ((bif->bif_flags & IFBIF_STP) != 0)
1616 			count++;
1617 	}
1618 
1619 	buflen = sizeof(bpreq) * count;
1620 	if (bifstp->ifbpstp_len == 0) {
1621 		bifstp->ifbpstp_len = buflen;
1622 		return (0);
1623 	}
1624 
1625 	BRIDGE_UNLOCK(sc);
1626 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1627 	BRIDGE_LOCK(sc);
1628 
1629 	count = 0;
1630 	buf = outbuf;
1631 	len = min(bifstp->ifbpstp_len, buflen);
1632 	bzero(&bpreq, sizeof(bpreq));
1633 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1634 		if (len < sizeof(bpreq))
1635 			break;
1636 
1637 		if ((bif->bif_flags & IFBIF_STP) == 0)
1638 			continue;
1639 
1640 		bp = &bif->bif_stp;
1641 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1642 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1643 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1644 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1645 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1646 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1647 
1648 		memcpy(buf, &bpreq, sizeof(bpreq));
1649 		count++;
1650 		buf += sizeof(bpreq);
1651 		len -= sizeof(bpreq);
1652 	}
1653 
1654 	BRIDGE_UNLOCK(sc);
1655 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1656 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1657 	BRIDGE_LOCK(sc);
1658 	free(outbuf, M_TEMP);
1659 	return (error);
1660 }
1661 
1662 static int
1663 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1664 {
1665 	struct ifbrparam *param = arg;
1666 
1667 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1668 }
1669 
1670 static int
1671 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1672 {
1673 	struct ifbrparam *param = arg;
1674 
1675 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1676 }
1677 
1678 /*
1679  * bridge_ifdetach:
1680  *
1681  *	Detach an interface from a bridge.  Called when a member
1682  *	interface is detaching.
1683  */
1684 static void
1685 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1686 {
1687 	struct bridge_softc *sc = ifp->if_bridge;
1688 	struct bridge_iflist *bif;
1689 
1690 	/* Check if the interface is a bridge member */
1691 	if (sc != NULL) {
1692 		BRIDGE_LOCK(sc);
1693 
1694 		bif = bridge_lookup_member_if(sc, ifp);
1695 		if (bif != NULL)
1696 			bridge_delete_member(sc, bif, 1);
1697 
1698 		BRIDGE_UNLOCK(sc);
1699 		return;
1700 	}
1701 
1702 	/* Check if the interface is a span port */
1703 	mtx_lock(&bridge_list_mtx);
1704 	LIST_FOREACH(sc, &bridge_list, sc_list) {
1705 		BRIDGE_LOCK(sc);
1706 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1707 			if (ifp == bif->bif_ifp) {
1708 				bridge_delete_span(sc, bif);
1709 				break;
1710 			}
1711 
1712 		BRIDGE_UNLOCK(sc);
1713 	}
1714 	mtx_unlock(&bridge_list_mtx);
1715 }
1716 
1717 /*
1718  * bridge_init:
1719  *
1720  *	Initialize a bridge interface.
1721  */
1722 static void
1723 bridge_init(void *xsc)
1724 {
1725 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1726 	struct ifnet *ifp = sc->sc_ifp;
1727 
1728 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1729 		return;
1730 
1731 	BRIDGE_LOCK(sc);
1732 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1733 	    bridge_timer, sc);
1734 
1735 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1736 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1737 
1738 	BRIDGE_UNLOCK(sc);
1739 }
1740 
1741 /*
1742  * bridge_stop:
1743  *
1744  *	Stop the bridge interface.
1745  */
1746 static void
1747 bridge_stop(struct ifnet *ifp, int disable)
1748 {
1749 	struct bridge_softc *sc = ifp->if_softc;
1750 
1751 	BRIDGE_LOCK_ASSERT(sc);
1752 
1753 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1754 		return;
1755 
1756 	callout_stop(&sc->sc_brcallout);
1757 	bstp_stop(&sc->sc_stp);
1758 
1759 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1760 
1761 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1762 }
1763 
1764 /*
1765  * bridge_enqueue:
1766  *
1767  *	Enqueue a packet on a bridge member interface.
1768  *
1769  */
1770 static void
1771 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1772 {
1773 	int len, err = 0;
1774 	short mflags;
1775 	struct mbuf *m0;
1776 
1777 	len = m->m_pkthdr.len;
1778 	mflags = m->m_flags;
1779 
1780 	/* We may be sending a fragment so traverse the mbuf */
1781 	for (; m; m = m0) {
1782 		m0 = m->m_nextpkt;
1783 		m->m_nextpkt = NULL;
1784 
1785 		/*
1786 		 * If underlying interface can not do VLAN tag insertion itself
1787 		 * then attach a packet tag that holds it.
1788 		 */
1789 		if ((m->m_flags & M_VLANTAG) &&
1790 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1791 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1792 			if (m == NULL) {
1793 				if_printf(dst_ifp,
1794 				    "unable to prepend VLAN header\n");
1795 				dst_ifp->if_oerrors++;
1796 				continue;
1797 			}
1798 			m->m_flags &= ~M_VLANTAG;
1799 		}
1800 
1801 		if (err == 0)
1802 			dst_ifp->if_transmit(dst_ifp, m);
1803 	}
1804 
1805 	if (err == 0) {
1806 		sc->sc_ifp->if_opackets++;
1807 		sc->sc_ifp->if_obytes += len;
1808 		if (mflags & M_MCAST)
1809 			sc->sc_ifp->if_omcasts++;
1810 	}
1811 }
1812 
1813 /*
1814  * bridge_dummynet:
1815  *
1816  * 	Receive a queued packet from dummynet and pass it on to the output
1817  * 	interface.
1818  *
1819  *	The mbuf has the Ethernet header already attached.
1820  */
1821 static void
1822 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1823 {
1824 	struct bridge_softc *sc;
1825 
1826 	sc = ifp->if_bridge;
1827 
1828 	/*
1829 	 * The packet didnt originate from a member interface. This should only
1830 	 * ever happen if a member interface is removed while packets are
1831 	 * queued for it.
1832 	 */
1833 	if (sc == NULL) {
1834 		m_freem(m);
1835 		return;
1836 	}
1837 
1838 	if (PFIL_HOOKED(&V_inet_pfil_hook)
1839 #ifdef INET6
1840 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1841 #endif
1842 	    ) {
1843 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1844 			return;
1845 		if (m == NULL)
1846 			return;
1847 	}
1848 
1849 	bridge_enqueue(sc, ifp, m);
1850 }
1851 
1852 /*
1853  * bridge_output:
1854  *
1855  *	Send output from a bridge member interface.  This
1856  *	performs the bridging function for locally originated
1857  *	packets.
1858  *
1859  *	The mbuf has the Ethernet header already attached.  We must
1860  *	enqueue or free the mbuf before returning.
