xref: /freebsd/sys/net/if_bridge.c (revision 8be96e101f2691b80ff9562b72f874da82e735aa)
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/time.h>
89 #include <sys/socket.h> /* for net/if.h */
90 #include <sys/sockio.h>
91 #include <sys/ctype.h>  /* string functions */
92 #include <sys/kernel.h>
93 #include <sys/random.h>
94 #include <sys/syslog.h>
95 #include <sys/sysctl.h>
96 #include <vm/uma.h>
97 #include <sys/module.h>
98 #include <sys/priv.h>
99 #include <sys/proc.h>
100 #include <sys/lock.h>
101 #include <sys/mutex.h>
102 #include <sys/rwlock.h>
103 
104 #include <net/bpf.h>
105 #include <net/if.h>
106 #include <net/if_clone.h>
107 #include <net/if_dl.h>
108 #include <net/if_types.h>
109 #include <net/if_var.h>
110 #include <net/pfil.h>
111 #include <net/vnet.h>
112 
113 #include <netinet/in.h> /* for struct arpcom */
114 #include <netinet/in_systm.h>
115 #include <netinet/in_var.h>
116 #include <netinet/ip.h>
117 #include <netinet/ip_var.h>
118 #ifdef INET6
119 #include <netinet/ip6.h>
120 #include <netinet6/ip6_var.h>
121 #endif
122 #if defined(INET) || defined(INET6)
123 #include <netinet/ip_carp.h>
124 #endif
125 #include <machine/in_cksum.h>
126 #include <netinet/if_ether.h> /* for struct arpcom */
127 #include <net/bridgestp.h>
128 #include <net/if_bridgevar.h>
129 #include <net/if_llc.h>
130 #include <net/if_vlan_var.h>
131 
132 #include <net/route.h>
133 #include <netinet/ip_fw.h>
134 #include <netinet/ipfw/ip_fw_private.h>
135 
136 /*
137  * Size of the route hash table.  Must be a power of two.
138  */
139 #ifndef BRIDGE_RTHASH_SIZE
140 #define	BRIDGE_RTHASH_SIZE		1024
141 #endif
142 
143 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
144 
145 /*
146  * Maximum number of addresses to cache.
147  */
148 #ifndef BRIDGE_RTABLE_MAX
149 #define	BRIDGE_RTABLE_MAX		100
150 #endif
151 
152 /*
153  * Timeout (in seconds) for entries learned dynamically.
154  */
155 #ifndef BRIDGE_RTABLE_TIMEOUT
156 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
157 #endif
158 
159 /*
160  * Number of seconds between walks of the route list.
161  */
162 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
163 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
164 #endif
165 
166 /*
167  * List of capabilities to possibly mask on the member interface.
168  */
169 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
170 
171 /*
172  * List of capabilities to strip
173  */
174 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
175 
176 /*
177  * Bridge interface list entry.
178  */
179 struct bridge_iflist {
180 	LIST_ENTRY(bridge_iflist) bif_next;
181 	struct ifnet		*bif_ifp;	/* member if */
182 	struct bstp_port	bif_stp;	/* STP state */
183 	uint32_t		bif_flags;	/* member if flags */
184 	int			bif_savedcaps;	/* saved capabilities */
185 	uint32_t		bif_addrmax;	/* max # of addresses */
186 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
187 	uint32_t		bif_addrexceeded;/* # of address violations */
188 };
189 
190 /*
191  * Bridge route node.
192  */
193 struct bridge_rtnode {
194 	LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
195 	LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
196 	struct bridge_iflist	*brt_dst;	/* destination if */
197 	unsigned long		brt_expire;	/* expiration time */
198 	uint8_t			brt_flags;	/* address flags */
199 	uint8_t			brt_addr[ETHER_ADDR_LEN];
200 	uint16_t		brt_vlan;	/* vlan id */
201 };
202 #define	brt_ifp			brt_dst->bif_ifp
203 
204 /*
205  * Software state for each bridge.
206  */
207 struct bridge_softc {
208 	struct ifnet		*sc_ifp;	/* make this an interface */
209 	LIST_ENTRY(bridge_softc) sc_list;
210 	struct mtx		sc_mtx;
211 	struct cv		sc_cv;
212 	uint32_t		sc_brtmax;	/* max # of addresses */
213 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
214 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
215 	struct callout		sc_brcallout;	/* bridge callout */
216 	uint32_t		sc_iflist_ref;	/* refcount for sc_iflist */
217 	uint32_t		sc_iflist_xcnt;	/* refcount for sc_iflist */
218 	LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
219 	LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
220 	LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
221 	uint32_t		sc_rthash_key;	/* key for hash */
222 	LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
223 	struct bstp_state	sc_stp;		/* STP state */
224 	uint32_t		sc_brtexceeded;	/* # of cache drops */
225 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
226 	u_char			sc_defaddr[6];	/* Default MAC address */
227 };
228 
229 static struct mtx 	bridge_list_mtx;
230 eventhandler_tag	bridge_detach_cookie = NULL;
231 
232 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
233 
234 uma_zone_t bridge_rtnode_zone;
235 
236 static int	bridge_clone_create(struct if_clone *, int, caddr_t);
237 static void	bridge_clone_destroy(struct ifnet *);
238 
239 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
240 static void	bridge_mutecaps(struct bridge_softc *);
241 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
242 		    int);
243 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
244 static void	bridge_init(void *);
245 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
246 static void	bridge_stop(struct ifnet *, int);
247 static void	bridge_start(struct ifnet *);
248 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
249 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
250 		    struct rtentry *);
251 static void	bridge_enqueue(struct bridge_softc *, struct ifnet *,
252 		    struct mbuf *);
253 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
254 
255 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
256 		    struct mbuf *m);
257 
258 static void	bridge_timer(void *);
259 
260 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
261 		    struct mbuf *, int);
262 static void	bridge_span(struct bridge_softc *, struct mbuf *);
263 
264 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
265 		    uint16_t, struct bridge_iflist *, int, uint8_t);
266 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
267 		    uint16_t);
268 static void	bridge_rttrim(struct bridge_softc *);
269 static void	bridge_rtage(struct bridge_softc *);
270 static void	bridge_rtflush(struct bridge_softc *, int);
271 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
272 		    uint16_t);
273 
274 static int	bridge_rtable_init(struct bridge_softc *);
275 static void	bridge_rtable_fini(struct bridge_softc *);
276 
277 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
278 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
279 		    const uint8_t *, uint16_t);
280 static int	bridge_rtnode_insert(struct bridge_softc *,
281 		    struct bridge_rtnode *);
282 static void	bridge_rtnode_destroy(struct bridge_softc *,
283 		    struct bridge_rtnode *);
284 static void	bridge_rtable_expire(struct ifnet *, int);
285 static void	bridge_state_change(struct ifnet *, int);
286 
287 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
288 		    const char *name);
289 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
290 		    struct ifnet *ifp);
291 static void	bridge_delete_member(struct bridge_softc *,
292 		    struct bridge_iflist *, int);
293 static void	bridge_delete_span(struct bridge_softc *,
294 		    struct bridge_iflist *);
295 
296 static int	bridge_ioctl_add(struct bridge_softc *, void *);
297 static int	bridge_ioctl_del(struct bridge_softc *, void *);
298 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
299 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
300 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
301 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
302 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
303 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
304 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
305 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
306 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
307 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
308 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
309 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
310 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
311 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
312 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
313 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
314 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
315 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
316 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
317 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
318 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
319 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
320 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
321 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
322 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
323 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
324 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
325 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
326 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
327 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
328 		    int);
329 static int	bridge_ip_checkbasic(struct mbuf **mp);
330 #ifdef INET6
331 static int	bridge_ip6_checkbasic(struct mbuf **mp);
332 #endif /* INET6 */
333 static int	bridge_fragment(struct ifnet *, struct mbuf *,
334 		    struct ether_header *, int, struct llc *);
335 
336 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
337 #define	VLANTAGOF(_m)	\
338     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
339 
340 static struct bstp_cb_ops bridge_ops = {
341 	.bcb_state = bridge_state_change,
342 	.bcb_rtage = bridge_rtable_expire
343 };
344 
345 SYSCTL_DECL(_net_link);
346 SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
347 
348 static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */
349 static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */
350 static int pfil_member = 1; /* run pfil hooks on the member interface */
351 static int pfil_ipfw = 0;   /* layer2 filter with ipfw */
352 static int pfil_ipfw_arp = 0;   /* layer2 filter with ipfw */
353 static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for
354                                    locally destined packets */
355 static int log_stp   = 0;   /* log STP state changes */
356 static int bridge_inherit_mac = 0;   /* share MAC with first bridge member */
357 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
358     &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
359 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
360     &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
361 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
362     &pfil_bridge, 0, "Packet filter on the bridge interface");
363 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
364     &pfil_member, 0, "Packet filter on the member interface");
365 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
366     &pfil_local_phys, 0,
367     "Packet filter on the physical interface for locally destined packets");
368 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
369     &log_stp, 0, "Log STP state changes");
370 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, CTLFLAG_RW,
371     &bridge_inherit_mac, 0,
372     "Inherit MAC address from the first bridge member");
373 
374 struct bridge_control {
375 	int	(*bc_func)(struct bridge_softc *, void *);
376 	int	bc_argsize;
377 	int	bc_flags;
378 };
379 
380 #define	BC_F_COPYIN		0x01	/* copy arguments in */
381 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
382 #define	BC_F_SUSER		0x04	/* do super-user check */
383 
384 const struct bridge_control bridge_control_table[] = {
385 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
386 	  BC_F_COPYIN|BC_F_SUSER },
387 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
388 	  BC_F_COPYIN|BC_F_SUSER },
389 
390 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
391 	  BC_F_COPYIN|BC_F_COPYOUT },
392 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
393 	  BC_F_COPYIN|BC_F_SUSER },
394 
395 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
396 	  BC_F_COPYIN|BC_F_SUSER },
397 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
398 	  BC_F_COPYOUT },
399 
400 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
401 	  BC_F_COPYIN|BC_F_COPYOUT },
402 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
403 	  BC_F_COPYIN|BC_F_COPYOUT },
404 
405 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
406 	  BC_F_COPYIN|BC_F_SUSER },
407 
408 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
409 	  BC_F_COPYIN|BC_F_SUSER },
410 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
411 	  BC_F_COPYOUT },
412 
413 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
414 	  BC_F_COPYIN|BC_F_SUSER },
415 
416 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
417 	  BC_F_COPYIN|BC_F_SUSER },
418 
419 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
420 	  BC_F_COPYOUT },
421 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
422 	  BC_F_COPYIN|BC_F_SUSER },
423 
424 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
425 	  BC_F_COPYOUT },
426 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
427 	  BC_F_COPYIN|BC_F_SUSER },
428 
429 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
430 	  BC_F_COPYOUT },
431 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
432 	  BC_F_COPYIN|BC_F_SUSER },
433 
434 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
435 	  BC_F_COPYOUT },
436 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
437 	  BC_F_COPYIN|BC_F_SUSER },
438 
439 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
440 	  BC_F_COPYIN|BC_F_SUSER },
441 
442 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
443 	  BC_F_COPYIN|BC_F_SUSER },
444 
445 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
446 	  BC_F_COPYIN|BC_F_SUSER },
447 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
448 	  BC_F_COPYIN|BC_F_SUSER },
449 
450 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
451 	  BC_F_COPYOUT },
452 
453 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
454 	  BC_F_COPYOUT },
455 
456 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
457 	  BC_F_COPYIN|BC_F_COPYOUT },
458 
459 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
460 	  BC_F_COPYIN|BC_F_SUSER },
461 
462 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
463 	  BC_F_COPYIN|BC_F_SUSER },
464 
465 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
466 	  BC_F_COPYIN|BC_F_SUSER },
467 
468 };
469 const int bridge_control_table_size =
470     sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
471 
472 LIST_HEAD(, bridge_softc) bridge_list;
473 
474 IFC_SIMPLE_DECLARE(bridge, 0);
475 
476 static int
477 bridge_modevent(module_t mod, int type, void *data)
478 {
479 
480 	switch (type) {
481 	case MOD_LOAD:
482 		mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
483 		if_clone_attach(&bridge_cloner);
484 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
485 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
486 		    UMA_ALIGN_PTR, 0);
487 		LIST_INIT(&bridge_list);
488 		bridge_input_p = bridge_input;
489 		bridge_output_p = bridge_output;
490 		bridge_dn_p = bridge_dummynet;
491 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
492 		    ifnet_departure_event, bridge_ifdetach, NULL,
493 		    EVENTHANDLER_PRI_ANY);
494 		break;
495 	case MOD_UNLOAD:
496 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
497 		    bridge_detach_cookie);
498 		if_clone_detach(&bridge_cloner);
499 		uma_zdestroy(bridge_rtnode_zone);
500 		bridge_input_p = NULL;
501 		bridge_output_p = NULL;
502 		bridge_dn_p = NULL;
503 		mtx_destroy(&bridge_list_mtx);
504 		break;
505 	default:
506 		return (EOPNOTSUPP);
507 	}
508 	return (0);
509 }
510 
511 static moduledata_t bridge_mod = {
512 	"if_bridge",
513 	bridge_modevent,
514 	0
515 };
516 
517 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
518 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
519 
520 /*
521  * handler for net.link.bridge.pfil_ipfw
522  */
523 static int
524 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
525 {
526 	int enable = pfil_ipfw;
527 	int error;
528 
529 	error = sysctl_handle_int(oidp, &enable, 0, req);
530 	enable = (enable) ? 1 : 0;
531 
532 	if (enable != pfil_ipfw) {
533 		pfil_ipfw = enable;
534 
535 		/*
536 		 * Disable pfil so that ipfw doesnt run twice, if the user
537 		 * really wants both then they can re-enable pfil_bridge and/or
538 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
539 		 * layer2 type.
