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: stable/10/sys/net/if_bridge.c 324116 2017-09-30 10:16:15Z kp $");
79
80#include "opt_inet.h"
81#include "opt_inet6.h"
82
83#include <sys/param.h>
84#include <sys/mbuf.h>
85#include <sys/malloc.h>
86#include <sys/protosw.h>
87#include <sys/systm.h>
88#include <sys/jail.h>
89#include <sys/time.h>
90#include <sys/socket.h> /* for net/if.h */
91#include <sys/sockio.h>
92#include <sys/ctype.h>  /* string functions */
93#include <sys/kernel.h>
94#include <sys/random.h>
95#include <sys/syslog.h>
96#include <sys/sysctl.h>
97#include <vm/uma.h>
98#include <sys/module.h>
99#include <sys/priv.h>
100#include <sys/proc.h>
101#include <sys/lock.h>
102#include <sys/mutex.h>
103
104#include <net/bpf.h>
105#include <net/if.h>
106#include <net/if_clone.h>
107#include <net/if_dl.h>
108#include <net/if_types.h>
109#include <net/if_var.h>
110#include <net/pfil.h>
111#include <net/vnet.h>
112
113#include <netinet/in.h> /* for struct arpcom */
114#include <netinet/in_systm.h>
115#include <netinet/in_var.h>
116#include <netinet/ip.h>
117#include <netinet/ip_var.h>
118#ifdef INET6
119#include <netinet/ip6.h>
120#include <netinet6/ip6_var.h>
121#include <netinet6/in6_ifattach.h>
122#endif
123#if defined(INET) || defined(INET6)
124#include <netinet/ip_carp.h>
125#endif
126#include <machine/in_cksum.h>
127#include <netinet/if_ether.h> /* for struct arpcom */
128#include <net/bridgestp.h>
129#include <net/if_bridgevar.h>
130#include <net/if_llc.h>
131#include <net/if_vlan_var.h>
132
133#include <net/route.h>
134
135/*
136 * Size of the route hash table.  Must be a power of two.
137 */
138#ifndef BRIDGE_RTHASH_SIZE
139#define	BRIDGE_RTHASH_SIZE		1024
140#endif
141
142#define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
143
144/*
145 * Default maximum number of addresses to cache.
146 */
147#ifndef BRIDGE_RTABLE_MAX
148#define	BRIDGE_RTABLE_MAX		2000
149#endif
150
151/*
152 * Timeout (in seconds) for entries learned dynamically.
153 */
154#ifndef BRIDGE_RTABLE_TIMEOUT
155#define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
156#endif
157
158/*
159 * Number of seconds between walks of the route list.
160 */
161#ifndef BRIDGE_RTABLE_PRUNE_PERIOD
162#define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
163#endif
164
165/*
166 * List of capabilities to possibly mask on the member interface.
167 */
168#define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
169					 IFCAP_TXCSUM_IPV6)
170
171/*
172 * List of capabilities to strip
173 */
174#define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
175
176/*
177 * Bridge interface list entry.
178 */
179struct 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 */
193struct 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 */
207struct 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
229static struct mtx 	bridge_list_mtx;
230eventhandler_tag	bridge_detach_cookie = NULL;
231
232int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
233
234uma_zone_t bridge_rtnode_zone;
235
236static int	bridge_clone_create(struct if_clone *, int, caddr_t);
237static void	bridge_clone_destroy(struct ifnet *);
238
239static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
240static void	bridge_mutecaps(struct bridge_softc *);
241static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
242		    int);
243static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
244static void	bridge_init(void *);
245static void	bridge_dummynet(struct mbuf *, struct ifnet *);
246static void	bridge_stop(struct ifnet *, int);
247static int	bridge_transmit(struct ifnet *, struct mbuf *);
248static void	bridge_qflush(struct ifnet *);
249static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
250static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
251		    struct rtentry *);
252static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
253		    struct mbuf *);
254static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
255
256static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
257		    struct mbuf *m);
258
259static void	bridge_timer(void *);
260
261static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
262		    struct mbuf *, int);
263static void	bridge_span(struct bridge_softc *, struct mbuf *);
264
265static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
266		    uint16_t, struct bridge_iflist *, int, uint8_t);
267static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
268		    uint16_t);
269static void	bridge_rttrim(struct bridge_softc *);
270static void	bridge_rtage(struct bridge_softc *);
271static void	bridge_rtflush(struct bridge_softc *, int);
272static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
273		    uint16_t);
274
275static void	bridge_rtable_init(struct bridge_softc *);
276static void	bridge_rtable_fini(struct bridge_softc *);
277
278static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
279static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
280		    const uint8_t *, uint16_t);
281static int	bridge_rtnode_insert(struct bridge_softc *,
282		    struct bridge_rtnode *);
283static void	bridge_rtnode_destroy(struct bridge_softc *,
284		    struct bridge_rtnode *);
285static void	bridge_rtable_expire(struct ifnet *, int);
286static void	bridge_state_change(struct ifnet *, int);
287
288static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
289		    const char *name);
290static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
291		    struct ifnet *ifp);
292static void	bridge_delete_member(struct bridge_softc *,