1861  */
1862 static int
1863 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1864     struct rtentry *rt)
1865 {
1866 	struct ether_header *eh;
1867 	struct ifnet *dst_if;
1868 	struct bridge_softc *sc;
1869 	uint16_t vlan;
1870 
1871 	if (m->m_len < ETHER_HDR_LEN) {
1872 		m = m_pullup(m, ETHER_HDR_LEN);
1873 		if (m == NULL)
1874 			return (0);
1875 	}
1876 
1877 	eh = mtod(m, struct ether_header *);
1878 	sc = ifp->if_bridge;
1879 	vlan = VLANTAGOF(m);
1880 
1881 	BRIDGE_LOCK(sc);
1882 
1883 	/*
1884 	 * If bridge is down, but the original output interface is up,
1885 	 * go ahead and send out that interface.  Otherwise, the packet
1886 	 * is dropped below.
1887 	 */
1888 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1889 		dst_if = ifp;
1890 		goto sendunicast;
1891 	}
1892 
1893 	/*
1894 	 * If the packet is a multicast, or we don't know a better way to
1895 	 * get there, send to all interfaces.
1896 	 */
1897 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
1898 		dst_if = NULL;
1899 	else
1900 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1901 	if (dst_if == NULL) {
1902 		struct bridge_iflist *bif;
1903 		struct mbuf *mc;
1904 		int error = 0, used = 0;
1905 
1906 		bridge_span(sc, m);
1907 
1908 		BRIDGE_LOCK2REF(sc, error);
1909 		if (error) {
1910 			m_freem(m);
1911 			return (0);
1912 		}
1913 
1914 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1915 			dst_if = bif->bif_ifp;
1916 
1917 			if (dst_if->if_type == IFT_GIF)
1918 				continue;
1919 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1920 				continue;
1921 
1922 			/*
1923 			 * If this is not the original output interface,
1924 			 * and the interface is participating in spanning
1925 			 * tree, make sure the port is in a state that
1926 			 * allows forwarding.
1927 			 */
1928 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
1929 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1930 				continue;
1931 
1932 			if (LIST_NEXT(bif, bif_next) == NULL) {
1933 				used = 1;
1934 				mc = m;
1935 			} else {
1936 				mc = m_copypacket(m, M_DONTWAIT);
1937 				if (mc == NULL) {
1938 					sc->sc_ifp->if_oerrors++;
1939 					continue;
1940 				}
1941 			}
1942 
1943 			bridge_enqueue(sc, dst_if, mc);
1944 		}
1945 		if (used == 0)
1946 			m_freem(m);
1947 		BRIDGE_UNREF(sc);
1948 		return (0);
1949 	}
1950 
1951 sendunicast:
1952 	/*
1953 	 * XXX Spanning tree consideration here?
1954 	 */
1955 
1956 	bridge_span(sc, m);
1957 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1958 		m_freem(m);
1959 		BRIDGE_UNLOCK(sc);
1960 		return (0);
1961 	}
1962 
1963 	BRIDGE_UNLOCK(sc);
1964 	bridge_enqueue(sc, dst_if, m);
1965 	return (0);
1966 }
1967 
1968 /*
1969  * bridge_start:
1970  *
1971  *	Start output on a bridge.
1972  *
1973  */
1974 static void
1975 bridge_start(struct ifnet *ifp)
1976 {
1977 	struct bridge_softc *sc;
1978 	struct mbuf *m;
1979 	struct ether_header *eh;
1980 	struct ifnet *dst_if;
1981 
1982 	sc = ifp->if_softc;
1983 
1984 	ifp->if_drv_flags |= IFF_DRV_OACTIVE;
1985 	for (;;) {
1986 		IFQ_DEQUEUE(&ifp->if_snd, m);
1987 		if (m == 0)
1988 			break;
1989 		ETHER_BPF_MTAP(ifp, m);
1990 
1991 		eh = mtod(m, struct ether_header *);
1992 		dst_if = NULL;
1993 
1994 		BRIDGE_LOCK(sc);
1995 		if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
1996 			dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1);
1997 		}
1998 
1999 		if (dst_if == NULL)
2000 			bridge_broadcast(sc, ifp, m, 0);
2001 		else {
2002 			BRIDGE_UNLOCK(sc);
2003 			bridge_enqueue(sc, dst_if, m);
2004 		}
2005 	}
2006 	ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
2007 }
2008 
2009 /*
2010  * bridge_forward:
2011  *
2012  *	The forwarding function of the bridge.
2013  *
2014  *	NOTE: Releases the lock on return.
2015  */
2016 static void
2017 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2018     struct mbuf *m)
2019 {
2020 	struct bridge_iflist *dbif;
2021 	struct ifnet *src_if, *dst_if, *ifp;
2022 	struct ether_header *eh;
2023 	uint16_t vlan;
2024 	uint8_t *dst;
2025 	int error;
2026 
2027 	src_if = m->m_pkthdr.rcvif;
2028 	ifp = sc->sc_ifp;
2029 
2030 	ifp->if_ipackets++;
2031 	ifp->if_ibytes += m->m_pkthdr.len;
2032 	vlan = VLANTAGOF(m);
2033 
2034 	if ((sbif->bif_flags & IFBIF_STP) &&
2035 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2036 		goto drop;
2037 
2038 	eh = mtod(m, struct ether_header *);
2039 	dst = eh->ether_dhost;
2040 
2041 	/* If the interface is learning, record the address. */
2042 	if (sbif->bif_flags & IFBIF_LEARNING) {
2043 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2044 		    sbif, 0, IFBAF_DYNAMIC);
2045 		/*
2046 		 * If the interface has addresses limits then deny any source
2047 		 * that is not in the cache.
2048 		 */
2049 		if (error && sbif->bif_addrmax)
2050 			goto drop;
2051 	}
2052 
2053 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2054 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2055 		goto drop;
2056 
2057 	/*
2058 	 * At this point, the port either doesn't participate
2059 	 * in spanning tree or it is in the forwarding state.
2060 	 */
2061 
2062 	/*
2063 	 * If the packet is unicast, destined for someone on
2064 	 * "this" side of the bridge, drop it.
2065 	 */
2066 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2067 		dst_if = bridge_rtlookup(sc, dst, vlan);
2068 		if (src_if == dst_if)
2069 			goto drop;
2070 	} else {
2071 		/*
2072 		 * Check if its a reserved multicast address, any address
2073 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2074 		 * bridge.
2075 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2076 		 */
2077 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2078 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2079 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2080 			goto drop;
2081 
2082 		/* ...forward it to all interfaces. */
2083 		ifp->if_imcasts++;
2084 		dst_if = NULL;
2085 	}
2086 
2087 	/*
2088 	 * If we have a destination interface which is a member of our bridge,
2089 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2090 	 * For broadcast or multicast packets, don't bother because it will
2091 	 * be reinjected into ether_input. We do this before we pass the packets
2092 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2093 	 * drop the packet, or possibly modify it, making it difficult to debug
2094 	 * firewall issues on the bridge.