540 		 */
541 		if (pfil_ipfw) {
542 			pfil_onlyip = 0;
543 			pfil_bridge = 0;
544 			pfil_member = 0;
545 		}
546 	}
547 
548 	return (error);
549 }
550 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
551 	    &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
552 
553 /*
554  * bridge_clone_create:
555  *
556  *	Create a new bridge instance.
557  */
558 static int
559 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
560 {
561 	struct bridge_softc *sc, *sc2;
562 	struct ifnet *bifp, *ifp;
563 	int retry;
564 
565 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
566 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
567 	if (ifp == NULL) {
568 		free(sc, M_DEVBUF);
569 		return (ENOSPC);
570 	}
571 
572 	BRIDGE_LOCK_INIT(sc);
573 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
574 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
575 
576 	/* Initialize our routing table. */
577 	bridge_rtable_init(sc);
578 
579 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
580 
581 	LIST_INIT(&sc->sc_iflist);
582 	LIST_INIT(&sc->sc_spanlist);
583 
584 	ifp->if_softc = sc;
585 	if_initname(ifp, ifc->ifc_name, unit);
586 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
587 	ifp->if_ioctl = bridge_ioctl;
588 	ifp->if_start = bridge_start;
589 	ifp->if_init = bridge_init;
590 	ifp->if_type = IFT_BRIDGE;
591 	IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
592 	ifp->if_snd.ifq_drv_maxlen = ifqmaxlen;
593 	IFQ_SET_READY(&ifp->if_snd);
594 
595 	/*
596 	 * Generate a random ethernet address with a locally administered
597 	 * address.
598 	 *
599 	 * Since we are using random ethernet addresses for the bridge, it is
600 	 * possible that we might have address collisions, so make sure that
601 	 * this hardware address isn't already in use on another bridge.
602 	 */
603 	for (retry = 1; retry != 0;) {
604 		arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
605 		sc->sc_defaddr[0] &= ~1;	/* clear multicast bit */
606 		sc->sc_defaddr[0] |= 2;		/* set the LAA bit */
607 		retry = 0;
608 		mtx_lock(&bridge_list_mtx);
609 		LIST_FOREACH(sc2, &bridge_list, sc_list) {
610 			bifp = sc2->sc_ifp;
611 			if (memcmp(sc->sc_defaddr,
612 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0)
613 				retry = 1;
614 		}
615 		mtx_unlock(&bridge_list_mtx);
616 	}
617 
618 	bstp_attach(&sc->sc_stp, &bridge_ops);
619 	ether_ifattach(ifp, sc->sc_defaddr);
620 	/* Now undo some of the damage... */
621 	ifp->if_baudrate = 0;
622 	ifp->if_type = IFT_BRIDGE;
623 
624 	mtx_lock(&bridge_list_mtx);
625 	LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
626 	mtx_unlock(&bridge_list_mtx);
627 
628 	return (0);
629 }
630 
631 /*
632  * bridge_clone_destroy:
633  *
634  *	Destroy a bridge instance.
635  */
636 static void
637 bridge_clone_destroy(struct ifnet *ifp)
638 {
639 	struct bridge_softc *sc = ifp->if_softc;
640 	struct bridge_iflist *bif;
641 
642 	BRIDGE_LOCK(sc);
643 
644 	bridge_stop(ifp, 1);
645 	ifp->if_flags &= ~IFF_UP;
646 
647 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
648 		bridge_delete_member(sc, bif, 0);
649 
650 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
651 		bridge_delete_span(sc, bif);
652 	}
653 
654 	BRIDGE_UNLOCK(sc);
655 
656 	callout_drain(&sc->sc_brcallout);
657 
658 	mtx_lock(&bridge_list_mtx);
659 	LIST_REMOVE(sc, sc_list);
660 	mtx_unlock(&bridge_list_mtx);
661 
662 	bstp_detach(&sc->sc_stp);
663 	ether_ifdetach(ifp);
664 	if_free_type(ifp, IFT_ETHER);
665 
666 	/* Tear down the routing table. */
667 	bridge_rtable_fini(sc);
668 
669 	BRIDGE_LOCK_DESTROY(sc);
670 	free(sc, M_DEVBUF);
671 }
672 
673 /*
674  * bridge_ioctl:
675  *
676  *	Handle a control request from the operator.
677  */
678 static int
679 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
680 {
681 	struct bridge_softc *sc = ifp->if_softc;
682 	struct ifreq *ifr = (struct ifreq *)data;
683 	struct bridge_iflist *bif;
684 	struct thread *td = curthread;
685 	union {
686 		struct ifbreq ifbreq;
687 		struct ifbifconf ifbifconf;
688 		struct ifbareq ifbareq;
689 		struct ifbaconf ifbaconf;
690 		struct ifbrparam ifbrparam;
691 		struct ifbropreq ifbropreq;
692 	} args;
693 	struct ifdrv *ifd = (struct ifdrv *) data;
694 	const struct bridge_control *bc;
695 	int error = 0;
696 
697 	switch (cmd) {
698 
699 	case SIOCADDMULTI:
700 	case SIOCDELMULTI:
701 		break;
702 
703 	case SIOCGDRVSPEC:
704 	case SIOCSDRVSPEC:
705 		if (ifd->ifd_cmd >= bridge_control_table_size) {
706 			error = EINVAL;
707 			break;
708 		}
709 		bc = &bridge_control_table[ifd->ifd_cmd];
710 
711 		if (cmd == SIOCGDRVSPEC &&
712 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
713 			error = EINVAL;
714 			break;
715 		}
716 		else if (cmd == SIOCSDRVSPEC &&
717 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
718 			error = EINVAL;
719 			break;
720 		}
721 
722 		if (bc->bc_flags & BC_F_SUSER) {
723 			error = priv_check(td, PRIV_NET_BRIDGE);
724 			if (error)
725 				break;
726 		}
727 
728 		if (ifd->ifd_len != bc->bc_argsize ||
729 		    ifd->ifd_len > sizeof(args)) {
730 			error = EINVAL;
731 			break;
732 		}
733 
734 		bzero(&args, sizeof(args));
735 		if (bc->bc_flags & BC_F_COPYIN) {
736 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
737 			if (error)
738 				break;
739 		}
740 
741 		BRIDGE_LOCK(sc);
742 		error = (*bc->bc_func)(sc, &args);
743 		BRIDGE_UNLOCK(sc);
744 		if (error)
745 			break;
746 
747 		if (bc->bc_flags & BC_F_COPYOUT)
748 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
749 
750 		break;
751 
752 	case SIOCSIFFLAGS:
753 		if (!(ifp->if_flags & IFF_UP) &&
754 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
755 			/*
756 			 * If interface is marked down and it is running,
757 			 * then stop and disable it.
758 			 */
759 			BRIDGE_LOCK(sc);
760 			bridge_stop(ifp, 1);
761 			BRIDGE_UNLOCK(sc);
762 		} else if ((ifp->if_flags & IFF_UP) &&
763 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
764 			/*
765 			 * If interface is marked up and it is stopped, then
766 			 * start it.
767 			 */
768 			(*ifp->if_init)(sc);
769 		}
770 		break;
771 
772 	case SIOCSIFMTU:
773 		if (ifr->ifr_mtu < 576) {
774 			error = EINVAL;
775 			break;
776 		}
777 		if (LIST_EMPTY(&sc->sc_iflist)) {
778 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
779 			break;
780 		}
781 		BRIDGE_LOCK(sc);
782 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
783 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
784 				log(LOG_NOTICE, "%s: invalid MTU: %lu(%s)"
785 				    " != %d\n", sc->sc_ifp->if_xname,
786 				    bif->bif_ifp->if_mtu,
787 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
788 				error = EINVAL;
789 				break;
790 			}
791 		}
792 		if (!error)
793 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
794 		BRIDGE_UNLOCK(sc);
795 		break;
796 	default:
797 		/*
798 		 * drop the lock as ether_ioctl() will call bridge_start() and
799 		 * cause the lock to be recursed.
800 		 */
801 		error = ether_ioctl(ifp, cmd, data);
802 		break;
803 	}
804 
805 	return (error);
806 }
807 
808 /*
809  * bridge_mutecaps:
810  *
811  *	Clear or restore unwanted capabilities on the member interface
812  */
813 static void
814 bridge_mutecaps(struct bridge_softc *sc)
815 {
816 	struct bridge_iflist *bif;
817 	int enabled, mask;
818 
819 	/* Initial bitmask of capabilities to test */
820 	mask = BRIDGE_IFCAPS_MASK;
821 
822 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
823 		/* Every member must support it or its disabled */
824 		mask &= bif->bif_savedcaps;
825 	}
826 
827 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
828 		enabled = bif->bif_ifp->if_capenable;
829 		enabled &= ~BRIDGE_IFCAPS_STRIP;
830 		/* strip off mask bits and enable them again if allowed */
831 		enabled &= ~BRIDGE_IFCAPS_MASK;
832 		enabled |= mask;
833 		bridge_set_ifcap(sc, bif, enabled);
834 	}
835 
836 }
837 
838 static void
839 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
840 {
841 	struct ifnet *ifp = bif->bif_ifp;
842 	struct ifreq ifr;
843 	int error;
844 
845 	bzero(&ifr, sizeof(ifr));
846 	ifr.ifr_reqcap = set;
847 
848 	if (ifp->if_capenable != set) {
849 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
850 		if (error)
851 			if_printf(sc->sc_ifp,
852 			    "error setting interface capabilities on %s\n",
853 			    ifp->if_xname);
854 	}
855 }
856 
857 /*
858  * bridge_lookup_member:
859  *
860  *	Lookup a bridge member interface.
861  */
862 static struct bridge_iflist *
863 bridge_lookup_member(struct bridge_softc *sc, const char *name)
864 {
865 	struct bridge_iflist *bif;
866 	struct ifnet *ifp;
867 
868 	BRIDGE_LOCK_ASSERT(sc);
869 
870 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
871 		ifp = bif->bif_ifp;
872 		if (strcmp(ifp->if_xname, name) == 0)
873 			return (bif);
874 	}
875 
876 	return (NULL);
877 }
878 
879 /*
880  * bridge_lookup_member_if:
881  *
882  *	Lookup a bridge member interface by ifnet*.
883  */
884 static struct bridge_iflist *
885 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
886 {
887 	struct bridge_iflist *bif;
888 
889 	BRIDGE_LOCK_ASSERT(sc);
890 
891 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
892 		if (bif->bif_ifp == member_ifp)
893 			return (bif);
894 	}
895 
896 	return (NULL);
897 }
898 
899 /*
900  * bridge_delete_member:
901  *
902  *	Delete the specified member interface.