293		    struct bridge_iflist *, int);
294static void	bridge_delete_span(struct bridge_softc *,
295		    struct bridge_iflist *);
296
297static int	bridge_ioctl_add(struct bridge_softc *, void *);
298static int	bridge_ioctl_del(struct bridge_softc *, void *);
299static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
300static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
301static int	bridge_ioctl_scache(struct bridge_softc *, void *);
302static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
303static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
304static int	bridge_ioctl_rts(struct bridge_softc *, void *);
305static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
306static int	bridge_ioctl_sto(struct bridge_softc *, void *);
307static int	bridge_ioctl_gto(struct bridge_softc *, void *);
308static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
309static int	bridge_ioctl_flush(struct bridge_softc *, void *);
310static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
311static int	bridge_ioctl_spri(struct bridge_softc *, void *);
312static int	bridge_ioctl_ght(struct bridge_softc *, void *);
313static int	bridge_ioctl_sht(struct bridge_softc *, void *);
314static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
315static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
316static int	bridge_ioctl_gma(struct bridge_softc *, void *);
317static int	bridge_ioctl_sma(struct bridge_softc *, void *);
318static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
319static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
320static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
321static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
322static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
323static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
324static int	bridge_ioctl_grte(struct bridge_softc *, void *);
325static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
326static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
327static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
328static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
329		    int);
330static int	bridge_ip_checkbasic(struct mbuf **mp);
331#ifdef INET6
332static int	bridge_ip6_checkbasic(struct mbuf **mp);
333#endif /* INET6 */
334static int	bridge_fragment(struct ifnet *, struct mbuf **mp,
335		    struct ether_header *, int, struct llc *);
336static void	bridge_linkstate(struct ifnet *ifp);
337static void	bridge_linkcheck(struct bridge_softc *sc);
338
339extern void (*bridge_linkstate_p)(struct ifnet *ifp);
340
341/* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
342#define	VLANTAGOF(_m)	\
343    (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
344
345static struct bstp_cb_ops bridge_ops = {
346	.bcb_state = bridge_state_change,
347	.bcb_rtage = bridge_rtable_expire
348};
349
350SYSCTL_DECL(_net_link);
351static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
352
353static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */
354static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */
355static int pfil_member = 1; /* run pfil hooks on the member interface */
356static int pfil_ipfw = 0;   /* layer2 filter with ipfw */
357static int pfil_ipfw_arp = 0;   /* layer2 filter with ipfw */
358static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for
359                                   locally destined packets */
360static int log_stp   = 0;   /* log STP state changes */
361static int bridge_inherit_mac = 0;   /* share MAC with first bridge member */
362TUNABLE_INT("net.link.bridge.pfil_onlyip", &pfil_onlyip);
363SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
364    &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
365TUNABLE_INT("net.link.bridge.ipfw_arp", &pfil_ipfw_arp);
366SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
367    &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
368TUNABLE_INT("net.link.bridge.pfil_bridge", &pfil_bridge);
369SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
370    &pfil_bridge, 0, "Packet filter on the bridge interface");
371TUNABLE_INT("net.link.bridge.pfil_member", &pfil_member);
372SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
373    &pfil_member, 0, "Packet filter on the member interface");
374TUNABLE_INT("net.link.bridge.pfil_local_phys", &pfil_local_phys);
375SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
376    &pfil_local_phys, 0,
377    "Packet filter on the physical interface for locally destined packets");
378TUNABLE_INT("net.link.bridge.log_stp", &log_stp);
379SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
380    &log_stp, 0, "Log STP state changes");
381TUNABLE_INT("net.link.bridge.inherit_mac", &bridge_inherit_mac);
382SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, CTLFLAG_RW,
383    &bridge_inherit_mac, 0,
384    "Inherit MAC address from the first bridge member");
385
386static VNET_DEFINE(int, allow_llz_overlap) = 0;
387#define	V_allow_llz_overlap	VNET(allow_llz_overlap)
388SYSCTL_VNET_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap, CTLFLAG_RW,
389    &VNET_NAME(allow_llz_overlap), 0, "Allow overlap of link-local scope "
390    "zones of a bridge interface and the member interfaces");
391
392struct bridge_control {
393	int	(*bc_func)(struct bridge_softc *, void *);
394	int	bc_argsize;
395	int	bc_flags;
396};
397
398#define	BC_F_COPYIN		0x01	/* copy arguments in */
399#define	BC_F_COPYOUT		0x02	/* copy arguments out */
400#define	BC_F_SUSER		0x04	/* do super-user check */
401