2095 	 */
2096 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2097 		ETHER_BPF_MTAP(ifp, m);
2098 
2099 	/* run the packet filter */
2100 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2101 #ifdef INET6
2102 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2103 #endif
2104 	    ) {
2105 		BRIDGE_UNLOCK(sc);
2106 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2107 			return;
2108 		if (m == NULL)
2109 			return;
2110 		BRIDGE_LOCK(sc);
2111 	}
2112 
2113 	if (dst_if == NULL) {
2114 		bridge_broadcast(sc, src_if, m, 1);
2115 		return;
2116 	}
2117 
2118 	/*
2119 	 * At this point, we're dealing with a unicast frame
2120 	 * going to a different interface.
2121 	 */
2122 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2123 		goto drop;
2124 
2125 	dbif = bridge_lookup_member_if(sc, dst_if);
2126 	if (dbif == NULL)
2127 		/* Not a member of the bridge (anymore?) */
2128 		goto drop;
2129 
2130 	/* Private segments can not talk to each other */
2131 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2132 		goto drop;
2133 
2134 	if ((dbif->bif_flags & IFBIF_STP) &&
2135 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2136 		goto drop;
2137 
2138 	BRIDGE_UNLOCK(sc);
2139 
2140 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2141 #ifdef INET6
2142 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2143 #endif
2144 	    ) {
2145 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2146 			return;
2147 		if (m == NULL)
2148 			return;
2149 	}
2150 
2151 	bridge_enqueue(sc, dst_if, m);
2152 	return;
2153 
2154 drop:
2155 	BRIDGE_UNLOCK(sc);
2156 	m_freem(m);
2157 }
2158 
2159 /*
2160  * bridge_input:
2161  *
2162  *	Receive input from a member interface.  Queue the packet for
2163  *	bridging if it is not for us.
2164  */
2165 static struct mbuf *
2166 bridge_input(struct ifnet *ifp, struct mbuf *m)
2167 {
2168 	struct bridge_softc *sc = ifp->if_bridge;
2169 	struct bridge_iflist *bif, *bif2;
2170 	struct ifnet *bifp;
2171 	struct ether_header *eh;
2172 	struct mbuf *mc, *mc2;
2173 	uint16_t vlan;
2174 	int error;
2175 
2176 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2177 		return (m);
2178 
2179 	bifp = sc->sc_ifp;
2180 	vlan = VLANTAGOF(m);
2181 
2182 	/*
2183 	 * Implement support for bridge monitoring. If this flag has been
2184 	 * set on this interface, discard the packet once we push it through
2185 	 * the bpf(4) machinery, but before we do, increment the byte and
2186 	 * packet counters associated with this interface.
2187 	 */
2188 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2189 		m->m_pkthdr.rcvif  = bifp;
2190 		ETHER_BPF_MTAP(bifp, m);
2191 		bifp->if_ipackets++;
2192 		bifp->if_ibytes += m->m_pkthdr.len;
2193 		m_freem(m);
2194 		return (NULL);
2195 	}
2196 	BRIDGE_LOCK(sc);
2197 	bif = bridge_lookup_member_if(sc, ifp);
2198 	if (bif == NULL) {
2199 		BRIDGE_UNLOCK(sc);
2200 		return (m);
2201 	}
2202 
2203 	eh = mtod(m, struct ether_header *);
2204 
2205 	bridge_span(sc, m);
2206 
2207 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2208 		/* Tap off 802.1D packets; they do not get forwarded. */
2209 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2210 		    ETHER_ADDR_LEN) == 0) {
2211 			m = bstp_input(&bif->bif_stp, ifp, m);
2212 			if (m == NULL) {
2213 				BRIDGE_UNLOCK(sc);
2214 				return (NULL);
2215 			}
2216 		}
2217 
2218 		if ((bif->bif_flags & IFBIF_STP) &&
2219 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2220 			BRIDGE_UNLOCK(sc);
2221 			return (m);
2222 		}
2223 
2224 		/*
2225 		 * Make a deep copy of the packet and enqueue the copy
2226 		 * for bridge processing; return the original packet for
2227 		 * local processing.
2228 		 */
2229 		mc = m_dup(m, M_DONTWAIT);
2230 		if (mc == NULL) {
2231 			BRIDGE_UNLOCK(sc);
2232 			return (m);
2233 		}
2234 
2235 		/* Perform the bridge forwarding function with the copy. */
2236 		bridge_forward(sc, bif, mc);
2237 
2238 		/*
2239 		 * Reinject the mbuf as arriving on the bridge so we have a
2240 		 * chance at claiming multicast packets. We can not loop back
2241 		 * here from ether_input as a bridge is never a member of a
2242 		 * bridge.
2243 		 */
2244 		KASSERT(bifp->if_bridge == NULL,
2245 		    ("loop created in bridge_input"));
2246 		mc2 = m_dup(m, M_DONTWAIT);
2247 		if (mc2 != NULL) {
2248 			/* Keep the layer3 header aligned */
2249 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2250 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2251 		}
2252 		if (mc2 != NULL) {
2253 			mc2->m_pkthdr.rcvif = bifp;
2254 			(*bifp->if_input)(bifp, mc2);
2255 		}
2256 
2257 		/* Return the original packet for local processing. */
2258 		return (m);
2259 	}
2260 
2261 	if ((bif->bif_flags & IFBIF_STP) &&
2262 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2263 		BRIDGE_UNLOCK(sc);
2264 		return (m);
2265 	}
2266 
2267 #if (defined(INET) || defined(INET6))
2268 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2269 	|| ((iface)->if_carp \
2270 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2271 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2272 	|| ((iface)->if_carp \
2273 	    && (*carp_forus_p)((iface), eh->ether_shost))
2274 #else
2275 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2276 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2277 #endif
2278 
2279 #ifdef INET6
2280 #   define OR_PFIL_HOOKED_INET6 \
2281 	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2282 #else
2283 #   define OR_PFIL_HOOKED_INET6
2284 #endif
2285 
2286 #define GRAB_OUR_PACKETS(iface) \
2287 	if ((iface)->if_type == IFT_GIF) \
2288 		continue; \
2289 	/* It is destined for us. */ \
2290 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2291 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2292 	    ) {								\
2293 		if ((iface)->if_type == IFT_BRIDGE) {			\
2294 			ETHER_BPF_MTAP(iface, m);			\
2295 			iface->if_ipackets++;				\
2296 			/* Filter on the physical interface. */		\
2297 			if (pfil_local_phys &&				\
2298 			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2299 			     OR_PFIL_HOOKED_INET6)) {			\
2300 				if (bridge_pfil(&m, NULL, ifp,		\
2301 				    PFIL_IN) != 0 || m == NULL) {	\
2302 					BRIDGE_UNLOCK(sc);		\
2303 					return (NULL);			\
2304 				}					\
2305 			}						\
2306 		}							\
2307 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2308 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2309 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2310 			if (error && bif->bif_addrmax) {		\
2311 				BRIDGE_UNLOCK(sc);			\
2312 				m_freem(m);				\
2313 				return (NULL);				\
2314 			}						\
2315 		}							\
2316 		m->m_pkthdr.rcvif = iface;				\
2317 		BRIDGE_UNLOCK(sc);					\
2318 		return (m);						\
2319 	}								\
2320 									\
2321 	/* We just received a packet that we sent out. */		\
2322 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2323 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2324 	    ) {								\
2325 		BRIDGE_UNLOCK(sc);					\
2326 		m_freem(m);						\
2327 		return (NULL);						\
2328 	}
2329 
2330 	/*
2331 	 * Unicast.  Make sure it's not for the bridge.