903  */
904 static void
905 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
906     int gone)
907 {
908 	struct ifnet *ifs = bif->bif_ifp;
909 	struct ifnet *fif = NULL;
910 
911 	BRIDGE_LOCK_ASSERT(sc);
912 
913 	if (bif->bif_flags & IFBIF_STP)
914 		bstp_disable(&bif->bif_stp);
915 
916 	ifs->if_bridge = NULL;
917 	BRIDGE_XLOCK(sc);
918 	LIST_REMOVE(bif, bif_next);
919 	BRIDGE_XDROP(sc);
920 
921 	/*
922 	 * If removing the interface that gave the bridge its mac address, set
923 	 * the mac address of the bridge to the address of the next member, or
924 	 * to its default address if no members are left.
925 	 */
926 	if (bridge_inherit_mac && sc->sc_ifaddr == ifs) {
927 		if (LIST_EMPTY(&sc->sc_iflist)) {
928 			bcopy(sc->sc_defaddr,
929 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
930 			sc->sc_ifaddr = NULL;
931 		} else {
932 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
933 			bcopy(IF_LLADDR(fif),
934 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
935 			sc->sc_ifaddr = fif;
936 		}
937 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
938 	}
939 
940 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
941 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
942 	KASSERT(bif->bif_addrcnt == 0,
943 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
944 
945 	BRIDGE_UNLOCK(sc);
946 	if (!gone) {
947 		switch (ifs->if_type) {
948 		case IFT_ETHER:
949 		case IFT_L2VLAN:
950 			/*
951 			 * Take the interface out of promiscuous mode.
952 			 */
953 			(void) ifpromisc(ifs, 0);
954 			break;
955 
956 		case IFT_GIF:
957 			break;
958 
959 		default:
960 #ifdef DIAGNOSTIC
961 			panic("bridge_delete_member: impossible");
962 #endif
963 			break;
964 		}
965 		/* reneable any interface capabilities */
966 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
967 	}
968 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
969 	BRIDGE_LOCK(sc);
970 	free(bif, M_DEVBUF);
971 }
972 
973 /*
974  * bridge_delete_span:
975  *
976  *	Delete the specified span interface.
977  */
978 static void
979 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
980 {
981 	BRIDGE_LOCK_ASSERT(sc);
982 
983 	KASSERT(bif->bif_ifp->if_bridge == NULL,
984 	    ("%s: not a span interface", __func__));
985 
986 	LIST_REMOVE(bif, bif_next);
987 	free(bif, M_DEVBUF);
988 }
989 
990 static int
991 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
992 {
993 	struct ifbreq *req = arg;
994 	struct bridge_iflist *bif = NULL;
995 	struct ifnet *ifs;
996 	int error = 0;
997 
998 	ifs = ifunit(req->ifbr_ifsname);
999 	if (ifs == NULL)
1000 		return (ENOENT);
1001 	if (ifs->if_ioctl == NULL)	/* must be supported */
1002 		return (EINVAL);
1003 
1004 	/* If it's in the span list, it can't be a member. */
1005 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1006 		if (ifs == bif->bif_ifp)
1007 			return (EBUSY);
1008 
1009 	if (ifs->if_bridge == sc)
1010 		return (EEXIST);
1011 
1012 	if (ifs->if_bridge != NULL)
1013 		return (EBUSY);
1014 
1015 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1016 	if (bif == NULL)
1017 		return (ENOMEM);
1018 
1019 	bif->bif_ifp = ifs;
1020 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1021 	bif->bif_savedcaps = ifs->if_capenable;
1022 
1023 	switch (ifs->if_type) {
1024 	case IFT_ETHER:
1025 	case IFT_L2VLAN:
1026 	case IFT_GIF:
1027 		/* permitted interface types */
1028 		break;
1029 	default:
1030 		error = EINVAL;
1031 		goto out;
1032 	}
1033 
1034 	/* Allow the first Ethernet member to define the MTU */
1035 	if (LIST_EMPTY(&sc->sc_iflist))
1036 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1037 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1038 		if_printf(sc->sc_ifp, "invalid MTU: %lu(%s) != %lu\n",
1039 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1040 		error = EINVAL;
1041 		goto out;
1042 	}
1043 
1044 	/*
1045 	 * Assign the interface's MAC address to the bridge if it's the first
1046 	 * member and the MAC address of the bridge has not been changed from
1047 	 * the default randomly generated one.
1048 	 */
1049 	if (bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1050 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1051 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1052 		sc->sc_ifaddr = ifs;
1053 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1054 	}
1055 
1056 	ifs->if_bridge = sc;
1057 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1058 	/*
1059 	 * XXX: XLOCK HERE!?!
1060 	 *
1061 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1062 	 */
1063 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1064 
1065 	/* Set interface capabilities to the intersection set of all members */
1066 	bridge_mutecaps(sc);
1067 
1068 	switch (ifs->if_type) {
1069 	case IFT_ETHER:
1070 	case IFT_L2VLAN:
1071 		/*
1072 		 * Place the interface into promiscuous mode.
1073 		 */
1074 		BRIDGE_UNLOCK(sc);
1075 		error = ifpromisc(ifs, 1);
1076 		BRIDGE_LOCK(sc);
1077 		break;
1078 	}
1079 	if (error)
1080 		bridge_delete_member(sc, bif, 0);
1081 out:
1082 	if (error) {
1083 		if (bif != NULL)
1084 			free(bif, M_DEVBUF);
1085 	}
1086 	return (error);
1087 }
1088 
1089 static int
1090 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1091 {
1092 	struct ifbreq *req = arg;
1093 	struct bridge_iflist *bif;
1094 
1095 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1096 	if (bif == NULL)
1097 		return (ENOENT);
1098 
1099 	bridge_delete_member(sc, bif, 0);
1100 
1101 	return (0);
1102 }
1103 
1104 static int
1105 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1106 {
1107 	struct ifbreq *req = arg;
1108 	struct bridge_iflist *bif;
1109 	struct bstp_port *bp;
1110 
1111 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1112 	if (bif == NULL)
1113 		return (ENOENT);
1114 
1115 	bp = &bif->bif_stp;
1116 	req->ifbr_ifsflags = bif->bif_flags;
1117 	req->ifbr_state = bp->bp_state;
1118 	req->ifbr_priority = bp->bp_priority;
1119 	req->ifbr_path_cost = bp->bp_path_cost;
1120 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1121 	req->ifbr_proto = bp->bp_protover;
1122 	req->ifbr_role = bp->bp_role;
1123 	req->ifbr_stpflags = bp->bp_flags;
1124 	req->ifbr_addrcnt = bif->bif_addrcnt;
1125 	req->ifbr_addrmax = bif->bif_addrmax;
1126 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1127 
1128 	/* Copy STP state options as flags */
1129 	if (bp->bp_operedge)
1130 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1131 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1132 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1133 	if (bp->bp_ptp_link)
1134 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1135 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1136 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1137 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1138 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1139 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1140 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1141 	return (0);
1142 }
1143 
1144 static int
1145 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1146 {
1147 	struct ifbreq *req = arg;
1148 	struct bridge_iflist *bif;
1149 	struct bstp_port *bp;
1150 	int error;
1151 
1152 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1153 	if (bif == NULL)
1154 		return (ENOENT);
1155 	bp = &bif->bif_stp;
1156 
1157 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1158 		/* SPAN is readonly */
1159 		return (EINVAL);
1160 
1161 	if (req->ifbr_ifsflags & IFBIF_STP) {
1162 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1163 			error = bstp_enable(&bif->bif_stp);
1164 			if (error)
1165 				return (error);
1166 		}
1167 	} else {
1168 		if ((bif->bif_flags & IFBIF_STP) != 0)
1169 			bstp_disable(&bif->bif_stp);
1170 	}
1171 
1172 	/* Pass on STP flags */
1173 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1174 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1175 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1176 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1177 
1178 	/* Save the bits relating to the bridge */
1179 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1180 
1181 	return (0);
1182 }
1183 
1184 static int
1185 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1186 {
1187 	struct ifbrparam *param = arg;
1188 
1189 	sc->sc_brtmax = param->ifbrp_csize;
1190 	bridge_rttrim(sc);
1191 
1192 	return (0);
1193 }
1194 
1195 static int
1196 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1197 {
1198 	struct ifbrparam *param = arg;
1199 
1200 	param->ifbrp_csize = sc->sc_brtmax;
1201 
1202 	return (0);
1203 }
1204 
1205 static int
1206 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1207 {
1208 	struct ifbifconf *bifc = arg;
1209 	struct bridge_iflist *bif;
1210 	struct ifbreq breq;
1211 	char *buf, *outbuf;
1212 	int count, buflen, len, error = 0;
1213 
1214 	count = 0;
1215 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1216 		count++;
1217 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1218 		count++;
1219 
1220 	buflen = sizeof(breq) * count;
1221 	if (bifc->ifbic_len == 0) {
1222 		bifc->ifbic_len = buflen;
1223 		return (0);
1224 	}
1225 	BRIDGE_UNLOCK(sc);
1226 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1227 	BRIDGE_LOCK(sc);
1228 
1229 	count = 0;
1230 	buf = outbuf;
1231 	len = min(bifc->ifbic_len, buflen);
1232 	bzero(&breq, sizeof(breq));
1233 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1234 		if (len < sizeof(breq))
1235 			break;
1236 
1237 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1238 		    sizeof(breq.ifbr_ifsname));
1239 		/* Fill in the ifbreq structure */
1240 		error = bridge_ioctl_gifflags(sc, &breq);
1241 		if (error)
1242 			break;
1243 		memcpy(buf, &breq, sizeof(breq));
1244 		count++;
1245 		buf += sizeof(breq);
1246 		len -= sizeof(breq);
1247 	}
1248 	LIST_FOREACH(bif, &sc->sc_spanlist, 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 		breq.ifbr_ifsflags = bif->bif_flags;
1255 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1256 		memcpy(buf, &breq, sizeof(breq));
1257 		count++;
1258 		buf += sizeof(breq);
1259 		len -= sizeof(breq);
1260 	}
1261 
1262 	BRIDGE_UNLOCK(sc);
1263 	bifc->ifbic_len = sizeof(breq) * count;
1264 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1265 	BRIDGE_LOCK(sc);
1266 	free(outbuf, M_TEMP);
1267 	return (error);
1268 }
1269 
1270 static int
1271 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1272 {
1273 	struct ifbaconf *bac = arg;
1274 	struct bridge_rtnode *brt;
1275 	struct ifbareq bareq;
1276 	char *buf, *outbuf;
1277 	int count, buflen, len, error = 0;
1278 
1279 	if (bac->ifbac_len == 0)
1280 		return (0);
1281 
1282 	count = 0;
1283 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1284 		count++;
1285 	buflen = sizeof(bareq) * count;
1286 
1287 	BRIDGE_UNLOCK(sc);
1288 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1289 	BRIDGE_LOCK(sc);
1290 
1291 	count = 0;
1292 	buf = outbuf;
1293 	len = min(bac->ifbac_len, buflen);
1294 	bzero(&bareq, sizeof(bareq));
1295 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1296 		if (len < sizeof(bareq))
1297 			goto out;
1298 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1299 		    sizeof(bareq.ifba_ifsname));
1300 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1301 		bareq.ifba_vlan = brt->brt_vlan;
1302 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1303 				time_uptime < brt->brt_expire)
1304 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1305 		else
1306 			bareq.ifba_expire = 0;
1307 		bareq.ifba_flags = brt->brt_flags;
1308 
1309 		memcpy(buf, &bareq, sizeof(bareq));
1310 		count++;
1311 		buf += sizeof(bareq);
1312 		len -= sizeof(bareq);
1313 	}
1314 out:
1315 	BRIDGE_UNLOCK(sc);
1316 	bac->ifbac_len = sizeof(bareq) * count;
1317 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1318 	BRIDGE_LOCK(sc);
1319 	free(outbuf, M_TEMP);
1320 	return (error);
1321 }
1322 
1323 static int
1324 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1325 {
1326 	struct ifbareq *req = arg;
1327 	struct bridge_iflist *bif;
1328 	int error;
1329 
1330 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1331 	if (bif == NULL)
1332 		return (ENOENT);
1333 
1334 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1335 	    req->ifba_flags);
1336 
1337 	return (error);
1338 }
1339 
1340 static int
1341 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1342 {
1343 	struct ifbrparam *param = arg;
1344 
1345 	sc->sc_brttimeout = param->ifbrp_ctime;
1346 	return (0);
1347 }
1348 
1349 static int
1350 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1351 {
1352 	struct ifbrparam *param = arg;
1353 
1354 	param->ifbrp_ctime = sc->sc_brttimeout;
1355 	return (0);
1356 }
1357 
1358 static int
1359 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1360 {
1361 	struct ifbareq *req = arg;
1362 
1363 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1364 }
1365 
1366 static int
1367 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1368 {
1369 	struct ifbreq *req = arg;
1370 
1371 	bridge_rtflush(sc, req->ifbr_ifsflags);
1372 	return (0);
1373 }
1374 
1375 static int
1376 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1377 {
1378 	struct ifbrparam *param = arg;
1379 	struct bstp_state *bs = &sc->sc_stp;
1380 
1381 	param->ifbrp_prio = bs->bs_bridge_priority;
1382 	return (0);
1383 }
1384 
1385 static int
1386 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1387 {
1388 	struct ifbrparam *param = arg;
1389 
1390 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1391 }
1392 
1393 static int
1394 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1395 {
1396 	struct ifbrparam *param = arg;
1397 	struct bstp_state *bs = &sc->sc_stp;
1398 
1399 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1400 	return (0);
1401 }
1402 
1403 static int
1404 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1405 {
1406 	struct ifbrparam *param = arg;
1407 
1408 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1409 }
1410 
1411 static int
1412 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1413 {
1414 	struct ifbrparam *param = arg;
1415 	struct bstp_state *bs = &sc->sc_stp;
1416 
1417 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1418 	return (0);
1419 }
1420 
1421 static int
1422 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1423 {
1424 	struct ifbrparam *param = arg;
1425 
1426 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1427 }
1428 
1429 static int
1430 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1431 {
1432 	struct ifbrparam *param = arg;
1433 	struct bstp_state *bs = &sc->sc_stp;
1434 
1435 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1436 	return (0);
1437 }
1438 
1439 static int
1440 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1441 {
1442 	struct ifbrparam *param = arg;
1443 
1444 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1445 }
1446 
1447 static int
1448 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1449 {
1450 	struct ifbreq *req = arg;
1451 	struct bridge_iflist *bif;
1452 
1453 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1454 	if (bif == NULL)
1455 		return (ENOENT);
1456 
1457 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1458 }
1459 
1460 static int
1461 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1462 {
1463 	struct ifbreq *req = arg;
1464 	struct bridge_iflist *bif;
1465 
1466 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1467 	if (bif == NULL)
1468 		return (ENOENT);
1469 
1470 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1471 }
1472 
1473 static int
1474 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1475 {
1476 	struct ifbreq *req = arg;
1477 	struct bridge_iflist *bif;
1478 
1479 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1480 	if (bif == NULL)
1481 		return (ENOENT);
1482 
1483 	bif->bif_addrmax = req->ifbr_addrmax;
1484 	return (0);
1485 }
1486 
1487 static int
1488 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1489 {
1490 	struct ifbreq *req = arg;
1491 	struct bridge_iflist *bif = NULL;
1492 	struct ifnet *ifs;
1493 
1494 	ifs = ifunit(req->ifbr_ifsname);
1495 	if (ifs == NULL)
1496 		return (ENOENT);
1497 
1498 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1499 		if (ifs == bif->bif_ifp)
1500 			return (EBUSY);
1501 
1502 	if (ifs->if_bridge != NULL)
1503 		return (EBUSY);
1504 
1505 	switch (ifs->if_type) {
1506 		case IFT_ETHER:
1507 		case IFT_GIF:
1508 		case IFT_L2VLAN:
1509 			break;
1510 		default:
1511 			return (EINVAL);
1512 	}
1513 
1514 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1515 	if (bif == NULL)
1516 		return (ENOMEM);
1517 
1518 	bif->bif_ifp = ifs;
1519 	bif->bif_flags = IFBIF_SPAN;
1520 
1521 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1522 
1523 	return (0);
1524 }
1525 
1526 static int
1527 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1528 {
1529 	struct ifbreq *req = arg;
1530 	struct bridge_iflist *bif;
1531 	struct ifnet *ifs;
1532 
1533 	ifs = ifunit(req->ifbr_ifsname);
1534 	if (ifs == NULL)
1535 		return (ENOENT);
1536 
1537 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1538 		if (ifs == bif->bif_ifp)
1539 			break;
1540 
1541 	if (bif == NULL)
1542 		return (ENOENT);
1543 
1544 	bridge_delete_span(sc, bif);
1545 
1546 	return (0);
1547 }
1548 
1549 static int
1550 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1551 {
1552 	struct ifbropreq *req = arg;
1553 	struct bstp_state *bs = &sc->sc_stp;
1554 	struct bstp_port *root_port;
1555 
1556 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1557 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1558 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1559 
1560 	root_port = bs->bs_root_port;
1561 	if (root_port == NULL)
1562 		req->ifbop_root_port = 0;
1563 	else
1564 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1565 
1566 	req->ifbop_holdcount = bs->bs_txholdcount;
1567 	req->ifbop_priority = bs->bs_bridge_priority;
1568 	req->ifbop_protocol = bs->bs_protover;
1569 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1570 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1571 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1572 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1573 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1574 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1575 
1576 	return (0);
1577 }
1578 
1579 static int
1580 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1581 {
1582 	struct ifbrparam *param = arg;
1583 
1584 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1585 	return (0);
1586 }
1587 
1588 static int
1589 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1590 {
1591 	struct ifbpstpconf *bifstp = arg;
1592 	struct bridge_iflist *bif;
1593 	struct bstp_port *bp;
1594 	struct ifbpstpreq bpreq;
1595 	char *buf, *outbuf;
1596 	int count, buflen, len, error = 0;
1597 
1598 	count = 0;
1599 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1600 		if ((bif->bif_flags & IFBIF_STP) != 0)
1601 			count++;
1602 	}
1603 
1604 	buflen = sizeof(bpreq) * count;
1605 	if (bifstp->ifbpstp_len == 0) {
1606 		bifstp->ifbpstp_len = buflen;
1607 		return (0);
1608 	}
1609 
1610 	BRIDGE_UNLOCK(sc);
1611 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1612 	BRIDGE_LOCK(sc);
1613 
1614 	count = 0;
1615 	buf = outbuf;
1616 	len = min(bifstp->ifbpstp_len, buflen);
1617 	bzero(&bpreq, sizeof(bpreq));
1618 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1619 		if (len < sizeof(bpreq))
1620 			break;
1621 
1622 		if ((bif->bif_flags & IFBIF_STP) == 0)
1623 			continue;
1624 
1625 		bp = &bif->bif_stp;
1626 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1627 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1628 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1629 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1630 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1631 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1632 
1633 		memcpy(buf, &bpreq, sizeof(bpreq));
1634 		count++;
1635 		buf += sizeof(bpreq);
1636 		len -= sizeof(bpreq);
1637 	}
1638 
1639 	BRIDGE_UNLOCK(sc);
1640 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1641 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1642 	BRIDGE_LOCK(sc);
1643 	free(outbuf, M_TEMP);
1644 	return (error);
1645 }
1646 
1647 static int
1648 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1649 {
1650 	struct ifbrparam *param = arg;
1651 
1652 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1653 }
1654 
1655 static int
1656 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1657 {
1658 	struct ifbrparam *param = arg;
1659 
1660 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1661 }
1662 
1663 /*
1664  * bridge_ifdetach:
1665  *
1666  *	Detach an interface from a bridge.  Called when a member
1667  *	interface is detaching.
1668  */
1669 static void
1670 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1671 {
1672 	struct bridge_softc *sc = ifp->if_bridge;
1673 	struct bridge_iflist *bif;
1674 
1675 	/* Check if the interface is a bridge member */
1676 	if (sc != NULL) {
1677 		BRIDGE_LOCK(sc);
1678 
1679 		bif = bridge_lookup_member_if(sc, ifp);
1680 		if (bif != NULL)
1681 			bridge_delete_member(sc, bif, 1);
1682 
1683 		BRIDGE_UNLOCK(sc);
1684 		return;
1685 	}
1686 
1687 	/* Check if the interface is a span port */
1688 	mtx_lock(&bridge_list_mtx);
1689 	LIST_FOREACH(sc, &bridge_list, sc_list) {
1690 		BRIDGE_LOCK(sc);
1691 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1692 			if (ifp == bif->bif_ifp) {
1693 				bridge_delete_span(sc, bif);
1694 				break;
1695 			}
1696 
1697 		BRIDGE_UNLOCK(sc);
1698 	}
1699 	mtx_unlock(&bridge_list_mtx);
1700 }
1701 
1702 /*
1703  * bridge_init:
1704  *
1705  *	Initialize a bridge interface.
1706  */
1707 static void
1708 bridge_init(void *xsc)
1709 {
1710 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1711 	struct ifnet *ifp = sc->sc_ifp;
1712 
1713 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1714 		return;
1715 
1716 	BRIDGE_LOCK(sc);
1717 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1718 	    bridge_timer, sc);
1719 
1720 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1721 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1722 
1723 	BRIDGE_UNLOCK(sc);
1724 }
1725 
1726 /*
1727  * bridge_stop:
1728  *
1729  *	Stop the bridge interface.
1730  */
1731 static void
1732 bridge_stop(struct ifnet *ifp, int disable)
1733 {
1734 	struct bridge_softc *sc = ifp->if_softc;
1735 
1736 	BRIDGE_LOCK_ASSERT(sc);
1737 
1738 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1739 		return;
1740 
1741 	callout_stop(&sc->sc_brcallout);
1742 	bstp_stop(&sc->sc_stp);
1743 
1744 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1745 
1746 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1747 }
1748 
1749 /*
1750  * bridge_enqueue:
1751  *
1752  *	Enqueue a packet on a bridge member interface.
1753  *
1754  */
1755 static void
1756 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1757 {
1758 	int len, err = 0;
1759 	short mflags;
1760 	struct mbuf *m0;
1761 
1762 	len = m->m_pkthdr.len;
1763 	mflags = m->m_flags;
1764 
1765 	/* We may be sending a fragment so traverse the mbuf */
1766 	for (; m; m = m0) {
1767 		m0 = m->m_nextpkt;
1768 		m->m_nextpkt = NULL;
1769 
1770 		/*
1771 		 * If underlying interface can not do VLAN tag insertion itself
1772 		 * then attach a packet tag that holds it.
1773 		 */
1774 		if ((m->m_flags & M_VLANTAG) &&
1775 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1776 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1777 			if (m == NULL) {
1778 				if_printf(dst_ifp,
1779 				    "unable to prepend VLAN header\n");
1780 				dst_ifp->if_oerrors++;
1781 				continue;
1782 			}
1783 			m->m_flags &= ~M_VLANTAG;
1784 		}
1785 
1786 		if (err == 0)
1787 			dst_ifp->if_transmit(dst_ifp, m);
1788 	}
1789 
1790 	if (err == 0) {
1791 		sc->sc_ifp->if_opackets++;
1792 		sc->sc_ifp->if_obytes += len;
1793 		if (mflags & M_MCAST)
1794 			sc->sc_ifp->if_omcasts++;
1795 	}
1796 }
1797 
1798 /*
1799  * bridge_dummynet:
1800  *
1801  * 	Receive a queued packet from dummynet and pass it on to the output
1802  * 	interface.
1803  *
1804  *	The mbuf has the Ethernet header already attached.
1805  */
1806 static void
1807 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1808 {
1809 	struct bridge_softc *sc;
1810 
1811 	sc = ifp->if_bridge;
1812 
1813 	/*
1814 	 * The packet didnt originate from a member interface. This should only
1815 	 * ever happen if a member interface is removed while packets are
1816 	 * queued for it.