2332 	 */
2333 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2334 
2335 	/*
2336 	 * Give a chance for ifp at first priority. This will help when	the
2337 	 * packet comes through the interface like VLAN's with the same MACs
2338 	 * on several interfaces from the same bridge. This also will save
2339 	 * some CPU cycles in case the destination interface and the input
2340 	 * interface (eq ifp) are the same.
2341 	 */
2342 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2343 
2344 	/* Now check the all bridge members. */
2345 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2346 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2347 	}
2348 
2349 #undef OR_CARP_CHECK_WE_ARE_DST
2350 #undef OR_CARP_CHECK_WE_ARE_SRC
2351 #undef OR_PFIL_HOOKED_INET6
2352 #undef GRAB_OUR_PACKETS
2353 
2354 	/* Perform the bridge forwarding function. */
2355 	bridge_forward(sc, bif, m);
2356 
2357 	return (NULL);
2358 }
2359 
2360 /*
2361  * bridge_broadcast:
2362  *
2363  *	Send a frame to all interfaces that are members of
2364  *	the bridge, except for the one on which the packet
2365  *	arrived.
2366  *
2367  *	NOTE: Releases the lock on return.
2368  */
2369 static void
2370 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2371     struct mbuf *m, int runfilt)
2372 {
2373 	struct bridge_iflist *dbif, *sbif;
2374 	struct mbuf *mc;
2375 	struct ifnet *dst_if;
2376 	int error = 0, used = 0, i;
2377 
2378 	sbif = bridge_lookup_member_if(sc, src_if);
2379 
2380 	BRIDGE_LOCK2REF(sc, error);
2381 	if (error) {
2382 		m_freem(m);
2383 		return;
2384 	}
2385 
2386 	/* Filter on the bridge interface before broadcasting */
2387 	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2388 #ifdef INET6
2389 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2390 #endif
2391 	    )) {
2392 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2393 			goto out;
2394 		if (m == NULL)
2395 			goto out;
2396 	}
2397 
2398 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2399 		dst_if = dbif->bif_ifp;
2400 		if (dst_if == src_if)
2401 			continue;
2402 
2403 		/* Private segments can not talk to each other */
2404 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2405 			continue;
2406 
2407 		if ((dbif->bif_flags & IFBIF_STP) &&
2408 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2409 			continue;
2410 
2411 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2412 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2413 			continue;
2414 
2415 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2416 			continue;
2417 
2418 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2419 			mc = m;
2420 			used = 1;
2421 		} else {
2422 			mc = m_dup(m, M_DONTWAIT);
2423 			if (mc == NULL) {
2424 				sc->sc_ifp->if_oerrors++;
2425 				continue;
2426 			}
2427 		}
2428 
2429 		/*
2430 		 * Filter on the output interface. Pass a NULL bridge interface
2431 		 * pointer so we do not redundantly filter on the bridge for
2432 		 * each interface we broadcast on.
2433 		 */
2434 		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2435 #ifdef INET6
2436 		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2437 #endif
2438 		    )) {
2439 			if (used == 0) {
2440 				/* Keep the layer3 header aligned */
2441 				i = min(mc->m_pkthdr.len, max_protohdr);
2442 				mc = m_copyup(mc, i, ETHER_ALIGN);
2443 				if (mc == NULL) {
2444 					sc->sc_ifp->if_oerrors++;
2445 					continue;
2446 				}
2447 			}
2448 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2449 				continue;
2450 			if (mc == NULL)
2451 				continue;
2452 		}
2453 
2454 		bridge_enqueue(sc, dst_if, mc);
2455 	}
2456 	if (used == 0)
2457 		m_freem(m);
2458 
2459 out:
2460 	BRIDGE_UNREF(sc);
2461 }
2462 
2463 /*
2464  * bridge_span:
2465  *
2466  *	Duplicate a packet out one or more interfaces that are in span mode,
2467  *	the original mbuf is unmodified.
2468  */
2469 static void
2470 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2471 {
2472 	struct bridge_iflist *bif;
2473 	struct ifnet *dst_if;
2474 	struct mbuf *mc;
2475 
2476 	if (LIST_EMPTY(&sc->sc_spanlist))
2477 		return;
2478 
2479 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2480 		dst_if = bif->bif_ifp;
2481 
2482 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2483 			continue;
2484 
2485 		mc = m_copypacket(m, M_DONTWAIT);
2486 		if (mc == NULL) {
2487 			sc->sc_ifp->if_oerrors++;
2488 			continue;
2489 		}
2490 
2491 		bridge_enqueue(sc, dst_if, mc);
2492 	}
2493 }
2494 
2495 /*
2496  * bridge_rtupdate:
2497  *
2498  *	Add a bridge routing entry.
2499  */
2500 static int
2501 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2502     struct bridge_iflist *bif, int setflags, uint8_t flags)
2503 {
2504 	struct bridge_rtnode *brt;
2505 	int error;
2506 
2507 	BRIDGE_LOCK_ASSERT(sc);
2508 
2509 	/* Check the source address is valid and not multicast. */
2510 	if (ETHER_IS_MULTICAST(dst) ||
2511 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2512 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2513 		return (EINVAL);
2514 
2515 	/* 802.1p frames map to vlan 1 */
2516 	if (vlan == 0)
2517 		vlan = 1;
2518 
2519 	/*
2520 	 * A route for this destination might already exist.  If so,
2521 	 * update it, otherwise create a new one.
2522 	 */
2523 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2524 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2525 			sc->sc_brtexceeded++;
2526 			return (ENOSPC);
2527 		}
2528 		/* Check per interface address limits (if enabled) */
2529 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2530 			bif->bif_addrexceeded++;
2531 			return (ENOSPC);
2532 		}
2533 
2534 		/*
2535 		 * Allocate a new bridge forwarding node, and
2536 		 * initialize the expiration time and Ethernet
2537 		 * address.