1817 	 */
1818 	if (sc == NULL) {
1819 		m_freem(m);
1820 		return;
1821 	}
1822 
1823 	if (PFIL_HOOKED(&V_inet_pfil_hook)
1824 #ifdef INET6
1825 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1826 #endif
1827 	    ) {
1828 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1829 			return;
1830 		if (m == NULL)
1831 			return;
1832 	}
1833 
1834 	bridge_enqueue(sc, ifp, m);
1835 }
1836 
1837 /*
1838  * bridge_output:
1839  *
1840  *	Send output from a bridge member interface.  This
1841  *	performs the bridging function for locally originated
1842  *	packets.
1843  *
1844  *	The mbuf has the Ethernet header already attached.  We must
1845  *	enqueue or free the mbuf before returning.
1846  */
1847 static int
1848 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1849     struct rtentry *rt)
1850 {
1851 	struct ether_header *eh;
1852 	struct ifnet *dst_if;
1853 	struct bridge_softc *sc;
1854 	uint16_t vlan;
1855 
1856 	if (m->m_len < ETHER_HDR_LEN) {
1857 		m = m_pullup(m, ETHER_HDR_LEN);
1858 		if (m == NULL)
1859 			return (0);
1860 	}
1861 
1862 	eh = mtod(m, struct ether_header *);
1863 	sc = ifp->if_bridge;
1864 	vlan = VLANTAGOF(m);
1865 
1866 	BRIDGE_LOCK(sc);
1867 
1868 	/*
1869 	 * If bridge is down, but the original output interface is up,
1870 	 * go ahead and send out that interface.  Otherwise, the packet
1871 	 * is dropped below.
1872 	 */
1873 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1874 		dst_if = ifp;
1875 		goto sendunicast;
1876 	}
1877 
1878 	/*
1879 	 * If the packet is a multicast, or we don't know a better way to
1880 	 * get there, send to all interfaces.
1881 	 */
1882 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
1883 		dst_if = NULL;
1884 	else
1885 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1886 	if (dst_if == NULL) {
1887 		struct bridge_iflist *bif;
1888 		struct mbuf *mc;
1889 		int error = 0, used = 0;
1890 
1891 		bridge_span(sc, m);
1892 
1893 		BRIDGE_LOCK2REF(sc, error);
1894 		if (error) {
1895 			m_freem(m);
1896 			return (0);
1897 		}
1898 
1899 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1900 			dst_if = bif->bif_ifp;
1901 
1902 			if (dst_if->if_type == IFT_GIF)
1903 				continue;
1904 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1905 				continue;
1906 
1907 			/*
1908 			 * If this is not the original output interface,
1909 			 * and the interface is participating in spanning
1910 			 * tree, make sure the port is in a state that
1911 			 * allows forwarding.
1912 			 */
1913 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
1914 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1915 				continue;
1916 
1917 			if (LIST_NEXT(bif, bif_next) == NULL) {
1918 				used = 1;
1919 				mc = m;
1920 			} else {
1921 				mc = m_copypacket(m, M_DONTWAIT);
1922 				if (mc == NULL) {
1923 					sc->sc_ifp->if_oerrors++;
1924 					continue;
1925 				}
1926 			}
1927 
1928 			bridge_enqueue(sc, dst_if, mc);
1929 		}
1930 		if (used == 0)
1931 			m_freem(m);
1932 		BRIDGE_UNREF(sc);
1933 		return (0);
1934 	}
1935 
1936 sendunicast:
1937 	/*
1938 	 * XXX Spanning tree consideration here?
1939 	 */
1940 
1941 	bridge_span(sc, m);
1942 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1943 		m_freem(m);
1944 		BRIDGE_UNLOCK(sc);
1945 		return (0);
1946 	}
1947 
1948 	BRIDGE_UNLOCK(sc);
1949 	bridge_enqueue(sc, dst_if, m);
1950 	return (0);
1951 }
1952 
1953 /*
1954  * bridge_start:
1955  *
1956  *	Start output on a bridge.
1957  *
1958  */
1959 static void
1960 bridge_start(struct ifnet *ifp)
1961 {
1962 	struct bridge_softc *sc;
1963 	struct mbuf *m;
1964 	struct ether_header *eh;
1965 	struct ifnet *dst_if;
1966 
1967 	sc = ifp->if_softc;
1968 
1969 	ifp->if_drv_flags |= IFF_DRV_OACTIVE;
1970 	for (;;) {
1971 		IFQ_DEQUEUE(&ifp->if_snd, m);
1972 		if (m == 0)
1973 			break;
1974 		ETHER_BPF_MTAP(ifp, m);
1975 
1976 		eh = mtod(m, struct ether_header *);
1977 		dst_if = NULL;
1978 
1979 		BRIDGE_LOCK(sc);
1980 		if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
1981 			dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1);
1982 		}
1983 
1984 		if (dst_if == NULL)
1985 			bridge_broadcast(sc, ifp, m, 0);
1986 		else {
1987 			BRIDGE_UNLOCK(sc);
1988 			bridge_enqueue(sc, dst_if, m);
1989 		}
1990 	}
1991 	ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
1992 }
1993 
1994 /*
1995  * bridge_forward:
1996  *
1997  *	The forwarding function of the bridge.
1998  *
1999  *	NOTE: Releases the lock on return.
2000  */
2001 static void
2002 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2003     struct mbuf *m)
2004 {
2005 	struct bridge_iflist *dbif;
2006 	struct ifnet *src_if, *dst_if, *ifp;
2007 	struct ether_header *eh;
2008 	uint16_t vlan;
2009 	uint8_t *dst;
2010 	int error;
2011 
2012 	src_if = m->m_pkthdr.rcvif;
2013 	ifp = sc->sc_ifp;
2014 
2015 	ifp->if_ipackets++;
2016 	ifp->if_ibytes += m->m_pkthdr.len;
2017 	vlan = VLANTAGOF(m);
2018 
2019 	if ((sbif->bif_flags & IFBIF_STP) &&
2020 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2021 		goto drop;
2022 
2023 	eh = mtod(m, struct ether_header *);
2024 	dst = eh->ether_dhost;
2025 
2026 	/* If the interface is learning, record the address. */
2027 	if (sbif->bif_flags & IFBIF_LEARNING) {
2028 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2029 		    sbif, 0, IFBAF_DYNAMIC);
2030 		/*
2031 		 * If the interface has addresses limits then deny any source
2032 		 * that is not in the cache.
2033 		 */
2034 		if (error && sbif->bif_addrmax)
2035 			goto drop;
2036 	}
2037 
2038 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2039 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2040 		goto drop;
2041 
2042 	/*
2043 	 * At this point, the port either doesn't participate
2044 	 * in spanning tree or it is in the forwarding state.
2045 	 */
2046 
2047 	/*
2048 	 * If the packet is unicast, destined for someone on
2049 	 * "this" side of the bridge, drop it.
2050 	 */
2051 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2052 		dst_if = bridge_rtlookup(sc, dst, vlan);
2053 		if (src_if == dst_if)
2054 			goto drop;
2055 	} else {
2056 		/*
2057 		 * Check if its a reserved multicast address, any address
2058 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2059 		 * bridge.
2060 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2061 		 */
2062 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2063 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2064 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2065 			goto drop;
2066 
2067 		/* ...forward it to all interfaces. */
2068 		ifp->if_imcasts++;
2069 		dst_if = NULL;
2070 	}
2071 
2072 	/*
2073 	 * If we have a destination interface which is a member of our bridge,
2074 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2075 	 * For broadcast or multicast packets, don't bother because it will
2076 	 * be reinjected into ether_input. We do this before we pass the packets
2077 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2078 	 * drop the packet, or possibly modify it, making it difficult to debug
2079 	 * firewall issues on the bridge.
2080 	 */
2081 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2082 		ETHER_BPF_MTAP(ifp, m);
2083 
2084 	/* run the packet filter */
2085 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2086 #ifdef INET6
2087 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2088 #endif
2089 	    ) {
2090 		BRIDGE_UNLOCK(sc);
2091 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2092 			return;
2093 		if (m == NULL)
2094 			return;
2095 		BRIDGE_LOCK(sc);
2096 	}
2097 
2098 	if (dst_if == NULL) {
2099 		bridge_broadcast(sc, src_if, m, 1);
2100 		return;
2101 	}
2102 
2103 	/*
2104 	 * At this point, we're dealing with a unicast frame
2105 	 * going to a different interface.
2106 	 */
2107 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2108 		goto drop;
2109 
2110 	dbif = bridge_lookup_member_if(sc, dst_if);
2111 	if (dbif == NULL)
2112 		/* Not a member of the bridge (anymore?) */
2113 		goto drop;
2114 
2115 	/* Private segments can not talk to each other */
2116 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2117 		goto drop;
2118 
2119 	if ((dbif->bif_flags & IFBIF_STP) &&
2120 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2121 		goto drop;
2122 
2123 	BRIDGE_UNLOCK(sc);
2124 
2125 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2126 #ifdef INET6
2127 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2128 #endif
2129 	    ) {
2130 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2131 			return;
2132 		if (m == NULL)
2133 			return;
2134 	}
2135 
2136 	bridge_enqueue(sc, dst_if, m);
2137 	return;
2138 
2139 drop:
2140 	BRIDGE_UNLOCK(sc);
2141 	m_freem(m);
2142 }
2143 
2144 #if defined(INET) || defined(INET6)
2145 int (*carp_forus_p)(struct carp_if *, u_char *);
2146 #endif
2147 
2148 /*
2149  * bridge_input:
2150  *
2151  *	Receive input from a member interface.  Queue the packet for
2152  *	bridging if it is not for us.
2153  */
2154 static struct mbuf *
2155 bridge_input(struct ifnet *ifp, struct mbuf *m)
2156 {
2157 	struct bridge_softc *sc = ifp->if_bridge;
2158 	struct bridge_iflist *bif, *bif2;
2159 	struct ifnet *bifp;
2160 	struct ether_header *eh;
2161 	struct mbuf *mc, *mc2;
2162 	uint16_t vlan;
2163 	int error;
2164 
2165 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2166 		return (m);
2167 
2168 	bifp = sc->sc_ifp;
2169 	vlan = VLANTAGOF(m);
2170 
2171 	/*
2172 	 * Implement support for bridge monitoring. If this flag has been
2173 	 * set on this interface, discard the packet once we push it through
2174 	 * the bpf(4) machinery, but before we do, increment the byte and
2175 	 * packet counters associated with this interface.
2176 	 */
2177 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2178 		m->m_pkthdr.rcvif  = bifp;
2179 		ETHER_BPF_MTAP(bifp, m);
2180 		bifp->if_ipackets++;
2181 		bifp->if_ibytes += m->m_pkthdr.len;
2182 		m_freem(m);
2183 		return (NULL);
2184 	}
2185 	BRIDGE_LOCK(sc);
2186 	bif = bridge_lookup_member_if(sc, ifp);
2187 	if (bif == NULL) {
2188 		BRIDGE_UNLOCK(sc);
2189 		return (m);
2190 	}
2191 
2192 	eh = mtod(m, struct ether_header *);
2193 
2194 	bridge_span(sc, m);
2195 
2196 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2197 		/* Tap off 802.1D packets; they do not get forwarded. */
2198 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2199 		    ETHER_ADDR_LEN) == 0) {
2200 			m = bstp_input(&bif->bif_stp, ifp, m);
2201 			if (m == NULL) {
2202 				BRIDGE_UNLOCK(sc);
2203 				return (NULL);
2204 			}
2205 		}
2206 
2207 		if ((bif->bif_flags & IFBIF_STP) &&
2208 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2209 			BRIDGE_UNLOCK(sc);
2210 			return (m);
2211 		}
2212 
2213 		/*
2214 		 * Make a deep copy of the packet and enqueue the copy
2215 		 * for bridge processing; return the original packet for
2216 		 * local processing.