2538 		 */
2539 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2540 		if (brt == NULL)
2541 			return (ENOMEM);
2542 
2543 		if (bif->bif_flags & IFBIF_STICKY)
2544 			brt->brt_flags = IFBAF_STICKY;
2545 		else
2546 			brt->brt_flags = IFBAF_DYNAMIC;
2547 
2548 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2549 		brt->brt_vlan = vlan;
2550 
2551 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2552 			uma_zfree(bridge_rtnode_zone, brt);
2553 			return (error);
2554 		}
2555 		brt->brt_dst = bif;
2556 		bif->bif_addrcnt++;
2557 	}
2558 
2559 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2560 	    brt->brt_dst != bif) {
2561 		brt->brt_dst->bif_addrcnt--;
2562 		brt->brt_dst = bif;
2563 		brt->brt_dst->bif_addrcnt++;
2564 	}
2565 
2566 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2567 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2568 	if (setflags)
2569 		brt->brt_flags = flags;
2570 
2571 	return (0);
2572 }
2573 
2574 /*
2575  * bridge_rtlookup:
2576  *
2577  *	Lookup the destination interface for an address.
2578  */
2579 static struct ifnet *
2580 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2581 {
2582 	struct bridge_rtnode *brt;
2583 
2584 	BRIDGE_LOCK_ASSERT(sc);
2585 
2586 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2587 		return (NULL);
2588 
2589 	return (brt->brt_ifp);
2590 }
2591 
2592 /*
2593  * bridge_rttrim:
2594  *
2595  *	Trim the routine table so that we have a number
2596  *	of routing entries less than or equal to the
2597  *	maximum number.
2598  */
2599 static void
2600 bridge_rttrim(struct bridge_softc *sc)
2601 {
2602 	struct bridge_rtnode *brt, *nbrt;
2603 
2604 	BRIDGE_LOCK_ASSERT(sc);
2605 
2606 	/* Make sure we actually need to do this. */
2607 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2608 		return;
2609 
2610 	/* Force an aging cycle; this might trim enough addresses. */
2611 	bridge_rtage(sc);
2612 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2613 		return;
2614 
2615 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2616 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2617 			bridge_rtnode_destroy(sc, brt);
2618 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2619 				return;
2620 		}
2621 	}
2622 }
2623 
2624 /*
2625  * bridge_timer:
2626  *
2627  *	Aging timer for the bridge.
2628  */
2629 static void
2630 bridge_timer(void *arg)
2631 {
2632 	struct bridge_softc *sc = arg;
2633 
2634 	BRIDGE_LOCK_ASSERT(sc);
2635 
2636 	bridge_rtage(sc);
2637 
2638 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2639 		callout_reset(&sc->sc_brcallout,
2640 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2641 }
2642 
2643 /*
2644  * bridge_rtage:
2645  *
2646  *	Perform an aging cycle.
2647  */
2648 static void
2649 bridge_rtage(struct bridge_softc *sc)
2650 {
2651 	struct bridge_rtnode *brt, *nbrt;
2652 
2653 	BRIDGE_LOCK_ASSERT(sc);
2654 
2655 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2656 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2657 			if (time_uptime >= brt->brt_expire)
2658 				bridge_rtnode_destroy(sc, brt);
2659 		}
2660 	}
2661 }
2662 
2663 /*
2664  * bridge_rtflush:
2665  *
2666  *	Remove all dynamic addresses from the bridge.
2667  */
2668 static void
2669 bridge_rtflush(struct bridge_softc *sc, int full)
2670 {
2671 	struct bridge_rtnode *brt, *nbrt;
2672 
2673 	BRIDGE_LOCK_ASSERT(sc);
2674 
2675 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2676 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2677 			bridge_rtnode_destroy(sc, brt);
2678 	}
2679 }
2680 
2681 /*
2682  * bridge_rtdaddr:
2683  *
2684  *	Remove an address from the table.
2685  */
2686 static int
2687 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2688 {
2689 	struct bridge_rtnode *brt;
2690 	int found = 0;
2691 
2692 	BRIDGE_LOCK_ASSERT(sc);
2693 
2694 	/*
2695 	 * If vlan is zero then we want to delete for all vlans so the lookup
2696 	 * may return more than one.
2697 	 */
2698 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2699 		bridge_rtnode_destroy(sc, brt);
2700 		found = 1;
2701 	}
2702 
2703 	return (found ? 0 : ENOENT);
2704 }
2705 
2706 /*
2707  * bridge_rtdelete:
2708  *
2709  *	Delete routes to a speicifc member interface.
2710  */
2711 static void
2712 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
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_ifp == ifp && (full ||
2720 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2721 			bridge_rtnode_destroy(sc, brt);
2722 	}
2723 }
2724 
2725 /*
2726  * bridge_rtable_init:
2727  *
2728  *	Initialize the route table for this bridge.
2729  */
2730 static int
2731 bridge_rtable_init(struct bridge_softc *sc)
2732 {
2733 	int i;
2734 
2735 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2736 	    M_DEVBUF, M_NOWAIT);
2737 	if (sc->sc_rthash == NULL)
2738 		return (ENOMEM);
2739 
2740 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2741 		LIST_INIT(&sc->sc_rthash[i]);
2742 
2743 	sc->sc_rthash_key = arc4random();
2744 
2745 	LIST_INIT(&sc->sc_rtlist);
2746 
2747 	return (0);
2748 }
2749 
2750 /*
2751  * bridge_rtable_fini:
2752  *
2753  *	Deconstruct the route table for this bridge.
2754  */
2755 static void
2756 bridge_rtable_fini(struct bridge_softc *sc)
2757 {
2758 
2759 	KASSERT(sc->sc_brtcnt == 0,
2760 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2761 	free(sc->sc_rthash, M_DEVBUF);
2762 }
2763 
2764 /*
2765  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2766  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2767  */
2768 #define	mix(a, b, c)							\
2769 do {									\
2770 	a -= b; a -= c; a ^= (c >> 13);					\
2771 	b -= c; b -= a; b ^= (a << 8);					\
2772 	c -= a; c -= b; c ^= (b >> 13);					\
2773 	a -= b; a -= c; a ^= (c >> 12);					\
2774 	b -= c; b -= a; b ^= (a << 16);					\
2775 	c -= a; c -= b; c ^= (b >> 5);					\
2776 	a -= b; a -= c; a ^= (c >> 3);					\
2777 	b -= c; b -= a; b ^= (a << 10);					\
2778 	c -= a; c -= b; c ^= (b >> 15);					\
2779 } while (/*CONSTCOND*/0)
2780 
2781 static __inline uint32_t
2782 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2783 {
2784 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2785 
2786 	b += addr[5] << 8;
2787 	b += addr[4];
2788 	a += addr[3] << 24;
2789 	a += addr[2] << 16;
2790 	a += addr[1] << 8;
2791 	a += addr[0];
2792 
2793 	mix(a, b, c);
2794 
2795 	return (c & BRIDGE_RTHASH_MASK);
2796 }
2797 
2798 #undef mix
2799 
2800 static int
2801 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2802 {
2803 	int i, d;
2804 
2805 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2806 		d = ((int)a[i]) - ((int)b[i]);
2807 	}
2808 
2809 	return (d);
2810 }
2811 
2812 /*
2813  * bridge_rtnode_lookup:
2814  *
2815  *	Look up a bridge route node for the specified destination. Compare the
2816  *	vlan id or if zero then just return the first match.
2817  */
2818 static struct bridge_rtnode *
2819 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2820 {
2821 	struct bridge_rtnode *brt;
2822 	uint32_t hash;
2823 	int dir;
2824 
2825 	BRIDGE_LOCK_ASSERT(sc);
2826 
2827 	hash = bridge_rthash(sc, addr);
2828 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2829 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2830 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2831 			return (brt);
2832 		if (dir > 0)
2833 			return (NULL);
2834 	}
2835 
2836 	return (NULL);
2837 }
2838 
2839 /*
2840  * bridge_rtnode_insert:
2841  *
2842  *	Insert the specified bridge node into the route table.  We
2843  *	assume the entry is not already in the table.
2844  */
2845 static int
2846 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2847 {
2848 	struct bridge_rtnode *lbrt;
2849 	uint32_t hash;
2850 	int dir;
2851 
2852 	BRIDGE_LOCK_ASSERT(sc);
2853 
2854 	hash = bridge_rthash(sc, brt->brt_addr);
2855 
2856 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2857 	if (lbrt == NULL) {
2858 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2859 		goto out;
2860 	}
2861 
2862 	do {
2863 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2864 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2865 			return (EEXIST);
2866 		if (dir > 0) {
2867 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2868 			goto out;
2869 		}
2870 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2871 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2872 			goto out;
2873 		}
2874 		lbrt = LIST_NEXT(lbrt, brt_hash);
2875 	} while (lbrt != NULL);
2876 
2877 #ifdef DIAGNOSTIC
2878 	panic("bridge_rtnode_insert: impossible");
2879 #endif
2880 
2881 out:
2882 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2883 	sc->sc_brtcnt++;
2884 
2885 	return (0);
2886 }
2887 
2888 /*
2889  * bridge_rtnode_destroy:
2890  *
2891  *	Destroy a bridge rtnode.
2892  */
2893 static void
2894 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2895 {
2896 	BRIDGE_LOCK_ASSERT(sc);
2897 
2898 	LIST_REMOVE(brt, brt_hash);
2899 
2900 	LIST_REMOVE(brt, brt_list);
2901 	sc->sc_brtcnt--;
2902 	brt->brt_dst->bif_addrcnt--;
2903 	uma_zfree(bridge_rtnode_zone, brt);
2904 }
2905 
2906 /*
2907  * bridge_rtable_expire:
2908  *
2909  *	Set the expiry time for all routes on an interface.
2910  */
2911 static void
2912 bridge_rtable_expire(struct ifnet *ifp, int age)
2913 {
2914 	struct bridge_softc *sc = ifp->if_bridge;
2915 	struct bridge_rtnode *brt;
2916 
2917 	BRIDGE_LOCK(sc);
2918 
2919 	/*
2920 	 * If the age is zero then flush, otherwise set all the expiry times to
2921 	 * age for the interface
2922 	 */
2923 	if (age == 0)
2924 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2925 	else {
2926 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2927 			/* Cap the expiry time to 'age' */
2928 			if (brt->brt_ifp == ifp &&
2929 			    brt->brt_expire > time_uptime + age &&
2930 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2931 				brt->brt_expire = time_uptime + age;
2932 		}
2933 	}
2934 	BRIDGE_UNLOCK(sc);
2935 }
2936 
2937 /*
2938  * bridge_state_change:
2939  *
2940  *	Callback from the bridgestp code when a port changes states.
2941  */
2942 static void
2943 bridge_state_change(struct ifnet *ifp, int state)
2944 {
2945 	struct bridge_softc *sc = ifp->if_bridge;
2946 	static const char *stpstates[] = {
2947 		"disabled",
2948 		"listening",
2949 		"learning",
2950 		"forwarding",
2951 		"blocking",
2952 		"discarding"
2953 	};
2954 
2955 	if (log_stp)
2956 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
2957 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
2958 }
2959 
2960 /*
2961  * Send bridge packets through pfil if they are one of the types pfil can deal
2962  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
2963  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
2964  * that interface.
2965  */
2966 static int
2967 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
2968 {
2969 	int snap, error, i, hlen;
2970 	struct ether_header *eh1, eh2;
2971 	struct ip_fw_args args;
2972 	struct ip *ip;
2973 	struct llc llc1;
2974 	u_int16_t ether_type;
2975 
2976 	snap = 0;
2977 	error = -1;	/* Default error if not error == 0 */
2978 
2979 #if 0
2980 	/* we may return with the IP fields swapped, ensure its not shared */
2981 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
2982 #endif
2983 
2984 	if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
2985 		return (0); /* filtering is disabled */
2986 
2987 	i = min((*mp)->m_pkthdr.len, max_protohdr);
2988 	if ((*mp)->m_len < i) {
2989 	    *mp = m_pullup(*mp, i);
2990 	    if (*mp == NULL) {
2991 		printf("%s: m_pullup failed\n", __func__);
2992 		return (-1);
2993 	    }
2994 	}
2995 
2996 	eh1 = mtod(*mp, struct ether_header *);
2997 	ether_type = ntohs(eh1->ether_type);
2998 
2999 	/*
3000 	 * Check for SNAP/LLC.
3001 	 */
3002 	if (ether_type < ETHERMTU) {
3003 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3004 
3005 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3006 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3007 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3008 		    llc2->llc_control == LLC_UI) {
3009 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3010 			snap = 1;
3011 		}
3012 	}
3013 
3014 	/*
3015 	 * If we're trying to filter bridge traffic, don't look at anything
3016 	 * other than IP and ARP traffic.  If the filter doesn't understand
3017 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3018 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3019 	 * but of course we don't have an AppleTalk filter to begin with.
3020 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3021 	 * ARP traffic.)
3022 	 */
3023 	switch (ether_type) {
3024 		case ETHERTYPE_ARP:
3025 		case ETHERTYPE_REVARP:
3026 			if (pfil_ipfw_arp == 0)
3027 				return (0); /* Automatically pass */
3028 			break;
3029 
3030 		case ETHERTYPE_IP:
3031 #ifdef INET6
3032 		case ETHERTYPE_IPV6:
3033 #endif /* INET6 */
3034 			break;
3035 		default:
3036 			/*
3037 			 * Check to see if the user wants to pass non-ip
3038 			 * packets, these will not be checked by pfil(9) and
3039 			 * passed unconditionally so the default is to drop.