2217 		 */
2218 		mc = m_dup(m, M_DONTWAIT);
2219 		if (mc == NULL) {
2220 			BRIDGE_UNLOCK(sc);
2221 			return (m);
2222 		}
2223 
2224 		/* Perform the bridge forwarding function with the copy. */
2225 		bridge_forward(sc, bif, mc);
2226 
2227 		/*
2228 		 * Reinject the mbuf as arriving on the bridge so we have a
2229 		 * chance at claiming multicast packets. We can not loop back
2230 		 * here from ether_input as a bridge is never a member of a
2231 		 * bridge.
2232 		 */
2233 		KASSERT(bifp->if_bridge == NULL,
2234 		    ("loop created in bridge_input"));
2235 		mc2 = m_dup(m, M_DONTWAIT);
2236 		if (mc2 != NULL) {
2237 			/* Keep the layer3 header aligned */
2238 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2239 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2240 		}
2241 		if (mc2 != NULL) {
2242 			mc2->m_pkthdr.rcvif = bifp;
2243 			(*bifp->if_input)(bifp, mc2);
2244 		}
2245 
2246 		/* Return the original packet for local processing. */
2247 		return (m);
2248 	}
2249 
2250 	if ((bif->bif_flags & IFBIF_STP) &&
2251 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2252 		BRIDGE_UNLOCK(sc);
2253 		return (m);
2254 	}
2255 
2256 #if (defined(INET) || defined(INET6))
2257 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2258 	|| ((iface)->if_carp \
2259 	    && (*carp_forus_p)((iface)->if_carp, eh->ether_dhost))
2260 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2261 	|| ((iface)->if_carp \
2262 	    && (*carp_forus_p)((iface)->if_carp, eh->ether_shost))
2263 #else
2264 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2265 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2266 #endif
2267 
2268 #ifdef INET6
2269 #   define OR_PFIL_HOOKED_INET6 \
2270 	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2271 #else
2272 #   define OR_PFIL_HOOKED_INET6
2273 #endif
2274 
2275 #define GRAB_OUR_PACKETS(iface) \
2276 	if ((iface)->if_type == IFT_GIF) \
2277 		continue; \
2278 	/* It is destined for us. */ \
2279 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2280 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2281 	    ) {								\
2282 		if ((iface)->if_type == IFT_BRIDGE) {			\
2283 			ETHER_BPF_MTAP(iface, m);			\
2284 			iface->if_ipackets++;				\
2285 			/* Filter on the physical interface. */		\
2286 			if (pfil_local_phys &&				\
2287 			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2288 			     OR_PFIL_HOOKED_INET6)) {			\
2289 				if (bridge_pfil(&m, NULL, ifp,		\
2290 				    PFIL_IN) != 0 || m == NULL) {	\
2291 					BRIDGE_UNLOCK(sc);		\
2292 					return (NULL);			\
2293 				}					\
2294 			}						\
2295 		}							\
2296 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2297 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2298 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2299 			if (error && bif->bif_addrmax) {		\
2300 				BRIDGE_UNLOCK(sc);			\
2301 				m_freem(m);				\
2302 				return (NULL);				\
2303 			}						\
2304 		}							\
2305 		m->m_pkthdr.rcvif = iface;				\
2306 		BRIDGE_UNLOCK(sc);					\
2307 		return (m);						\
2308 	}								\
2309 									\
2310 	/* We just received a packet that we sent out. */		\
2311 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2312 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2313 	    ) {								\
2314 		BRIDGE_UNLOCK(sc);					\
2315 		m_freem(m);						\
2316 		return (NULL);						\
2317 	}
2318 
2319 	/*
2320 	 * Unicast.  Make sure it's not for the bridge.
2321 	 */
2322 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2323 
2324 	/*
2325 	 * Give a chance for ifp at first priority. This will help when	the
2326 	 * packet comes through the interface like VLAN's with the same MACs
2327 	 * on several interfaces from the same bridge. This also will save
2328 	 * some CPU cycles in case the destination interface and the input
2329 	 * interface (eq ifp) are the same.
2330 	 */
2331 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2332 
2333 	/* Now check the all bridge members. */
2334 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2335 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2336 	}
2337 
2338 #undef OR_CARP_CHECK_WE_ARE_DST
2339 #undef OR_CARP_CHECK_WE_ARE_SRC
2340 #undef OR_PFIL_HOOKED_INET6
2341 #undef GRAB_OUR_PACKETS
2342 
2343 	/* Perform the bridge forwarding function. */
2344 	bridge_forward(sc, bif, m);
2345 
2346 	return (NULL);
2347 }
2348 
2349 /*
2350  * bridge_broadcast:
2351  *
2352  *	Send a frame to all interfaces that are members of
2353  *	the bridge, except for the one on which the packet
2354  *	arrived.
2355  *
2356  *	NOTE: Releases the lock on return.
2357  */
2358 static void
2359 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2360     struct mbuf *m, int runfilt)
2361 {
2362 	struct bridge_iflist *dbif, *sbif;
2363 	struct mbuf *mc;
2364 	struct ifnet *dst_if;
2365 	int error = 0, used = 0, i;
2366 
2367 	sbif = bridge_lookup_member_if(sc, src_if);
2368 
2369 	BRIDGE_LOCK2REF(sc, error);
2370 	if (error) {
2371 		m_freem(m);
2372 		return;
2373 	}
2374 
2375 	/* Filter on the bridge interface before broadcasting */
2376 	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2377 #ifdef INET6
2378 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2379 #endif
2380 	    )) {
2381 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2382 			goto out;
2383 		if (m == NULL)
2384 			goto out;
2385 	}
2386 
2387 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2388 		dst_if = dbif->bif_ifp;
2389 		if (dst_if == src_if)
2390 			continue;
2391 
2392 		/* Private segments can not talk to each other */
2393 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2394 			continue;
2395 
2396 		if ((dbif->bif_flags & IFBIF_STP) &&
2397 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2398 			continue;
2399 
2400 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2401 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2402 			continue;
2403 
2404 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2405 			continue;
2406 
2407 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2408 			mc = m;
2409 			used = 1;
2410 		} else {
2411 			mc = m_dup(m, M_DONTWAIT);
2412 			if (mc == NULL) {
2413 				sc->sc_ifp->if_oerrors++;
2414 				continue;
2415 			}
2416 		}
2417 
2418 		/*
2419 		 * Filter on the output interface. Pass a NULL bridge interface
2420 		 * pointer so we do not redundantly filter on the bridge for
2421 		 * each interface we broadcast on.
2422 		 */
2423 		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2424 #ifdef INET6
2425 		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2426 #endif
2427 		    )) {
2428 			if (used == 0) {
2429 				/* Keep the layer3 header aligned */
2430 				i = min(mc->m_pkthdr.len, max_protohdr);
2431 				mc = m_copyup(mc, i, ETHER_ALIGN);
2432 				if (mc == NULL) {
2433 					sc->sc_ifp->if_oerrors++;
2434 					continue;
2435 				}
2436 			}
2437 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2438 				continue;
2439 			if (mc == NULL)
2440 				continue;
2441 		}
2442 
2443 		bridge_enqueue(sc, dst_if, mc);
2444 	}
2445 	if (used == 0)
2446 		m_freem(m);
2447 
2448 out:
2449 	BRIDGE_UNREF(sc);
2450 }
2451 
2452 /*
2453  * bridge_span:
2454  *
2455  *	Duplicate a packet out one or more interfaces that are in span mode,
2456  *	the original mbuf is unmodified.
2457  */
2458 static void
2459 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2460 {
2461 	struct bridge_iflist *bif;
2462 	struct ifnet *dst_if;
2463 	struct mbuf *mc;
2464 
2465 	if (LIST_EMPTY(&sc->sc_spanlist))
2466 		return;
2467 
2468 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2469 		dst_if = bif->bif_ifp;
2470 
2471 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2472 			continue;
2473 
2474 		mc = m_copypacket(m, M_DONTWAIT);
2475 		if (mc == NULL) {
2476 			sc->sc_ifp->if_oerrors++;
2477 			continue;
2478 		}
2479 
2480 		bridge_enqueue(sc, dst_if, mc);
2481 	}
2482 }
2483 
2484 /*
2485  * bridge_rtupdate:
2486  *
2487  *	Add a bridge routing entry.
2488  */
2489 static int
2490 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2491     struct bridge_iflist *bif, int setflags, uint8_t flags)
2492 {
2493 	struct bridge_rtnode *brt;
2494 	int error;
2495 
2496 	BRIDGE_LOCK_ASSERT(sc);
2497 
2498 	/* Check the source address is valid and not multicast. */
2499 	if (ETHER_IS_MULTICAST(dst) ||
2500 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2501 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2502 		return (EINVAL);
2503 
2504 	/* 802.1p frames map to vlan 1 */
2505 	if (vlan == 0)
2506 		vlan = 1;
2507 
2508 	/*
2509 	 * A route for this destination might already exist.  If so,
2510 	 * update it, otherwise create a new one.
2511 	 */
2512 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2513 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2514 			sc->sc_brtexceeded++;
2515 			return (ENOSPC);
2516 		}
2517 		/* Check per interface address limits (if enabled) */
2518 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2519 			bif->bif_addrexceeded++;
2520 			return (ENOSPC);
2521 		}
2522 
2523 		/*
2524 		 * Allocate a new bridge forwarding node, and
2525 		 * initialize the expiration time and Ethernet
2526 		 * address.
2527 		 */
2528 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2529 		if (brt == NULL)
2530 			return (ENOMEM);
2531 
2532 		if (bif->bif_flags & IFBIF_STICKY)
2533 			brt->brt_flags = IFBAF_STICKY;
2534 		else
2535 			brt->brt_flags = IFBAF_DYNAMIC;
2536 
2537 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2538 		brt->brt_vlan = vlan;
2539 
2540 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2541 			uma_zfree(bridge_rtnode_zone, brt);
2542 			return (error);
2543 		}
2544 		brt->brt_dst = bif;
2545 		bif->bif_addrcnt++;
2546 	}
2547 
2548 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2549 	    brt->brt_dst != bif) {
2550 		brt->brt_dst->bif_addrcnt--;
2551 		brt->brt_dst = bif;
2552 		brt->brt_dst->bif_addrcnt++;
2553 	}
2554 
2555 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2556 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2557 	if (setflags)
2558 		brt->brt_flags = flags;
2559 
2560 	return (0);
2561 }
2562 
2563 /*
2564  * bridge_rtlookup:
2565  *
2566  *	Lookup the destination interface for an address.
2567  */
2568 static struct ifnet *
2569 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2570 {
2571 	struct bridge_rtnode *brt;
2572 
2573 	BRIDGE_LOCK_ASSERT(sc);
2574 
2575 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2576 		return (NULL);
2577 
2578 	return (brt->brt_ifp);
2579 }
2580 
2581 /*
2582  * bridge_rttrim:
2583  *
2584  *	Trim the routine table so that we have a number
2585  *	of routing entries less than or equal to the
2586  *	maximum number.
2587  */
2588 static void
2589 bridge_rttrim(struct bridge_softc *sc)
2590 {
2591 	struct bridge_rtnode *brt, *nbrt;
2592 
2593 	BRIDGE_LOCK_ASSERT(sc);
2594 
2595 	/* Make sure we actually need to do this. */
2596 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2597 		return;
2598 
2599 	/* Force an aging cycle; this might trim enough addresses. */
2600 	bridge_rtage(sc);
2601 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2602 		return;
2603 
2604 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2605 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2606 			bridge_rtnode_destroy(sc, brt);
2607 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2608 				return;
2609 		}
2610 	}
2611 }
2612 
2613 /*
2614  * bridge_timer:
2615  *
2616  *	Aging timer for the bridge.
2617  */
2618 static void
2619 bridge_timer(void *arg)
2620 {
2621 	struct bridge_softc *sc = arg;
2622 
2623 	BRIDGE_LOCK_ASSERT(sc);
2624 
2625 	bridge_rtage(sc);
2626 
2627 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2628 		callout_reset(&sc->sc_brcallout,
2629 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2630 }
2631 
2632 /*
2633  * bridge_rtage:
2634  *
2635  *	Perform an aging cycle.
2636  */
2637 static void
2638 bridge_rtage(struct bridge_softc *sc)
2639 {
2640 	struct bridge_rtnode *brt, *nbrt;
2641 
2642 	BRIDGE_LOCK_ASSERT(sc);
2643 
2644 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2645 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2646 			if (time_uptime >= brt->brt_expire)
2647 				bridge_rtnode_destroy(sc, brt);
2648 		}
2649 	}
2650 }
2651 
2652 /*
2653  * bridge_rtflush:
2654  *
2655  *	Remove all dynamic addresses from the bridge.
2656  */
2657 static void
2658 bridge_rtflush(struct bridge_softc *sc, int full)
2659 {
2660 	struct bridge_rtnode *brt, *nbrt;
2661 
2662 	BRIDGE_LOCK_ASSERT(sc);
2663 
2664 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2665 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2666 			bridge_rtnode_destroy(sc, brt);
2667 	}
2668 }
2669 
2670 /*
2671  * bridge_rtdaddr:
2672  *
2673  *	Remove an address from the table.
2674  */
2675 static int
2676 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2677 {
2678 	struct bridge_rtnode *brt;
2679 	int found = 0;
2680 
2681 	BRIDGE_LOCK_ASSERT(sc);
2682 
2683 	/*
2684 	 * If vlan is zero then we want to delete for all vlans so the lookup
2685 	 * may return more than one.
2686 	 */
2687 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2688 		bridge_rtnode_destroy(sc, brt);
2689 		found = 1;
2690 	}
2691 
2692 	return (found ? 0 : ENOENT);
2693 }
2694 
2695 /*
2696  * bridge_rtdelete:
2697  *
2698  *	Delete routes to a speicifc member interface.
2699  */
2700 static void
2701 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2702 {
2703 	struct bridge_rtnode *brt, *nbrt;
2704 
2705 	BRIDGE_LOCK_ASSERT(sc);
2706 
2707 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2708 		if (brt->brt_ifp == ifp && (full ||
2709 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2710 			bridge_rtnode_destroy(sc, brt);
2711 	}
2712 }
2713 
2714 /*
2715  * bridge_rtable_init:
2716  *
2717  *	Initialize the route table for this bridge.
2718  */
2719 static int
2720 bridge_rtable_init(struct bridge_softc *sc)
2721 {
2722 	int i;
2723 
2724 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2725 	    M_DEVBUF, M_NOWAIT);
2726 	if (sc->sc_rthash == NULL)
2727 		return (ENOMEM);
2728 
2729 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2730 		LIST_INIT(&sc->sc_rthash[i]);
2731 
2732 	sc->sc_rthash_key = arc4random();
2733 
2734 	LIST_INIT(&sc->sc_rtlist);
2735 
2736 	return (0);
2737 }
2738 
2739 /*
2740  * bridge_rtable_fini:
2741  *
2742  *	Deconstruct the route table for this bridge.
2743  */
2744 static void
2745 bridge_rtable_fini(struct bridge_softc *sc)
2746 {
2747 
2748 	KASSERT(sc->sc_brtcnt == 0,
2749 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2750 	free(sc->sc_rthash, M_DEVBUF);
2751 }
2752 
2753 /*
2754  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2755  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2756  */
2757 #define	mix(a, b, c)							\
2758 do {									\
2759 	a -= b; a -= c; a ^= (c >> 13);					\
2760 	b -= c; b -= a; b ^= (a << 8);					\
2761 	c -= a; c -= b; c ^= (b >> 13);					\
2762 	a -= b; a -= c; a ^= (c >> 12);					\
2763 	b -= c; b -= a; b ^= (a << 16);					\
2764 	c -= a; c -= b; c ^= (b >> 5);					\
2765 	a -= b; a -= c; a ^= (c >> 3);					\
2766 	b -= c; b -= a; b ^= (a << 10);					\
2767 	c -= a; c -= b; c ^= (b >> 15);					\
2768 } while (/*CONSTCOND*/0)
2769 
2770 static __inline uint32_t
2771 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2772 {
2773 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2774 
2775 	b += addr[5] << 8;
2776 	b += addr[4];
2777 	a += addr[3] << 24;
2778 	a += addr[2] << 16;
2779 	a += addr[1] << 8;
2780 	a += addr[0];
2781 
2782 	mix(a, b, c);
2783 
2784 	return (c & BRIDGE_RTHASH_MASK);
2785 }
2786 
2787 #undef mix
2788 
2789 static int
2790 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2791 {
2792 	int i, d;
2793 
2794 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2795 		d = ((int)a[i]) - ((int)b[i]);
2796 	}
2797 
2798 	return (d);
2799 }
2800 
2801 /*
2802  * bridge_rtnode_lookup:
2803  *
2804  *	Look up a bridge route node for the specified destination. Compare the
2805  *	vlan id or if zero then just return the first match.
2806  */
2807 static struct bridge_rtnode *
2808 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2809 {
2810 	struct bridge_rtnode *brt;
2811 	uint32_t hash;
2812 	int dir;
2813 
2814 	BRIDGE_LOCK_ASSERT(sc);
2815 
2816 	hash = bridge_rthash(sc, addr);
2817 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2818 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2819 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2820 			return (brt);
2821 		if (dir > 0)
2822 			return (NULL);
2823 	}
2824 
2825 	return (NULL);
2826 }
2827 
2828 /*
2829  * bridge_rtnode_insert:
2830  *
2831  *	Insert the specified bridge node into the route table.  We
2832  *	assume the entry is not already in the table.
2833  */
2834 static int
2835 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2836 {
2837 	struct bridge_rtnode *lbrt;
2838 	uint32_t hash;
2839 	int dir;
2840 
2841 	BRIDGE_LOCK_ASSERT(sc);
2842 
2843 	hash = bridge_rthash(sc, brt->brt_addr);
2844 
2845 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2846 	if (lbrt == NULL) {
2847 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2848 		goto out;
2849 	}
2850 
2851 	do {
2852 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2853 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2854 			return (EEXIST);
2855 		if (dir > 0) {
2856 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2857 			goto out;
2858 		}
2859 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2860 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2861 			goto out;
2862 		}
2863 		lbrt = LIST_NEXT(lbrt, brt_hash);
2864 	} while (lbrt != NULL);
2865 
2866 #ifdef DIAGNOSTIC
2867 	panic("bridge_rtnode_insert: impossible");
2868 #endif
2869 
2870 out:
2871 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2872 	sc->sc_brtcnt++;
2873 
2874 	return (0);
2875 }
2876 
2877 /*
2878  * bridge_rtnode_destroy:
2879  *
2880  *	Destroy a bridge rtnode.
2881  */
2882 static void
2883 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2884 {
2885 	BRIDGE_LOCK_ASSERT(sc);
2886 
2887 	LIST_REMOVE(brt, brt_hash);
2888 
2889 	LIST_REMOVE(brt, brt_list);
2890 	sc->sc_brtcnt--;
2891 	brt->brt_dst->bif_addrcnt--;
2892 	uma_zfree(bridge_rtnode_zone, brt);
2893 }
2894 
2895 /*
2896  * bridge_rtable_expire:
2897  *
2898  *	Set the expiry time for all routes on an interface.
2899  */
2900 static void
2901 bridge_rtable_expire(struct ifnet *ifp, int age)
2902 {
2903 	struct bridge_softc *sc = ifp->if_bridge;
2904 	struct bridge_rtnode *brt;
2905 
2906 	BRIDGE_LOCK(sc);
2907 
2908 	/*
2909 	 * If the age is zero then flush, otherwise set all the expiry times to
2910 	 * age for the interface
2911 	 */
2912 	if (age == 0)
2913 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2914 	else {
2915 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2916 			/* Cap the expiry time to 'age' */
2917 			if (brt->brt_ifp == ifp &&
2918 			    brt->brt_expire > time_uptime + age &&
2919 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2920 				brt->brt_expire = time_uptime + age;
2921 		}
2922 	}
2923 	BRIDGE_UNLOCK(sc);
2924 }
2925 
2926 /*
2927  * bridge_state_change:
2928  *
2929  *	Callback from the bridgestp code when a port changes states.
2930  */
2931 static void
2932 bridge_state_change(struct ifnet *ifp, int state)
2933 {
2934 	struct bridge_softc *sc = ifp->if_bridge;
2935 	static const char *stpstates[] = {
2936 		"disabled",
2937 		"listening",
2938 		"learning",
2939 		"forwarding",
2940 		"blocking",
2941 		"discarding"
2942 	};
2943 
2944 	if (log_stp)
2945 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
2946 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
2947 }
2948 
2949 /*
2950  * Send bridge packets through pfil if they are one of the types pfil can deal
2951  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
2952  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
2953  * that interface.
2954  */
2955 static int
2956 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
2957 {
2958 	int snap, error, i, hlen;
2959 	struct ether_header *eh1, eh2;
2960 	struct ip_fw_args args;
2961 	struct ip *ip;
2962 	struct llc llc1;
2963 	u_int16_t ether_type;
2964 
2965 	snap = 0;
2966 	error = -1;	/* Default error if not error == 0 */
2967 
2968 #if 0
2969 	/* we may return with the IP fields swapped, ensure its not shared */
2970 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
2971 #endif
2972 
2973 	if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
2974 		return (0); /* filtering is disabled */
2975 
2976 	i = min((*mp)->m_pkthdr.len, max_protohdr);
2977 	if ((*mp)->m_len < i) {
2978 	    *mp = m_pullup(*mp, i);
2979 	    if (*mp == NULL) {
2980 		printf("%s: m_pullup failed\n", __func__);
2981 		return (-1);
2982 	    }
2983 	}
2984 
2985 	eh1 = mtod(*mp, struct ether_header *);
2986 	ether_type = ntohs(eh1->ether_type);
2987 
2988 	/*
2989 	 * Check for SNAP/LLC.
2990 	 */
2991 	if (ether_type < ETHERMTU) {
2992 		struct llc *llc2 = (struct llc *)(eh1 + 1);
2993 
2994 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
2995 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
2996 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
2997 		    llc2->llc_control == LLC_UI) {
2998 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
2999 			snap = 1;
3000 		}
3001 	}
3002 
3003 	/*
3004 	 * If we're trying to filter bridge traffic, don't look at anything
3005 	 * other than IP and ARP traffic.  If the filter doesn't understand
3006 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3007 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3008 	 * but of course we don't have an AppleTalk filter to begin with.
3009 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3010 	 * ARP traffic.)
3011 	 */
3012 	switch (ether_type) {
3013 		case ETHERTYPE_ARP:
3014 		case ETHERTYPE_REVARP:
3015 			if (pfil_ipfw_arp == 0)
3016 				return (0); /* Automatically pass */
3017 			break;
3018 
3019 		case ETHERTYPE_IP:
3020 #ifdef INET6
3021 		case ETHERTYPE_IPV6:
3022 #endif /* INET6 */
3023 			break;
3024 		default:
3025 			/*
3026 			 * Check to see if the user wants to pass non-ip
3027 			 * packets, these will not be checked by pfil(9) and
3028 			 * passed unconditionally so the default is to drop.
3029 			 */
3030 			if (pfil_onlyip)
3031 				goto bad;
3032 	}
3033 
3034 	/* Strip off the Ethernet header and keep a copy. */
3035 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3036 	m_adj(*mp, ETHER_HDR_LEN);
3037 
3038 	/* Strip off snap header, if present */
3039 	if (snap) {
3040 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3041 		m_adj(*mp, sizeof(struct llc));
3042 	}
3043 
3044 	/*
3045 	 * Check the IP header for alignment and errors
3046 	 */
3047 	if (dir == PFIL_IN) {
3048 		switch (ether_type) {
3049 			case ETHERTYPE_IP:
3050 				error = bridge_ip_checkbasic(mp);
3051 				break;
3052 #ifdef INET6
3053 			case ETHERTYPE_IPV6:
3054 				error = bridge_ip6_checkbasic(mp);
3055 				break;
3056 #endif /* INET6 */
3057 			default:
3058 				error = 0;
3059 		}
3060 		if (error)
3061 			goto bad;
3062 	}
3063 
3064 	/* XXX this section is also in if_ethersubr.c */
3065 	// XXX PFIL_OUT or DIR_OUT ?