3040 			 */
3041 			if (pfil_onlyip)
3042 				goto bad;
3043 	}
3044 
3045 	/* Strip off the Ethernet header and keep a copy. */
3046 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3047 	m_adj(*mp, ETHER_HDR_LEN);
3048 
3049 	/* Strip off snap header, if present */
3050 	if (snap) {
3051 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3052 		m_adj(*mp, sizeof(struct llc));
3053 	}
3054 
3055 	/*
3056 	 * Check the IP header for alignment and errors
3057 	 */
3058 	if (dir == PFIL_IN) {
3059 		switch (ether_type) {
3060 			case ETHERTYPE_IP:
3061 				error = bridge_ip_checkbasic(mp);
3062 				break;
3063 #ifdef INET6
3064 			case ETHERTYPE_IPV6:
3065 				error = bridge_ip6_checkbasic(mp);
3066 				break;
3067 #endif /* INET6 */
3068 			default:
3069 				error = 0;
3070 		}
3071 		if (error)
3072 			goto bad;
3073 	}
3074 
3075 	/* XXX this section is also in if_ethersubr.c */
3076 	// XXX PFIL_OUT or DIR_OUT ?
3077 	if (V_ip_fw_chk_ptr && pfil_ipfw != 0 &&
3078 			dir == PFIL_OUT && ifp != NULL) {
3079 		struct m_tag *mtag;
3080 
3081 		error = -1;
3082 		/* fetch the start point from existing tags, if any */
3083 		mtag = m_tag_locate(*mp, MTAG_IPFW_RULE, 0, NULL);
3084 		if (mtag == NULL) {
3085 			args.rule.slot = 0;
3086 		} else {
3087 			struct ipfw_rule_ref *r;
3088 
3089 			/* XXX can we free the tag after use ? */
3090 			mtag->m_tag_id = PACKET_TAG_NONE;
3091 			r = (struct ipfw_rule_ref *)(mtag + 1);
3092 			/* packet already partially processed ? */
3093 			if (r->info & IPFW_ONEPASS)
3094 				goto ipfwpass;
3095 			args.rule = *r;
3096 		}
3097 
3098 		args.m = *mp;
3099 		args.oif = ifp;
3100 		args.next_hop = NULL;
3101 		args.next_hop6 = NULL;
3102 		args.eh = &eh2;
3103 		args.inp = NULL;	/* used by ipfw uid/gid/jail rules */
3104 		i = V_ip_fw_chk_ptr(&args);
3105 		*mp = args.m;
3106 
3107 		if (*mp == NULL)
3108 			return (error);
3109 
3110 		if (ip_dn_io_ptr && (i == IP_FW_DUMMYNET)) {
3111 
3112 			/* put the Ethernet header back on */
3113 			M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3114 			if (*mp == NULL)
3115 				return (error);
3116 			bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3117 
3118 			/*
3119 			 * Pass the pkt to dummynet, which consumes it. The
3120 			 * packet will return to us via bridge_dummynet().
3121 			 */
3122 			args.oif = ifp;
3123 			ip_dn_io_ptr(mp, DIR_FWD | PROTO_IFB, &args);
3124 			return (error);
3125 		}
3126 
3127 		if (i != IP_FW_PASS) /* drop */
3128 			goto bad;
3129 	}
3130 
3131 ipfwpass:
3132 	error = 0;
3133 
3134 	/*
3135 	 * Run the packet through pfil
3136 	 */
3137 	switch (ether_type) {
3138 	case ETHERTYPE_IP:
3139 		/*
3140 		 * before calling the firewall, swap fields the same as
3141 		 * IP does. here we assume the header is contiguous
3142 		 */
3143 		ip = mtod(*mp, struct ip *);
3144 
3145 		ip->ip_len = ntohs(ip->ip_len);
3146 		ip->ip_off = ntohs(ip->ip_off);
3147 
3148 		/*
3149 		 * Run pfil on the member interface and the bridge, both can
3150 		 * be skipped by clearing pfil_member or pfil_bridge.
3151 		 *
3152 		 * Keep the order:
3153 		 *   in_if -> bridge_if -> out_if
3154 		 */
3155 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3156 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3157 					dir, NULL);
3158 
3159 		if (*mp == NULL || error != 0) /* filter may consume */
3160 			break;
3161 
3162 		if (pfil_member && ifp != NULL)
3163 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3164 					dir, NULL);
3165 
3166 		if (*mp == NULL || error != 0) /* filter may consume */
3167 			break;
3168 
3169 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3170 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3171 					dir, NULL);
3172 
3173 		if (*mp == NULL || error != 0) /* filter may consume */
3174 			break;
3175 
3176 		/* check if we need to fragment the packet */
3177 		if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3178 			i = (*mp)->m_pkthdr.len;
3179 			if (i > ifp->if_mtu) {
3180 				error = bridge_fragment(ifp, *mp, &eh2, snap,
3181 					    &llc1);
3182 				return (error);
3183 			}
3184 		}
3185 
3186 		/* Recalculate the ip checksum and restore byte ordering */
3187 		ip = mtod(*mp, struct ip *);
3188 		hlen = ip->ip_hl << 2;
3189 		if (hlen < sizeof(struct ip))
3190 			goto bad;
3191 		if (hlen > (*mp)->m_len) {
3192 			if ((*mp = m_pullup(*mp, hlen)) == 0)
3193 				goto bad;
3194 			ip = mtod(*mp, struct ip *);
3195 			if (ip == NULL)
3196 				goto bad;
3197 		}
3198 		ip->ip_len = htons(ip->ip_len);
3199 		ip->ip_off = htons(ip->ip_off);
3200 		ip->ip_sum = 0;
3201 		if (hlen == sizeof(struct ip))
3202 			ip->ip_sum = in_cksum_hdr(ip);
3203 		else
3204 			ip->ip_sum = in_cksum(*mp, hlen);
3205 
3206 		break;
3207 #ifdef INET6
3208 	case ETHERTYPE_IPV6:
3209 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3210 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3211 					dir, NULL);
3212 
3213 		if (*mp == NULL || error != 0) /* filter may consume */
3214 			break;
3215 
3216 		if (pfil_member && ifp != NULL)
3217 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3218 					dir, NULL);
3219 
3220 		if (*mp == NULL || error != 0) /* filter may consume */
3221 			break;
3222 
3223 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3224 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3225 					dir, NULL);
3226 		break;
3227 #endif
3228 	default:
3229 		error = 0;
3230 		break;
3231 	}
3232 
3233 	if (*mp == NULL)
3234 		return (error);
3235 	if (error != 0)
3236 		goto bad;
3237 
3238 	error = -1;
3239 
3240 	/*
3241 	 * Finally, put everything back the way it was and return
3242 	 */
3243 	if (snap) {
3244 		M_PREPEND(*mp, sizeof(struct llc), M_DONTWAIT);
3245 		if (*mp == NULL)
3246 			return (error);
3247 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3248 	}
3249 
3250 	M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3251 	if (*mp == NULL)
3252 		return (error);
3253 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3254 
3255 	return (0);
3256 
3257 bad:
3258 	m_freem(*mp);
3259 	*mp = NULL;
3260 	return (error);
3261 }
3262 
3263 /*
3264  * Perform basic checks on header size since
3265  * pfil assumes ip_input has already processed
3266  * it for it.  Cut-and-pasted from ip_input.c.