3066 	if (V_ip_fw_chk_ptr && pfil_ipfw != 0 &&
3067 			dir == PFIL_OUT && ifp != NULL) {
3068 		struct m_tag *mtag;
3069 
3070 		error = -1;
3071 		/* fetch the start point from existing tags, if any */
3072 		mtag = m_tag_locate(*mp, MTAG_IPFW_RULE, 0, NULL);
3073 		if (mtag == NULL) {
3074 			args.rule.slot = 0;
3075 		} else {
3076 			struct ipfw_rule_ref *r;
3077 
3078 			/* XXX can we free the tag after use ? */
3079 			mtag->m_tag_id = PACKET_TAG_NONE;
3080 			r = (struct ipfw_rule_ref *)(mtag + 1);
3081 			/* packet already partially processed ? */
3082 			if (r->info & IPFW_ONEPASS)
3083 				goto ipfwpass;
3084 			args.rule = *r;
3085 		}
3086 
3087 		args.m = *mp;
3088 		args.oif = ifp;
3089 		args.next_hop = NULL;
3090 		args.eh = &eh2;
3091 		args.inp = NULL;	/* used by ipfw uid/gid/jail rules */
3092 		i = V_ip_fw_chk_ptr(&args);
3093 		*mp = args.m;
3094 
3095 		if (*mp == NULL)
3096 			return (error);
3097 
3098 		if (ip_dn_io_ptr && (i == IP_FW_DUMMYNET)) {
3099 
3100 			/* put the Ethernet header back on */
3101 			M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3102 			if (*mp == NULL)
3103 				return (error);
3104 			bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3105 
3106 			/*
3107 			 * Pass the pkt to dummynet, which consumes it. The
3108 			 * packet will return to us via bridge_dummynet().
3109 			 */
3110 			args.oif = ifp;
3111 			ip_dn_io_ptr(mp, DIR_FWD | PROTO_IFB, &args);
3112 			return (error);
3113 		}
3114 
3115 		if (i != IP_FW_PASS) /* drop */
3116 			goto bad;
3117 	}
3118 
3119 ipfwpass:
3120 	error = 0;
3121 
3122 	/*
3123 	 * Run the packet through pfil
3124 	 */
3125 	switch (ether_type) {
3126 	case ETHERTYPE_IP:
3127 		/*
3128 		 * before calling the firewall, swap fields the same as
3129 		 * IP does. here we assume the header is contiguous
3130 		 */
3131 		ip = mtod(*mp, struct ip *);
3132 
3133 		ip->ip_len = ntohs(ip->ip_len);
3134 		ip->ip_off = ntohs(ip->ip_off);
3135 
3136 		/*
3137 		 * Run pfil on the member interface and the bridge, both can
3138 		 * be skipped by clearing pfil_member or pfil_bridge.
3139 		 *
3140 		 * Keep the order:
3141 		 *   in_if -> bridge_if -> out_if
3142 		 */
3143 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3144 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3145 					dir, NULL);
3146 
3147 		if (*mp == NULL || error != 0) /* filter may consume */
3148 			break;
3149 
3150 		if (pfil_member && ifp != NULL)
3151 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3152 					dir, NULL);
3153 
3154 		if (*mp == NULL || error != 0) /* filter may consume */
3155 			break;
3156 
3157 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3158 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3159 					dir, NULL);
3160 
3161 		if (*mp == NULL || error != 0) /* filter may consume */
3162 			break;
3163 
3164 		/* check if we need to fragment the packet */
3165 		if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3166 			i = (*mp)->m_pkthdr.len;
3167 			if (i > ifp->if_mtu) {
3168 				error = bridge_fragment(ifp, *mp, &eh2, snap,
3169 					    &llc1);
3170 				return (error);
3171 			}
3172 		}
3173 
3174 		/* Recalculate the ip checksum and restore byte ordering */
3175 		ip = mtod(*mp, struct ip *);
3176 		hlen = ip->ip_hl << 2;
3177 		if (hlen < sizeof(struct ip))
3178 			goto bad;
3179 		if (hlen > (*mp)->m_len) {
3180 			if ((*mp = m_pullup(*mp, hlen)) == 0)
3181 				goto bad;
3182 			ip = mtod(*mp, struct ip *);
3183 			if (ip == NULL)
3184 				goto bad;
3185 		}
3186 		ip->ip_len = htons(ip->ip_len);
3187 		ip->ip_off = htons(ip->ip_off);
3188 		ip->ip_sum = 0;
3189 		if (hlen == sizeof(struct ip))
3190 			ip->ip_sum = in_cksum_hdr(ip);
3191 		else
3192 			ip->ip_sum = in_cksum(*mp, hlen);
3193 
3194 		break;
3195 #ifdef INET6
3196 	case ETHERTYPE_IPV6:
3197 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3198 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3199 					dir, NULL);
3200 
3201 		if (*mp == NULL || error != 0) /* filter may consume */
3202 			break;
3203 
3204 		if (pfil_member && ifp != NULL)
3205 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3206 					dir, NULL);
3207 
3208 		if (*mp == NULL || error != 0) /* filter may consume */
3209 			break;
3210 
3211 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3212 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3213 					dir, NULL);
3214 		break;
3215 #endif
3216 	default:
3217 		error = 0;
3218 		break;
3219 	}
3220 
3221 	if (*mp == NULL)
3222 		return (error);
3223 	if (error != 0)
3224 		goto bad;
3225 
3226 	error = -1;
3227 
3228 	/*
3229 	 * Finally, put everything back the way it was and return
3230 	 */
3231 	if (snap) {
3232 		M_PREPEND(*mp, sizeof(struct llc), M_DONTWAIT);
3233 		if (*mp == NULL)
3234 			return (error);
3235 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3236 	}
3237 
3238 	M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3239 	if (*mp == NULL)
3240 		return (error);
3241 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3242 
3243 	return (0);
3244 
3245 bad:
3246 	m_freem(*mp);
3247 	*mp = NULL;
3248 	return (error);
3249 }
3250 
3251 /*
3252  * Perform basic checks on header size since
3253  * pfil assumes ip_input has already processed
3254  * it for it.  Cut-and-pasted from ip_input.c.
3255  * Given how simple the IPv6 version is,
3256  * does the IPv4 version really need to be
3257  * this complicated?
3258  *
3259  * XXX Should we update ipstat here, or not?
3260  * XXX Right now we update ipstat but not
3261  * XXX csum_counter.
3262  */
3263 static int
3264 bridge_ip_checkbasic(struct mbuf **mp)
3265 {
3266 	struct mbuf *m = *mp;
3267 	struct ip *ip;
3268 	int len, hlen;
3269 	u_short sum;
3270 
3271 	if (*mp == NULL)
3272 		return (-1);
3273 
3274 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3275 		if ((m = m_copyup(m, sizeof(struct ip),
3276 			(max_linkhdr + 3) & ~3)) == NULL) {
3277 			/* XXXJRT new stat, please */
3278 			KMOD_IPSTAT_INC(ips_toosmall);
3279 			goto bad;
3280 		}
3281 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3282 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3283 			KMOD_IPSTAT_INC(ips_toosmall);
3284 			goto bad;
3285 		}
3286 	}
3287 	ip = mtod(m, struct ip *);
3288 	if (ip == NULL) goto bad;
3289 
3290 	if (ip->ip_v != IPVERSION) {
3291 		KMOD_IPSTAT_INC(ips_badvers);
3292 		goto bad;
3293 	}
3294 	hlen = ip->ip_hl << 2;
3295 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3296 		KMOD_IPSTAT_INC(ips_badhlen);
3297 		goto bad;
3298 	}
3299 	if (hlen > m->m_len) {
3300 		if ((m = m_pullup(m, hlen)) == 0) {
3301 			KMOD_IPSTAT_INC(ips_badhlen);
3302 			goto bad;
3303 		}
3304 		ip = mtod(m, struct ip *);
3305 		if (ip == NULL) goto bad;
3306 	}
3307 
3308 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3309 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3310 	} else {
3311 		if (hlen == sizeof(struct ip)) {
3312 			sum = in_cksum_hdr(ip);
3313 		} else {
3314 			sum = in_cksum(m, hlen);
3315 		}
3316 	}
3317 	if (sum) {
3318 		KMOD_IPSTAT_INC(ips_badsum);
3319 		goto bad;
3320 	}
3321 
3322 	/* Retrieve the packet length. */
3323 	len = ntohs(ip->ip_len);
3324 
3325 	/*
3326 	 * Check for additional length bogosity
3327 	 */
3328 	if (len < hlen) {
3329 		KMOD_IPSTAT_INC(ips_badlen);
3330 		goto bad;
3331 	}
3332 
3333 	/*
3334 	 * Check that the amount of data in the buffers
3335 	 * is as at least much as the IP header would have us expect.
3336 	 * Drop packet if shorter than we expect.
3337 	 */
3338 	if (m->m_pkthdr.len < len) {
3339 		KMOD_IPSTAT_INC(ips_tooshort);
3340 		goto bad;
3341 	}
3342 
3343 	/* Checks out, proceed */
3344 	*mp = m;
3345 	return (0);
3346 
3347 bad:
3348 	*mp = m;
3349 	return (-1);
3350 }
3351 
3352 #ifdef INET6
3353 /*
3354  * Same as above, but for IPv6.
3355  * Cut-and-pasted from ip6_input.c.
3356  * XXX Should we update ip6stat, or not?
3357  */
3358 static int
3359 bridge_ip6_checkbasic(struct mbuf **mp)
3360 {
3361 	struct mbuf *m = *mp;
3362 	struct ip6_hdr *ip6;
3363 
3364 	/*
3365 	 * If the IPv6 header is not aligned, slurp it up into a new
3366 	 * mbuf with space for link headers, in the event we forward
3367 	 * it.  Otherwise, if it is aligned, make sure the entire base
3368 	 * IPv6 header is in the first mbuf of the chain.
3369 	 */
3370 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3371 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3372 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3373 			    (max_linkhdr + 3) & ~3)) == NULL) {
3374 			/* XXXJRT new stat, please */
3375 			V_ip6stat.ip6s_toosmall++;
3376 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3377 			goto bad;
3378 		}
3379 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3380 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3381 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3382 			V_ip6stat.ip6s_toosmall++;
3383 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3384 			goto bad;
3385 		}
3386 	}
3387 
3388 	ip6 = mtod(m, struct ip6_hdr *);
3389 
3390 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3391 		V_ip6stat.ip6s_badvers++;
3392 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3393 		goto bad;
3394 	}
3395 
3396 	/* Checks out, proceed */
3397 	*mp = m;
3398 	return (0);
3399 
3400 bad:
3401 	*mp = m;
3402 	return (-1);
3403 }
3404 #endif /* INET6 */
3405 
3406 /*
3407  * bridge_fragment:
3408  *
3409  *	Return a fragmented mbuf chain.
3410  */
3411 static int
3412 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3413     int snap, struct llc *llc)
3414 {
3415 	struct mbuf *m0;
3416 	struct ip *ip;
3417 	int error = -1;
3418 
3419 	if (m->m_len < sizeof(struct ip) &&
3420 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3421 		goto out;
3422 	ip = mtod(m, struct ip *);
3423 
3424 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist,
3425 		    CSUM_DELAY_IP);
3426 	if (error)
3427 		goto out;
3428 
3429 	/* walk the chain and re-add the Ethernet header */
3430 	for (m0 = m; m0; m0 = m0->m_nextpkt) {
3431 		if (error == 0) {
3432 			if (snap) {
3433 				M_PREPEND(m0, sizeof(struct llc), M_DONTWAIT);
3434 				if (m0 == NULL) {
3435 					error = ENOBUFS;
3436 					continue;
3437 				}
3438 				bcopy(llc, mtod(m0, caddr_t),
3439 				    sizeof(struct llc));
3440 			}
3441 			M_PREPEND(m0, ETHER_HDR_LEN, M_DONTWAIT);
3442 			if (m0 == NULL) {
3443 				error = ENOBUFS;
3444 				continue;
3445 			}
3446 			bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3447 		} else
3448 			m_freem(m);
3449 	}
3450 
3451 	if (error == 0)
3452 		KMOD_IPSTAT_INC(ips_fragmented);
3453 
3454 	return (error);
3455 
3456 out:
3457 	if (m != NULL)
3458 		m_freem(m);
3459 	return (error);
3460 }
3461