3267  * Given how simple the IPv6 version is,
3268  * does the IPv4 version really need to be
3269  * this complicated?
3270  *
3271  * XXX Should we update ipstat here, or not?
3272  * XXX Right now we update ipstat but not
3273  * XXX csum_counter.
3274  */
3275 static int
3276 bridge_ip_checkbasic(struct mbuf **mp)
3277 {
3278 	struct mbuf *m = *mp;
3279 	struct ip *ip;
3280 	int len, hlen;
3281 	u_short sum;
3282 
3283 	if (*mp == NULL)
3284 		return (-1);
3285 
3286 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3287 		if ((m = m_copyup(m, sizeof(struct ip),
3288 			(max_linkhdr + 3) & ~3)) == NULL) {
3289 			/* XXXJRT new stat, please */
3290 			KMOD_IPSTAT_INC(ips_toosmall);
3291 			goto bad;
3292 		}
3293 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3294 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3295 			KMOD_IPSTAT_INC(ips_toosmall);
3296 			goto bad;
3297 		}
3298 	}
3299 	ip = mtod(m, struct ip *);
3300 	if (ip == NULL) goto bad;
3301 
3302 	if (ip->ip_v != IPVERSION) {
3303 		KMOD_IPSTAT_INC(ips_badvers);
3304 		goto bad;
3305 	}
3306 	hlen = ip->ip_hl << 2;
3307 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3308 		KMOD_IPSTAT_INC(ips_badhlen);
3309 		goto bad;
3310 	}
3311 	if (hlen > m->m_len) {
3312 		if ((m = m_pullup(m, hlen)) == 0) {
3313 			KMOD_IPSTAT_INC(ips_badhlen);
3314 			goto bad;
3315 		}
3316 		ip = mtod(m, struct ip *);
3317 		if (ip == NULL) goto bad;
3318 	}
3319 
3320 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3321 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3322 	} else {
3323 		if (hlen == sizeof(struct ip)) {
3324 			sum = in_cksum_hdr(ip);
3325 		} else {
3326 			sum = in_cksum(m, hlen);
3327 		}
3328 	}
3329 	if (sum) {
3330 		KMOD_IPSTAT_INC(ips_badsum);
3331 		goto bad;
3332 	}
3333 
3334 	/* Retrieve the packet length. */
3335 	len = ntohs(ip->ip_len);
3336 
3337 	/*
3338 	 * Check for additional length bogosity
3339 	 */
3340 	if (len < hlen) {
3341 		KMOD_IPSTAT_INC(ips_badlen);
3342 		goto bad;
3343 	}
3344 
3345 	/*
3346 	 * Check that the amount of data in the buffers
3347 	 * is as at least much as the IP header would have us expect.
3348 	 * Drop packet if shorter than we expect.
3349 	 */
3350 	if (m->m_pkthdr.len < len) {
3351 		KMOD_IPSTAT_INC(ips_tooshort);
3352 		goto bad;
3353 	}
3354 
3355 	/* Checks out, proceed */
3356 	*mp = m;
3357 	return (0);
3358 
3359 bad:
3360 	*mp = m;
3361 	return (-1);
3362 }
3363 
3364 #ifdef INET6
3365 /*
3366  * Same as above, but for IPv6.
3367  * Cut-and-pasted from ip6_input.c.
3368  * XXX Should we update ip6stat, or not?
3369  */
3370 static int
3371 bridge_ip6_checkbasic(struct mbuf **mp)
3372 {
3373 	struct mbuf *m = *mp;
3374 	struct ip6_hdr *ip6;
3375 
3376 	/*
3377 	 * If the IPv6 header is not aligned, slurp it up into a new
3378 	 * mbuf with space for link headers, in the event we forward
3379 	 * it.  Otherwise, if it is aligned, make sure the entire base
3380 	 * IPv6 header is in the first mbuf of the chain.
3381 	 */
3382 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3383 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3384 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3385 			    (max_linkhdr + 3) & ~3)) == NULL) {
3386 			/* XXXJRT new stat, please */
3387 			V_ip6stat.ip6s_toosmall++;
3388 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3389 			goto bad;
3390 		}
3391 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3392 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3393 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3394 			V_ip6stat.ip6s_toosmall++;
3395 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3396 			goto bad;
3397 		}
3398 	}
3399 
3400 	ip6 = mtod(m, struct ip6_hdr *);
3401 
3402 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3403 		V_ip6stat.ip6s_badvers++;
3404 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3405 		goto bad;
3406 	}
3407 
3408 	/* Checks out, proceed */
3409 	*mp = m;
3410 	return (0);
3411 
3412 bad:
3413 	*mp = m;
3414 	return (-1);
3415 }
3416 #endif /* INET6 */
3417 
3418 /*
3419  * bridge_fragment:
3420  *
3421  *	Return a fragmented mbuf chain.
3422  */
3423 static int
3424 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3425     int snap, struct llc *llc)
3426 {
3427 	struct mbuf *m0;
3428 	struct ip *ip;
3429 	int error = -1;
3430 
3431 	if (m->m_len < sizeof(struct ip) &&
3432 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3433 		goto out;
3434 	ip = mtod(m, struct ip *);
3435 
3436 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist,
3437 		    CSUM_DELAY_IP);
3438 	if (error)
3439 		goto out;
3440 
3441 	/* walk the chain and re-add the Ethernet header */
3442 	for (m0 = m; m0; m0 = m0->m_nextpkt) {
3443 		if (error == 0) {
3444 			if (snap) {
3445 				M_PREPEND(m0, sizeof(struct llc), M_DONTWAIT);
3446 				if (m0 == NULL) {
3447 					error = ENOBUFS;
3448 					continue;
3449 				}
3450 				bcopy(llc, mtod(m0, caddr_t),
3451 				    sizeof(struct llc));
3452 			}
3453 			M_PREPEND(m0, ETHER_HDR_LEN, M_DONTWAIT);
3454 			if (m0 == NULL) {
3455 				error = ENOBUFS;
3456 				continue;
3457 			}
3458 			bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3459 		} else
3460 			m_freem(m);
3461 	}
3462 
3463 	if (error == 0)
3464 		KMOD_IPSTAT_INC(ips_fragmented);
3465 
3466 	return (error);
3467 
3468 out:
3469 	if (m != NULL)
3470 		m_freem(m);
3471 	return (error);
3472 }
3473