402const struct bridge_control bridge_control_table[] = {
403	{ bridge_ioctl_add,		sizeof(struct ifbreq),
404	  BC_F_COPYIN|BC_F_SUSER },
405	{ bridge_ioctl_del,		sizeof(struct ifbreq),
406	  BC_F_COPYIN|BC_F_SUSER },
407
408	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
409	  BC_F_COPYIN|BC_F_COPYOUT },
410	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
411	  BC_F_COPYIN|BC_F_SUSER },
412
413	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
414	  BC_F_COPYIN|BC_F_SUSER },
415	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
416	  BC_F_COPYOUT },
417
418	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
419	  BC_F_COPYIN|BC_F_COPYOUT },
420	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
421	  BC_F_COPYIN|BC_F_COPYOUT },
422
423	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
424	  BC_F_COPYIN|BC_F_SUSER },
425
426	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
427	  BC_F_COPYIN|BC_F_SUSER },
428	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
429	  BC_F_COPYOUT },
430
431	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
432	  BC_F_COPYIN|BC_F_SUSER },
433
434	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
435	  BC_F_COPYIN|BC_F_SUSER },
436
437	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
438	  BC_F_COPYOUT },
439	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
440	  BC_F_COPYIN|BC_F_SUSER },
441
442	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
443	  BC_F_COPYOUT },
444	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
445	  BC_F_COPYIN|BC_F_SUSER },
446
447	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
448	  BC_F_COPYOUT },
449	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
450	  BC_F_COPYIN|BC_F_SUSER },
451
452	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
453	  BC_F_COPYOUT },
454	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
455	  BC_F_COPYIN|BC_F_SUSER },
456
457	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
458	  BC_F_COPYIN|BC_F_SUSER },
459
460	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
461	  BC_F_COPYIN|BC_F_SUSER },
462
463	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
464	  BC_F_COPYIN|BC_F_SUSER },
465	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
466	  BC_F_COPYIN|BC_F_SUSER },
467
468	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
469	  BC_F_COPYOUT },
470
471	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
472	  BC_F_COPYOUT },
473
474	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
475	  BC_F_COPYIN|BC_F_COPYOUT },
476
477	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
478	  BC_F_COPYIN|BC_F_SUSER },
479
480	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
481	  BC_F_COPYIN|BC_F_SUSER },
482
483	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
484	  BC_F_COPYIN|BC_F_SUSER },
485
486};
487const int bridge_control_table_size =
488    sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
489
490LIST_HEAD(, bridge_softc) bridge_list;
491
492static struct if_clone *bridge_cloner;
493static const char bridge_name[] = "bridge";
494
495static int
496bridge_modevent(module_t mod, int type, void *data)
497{
498
499	switch (type) {
500	case MOD_LOAD:
501		mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
502		bridge_cloner = if_clone_simple(bridge_name,
503		    bridge_clone_create, bridge_clone_destroy, 0);
504		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
505		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
506		    UMA_ALIGN_PTR, 0);
507		LIST_INIT(&bridge_list);
508		bridge_input_p = bridge_input;
509		bridge_output_p = bridge_output;
510		bridge_dn_p = bridge_dummynet;
511		bridge_linkstate_p = bridge_linkstate;
512		bridge_detach_cookie = EVENTHANDLER_REGISTER(
513		    ifnet_departure_event, bridge_ifdetach, NULL,
514		    EVENTHANDLER_PRI_ANY);
515		break;
516	case MOD_UNLOAD:
517		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
518		    bridge_detach_cookie);
519		if_clone_detach(bridge_cloner);
520		uma_zdestroy(bridge_rtnode_zone);
521		bridge_input_p = NULL;
522		bridge_output_p = NULL;
523		bridge_dn_p = NULL;
524		bridge_linkstate_p = NULL;
525		mtx_destroy(&bridge_list_mtx);
526		break;
527	default:
528		return (EOPNOTSUPP);
529	}
530	return (0);
531}
532
533static moduledata_t bridge_mod = {
534	"if_bridge",
535	bridge_modevent,
536	0
537};
538
539DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
540MODULE_VERSION(if_bridge, 1);
541MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
542
543/*
544 * handler for net.link.bridge.ipfw
545 */
546static int
547sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
548{
549	int enable = pfil_ipfw;
550	int error;
551
552	error = sysctl_handle_int(oidp, &enable, 0, req);
553	enable = (enable) ? 1 : 0;
554
555	if (enable != pfil_ipfw) {
556		pfil_ipfw = enable;
557
558		/*
559		 * Disable pfil so that ipfw doesnt run twice, if the user
560		 * really wants both then they can re-enable pfil_bridge and/or
561		 * pfil_member. Also allow non-ip packets as ipfw can filter by
562		 * layer2 type.
563		 */
564		if (pfil_ipfw) {
565			pfil_onlyip = 0;
566			pfil_bridge = 0;
567			pfil_member = 0;
568		}
569	}
570
571	return (error);
572}
573SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
574	    &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
575
576/*
577 * bridge_clone_create:
578 *
579 *	Create a new bridge instance.
580 */
581static int
582bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
583{
584	struct bridge_softc *sc, *sc2;
585	struct ifnet *bifp, *ifp;
586	int fb, retry;
587	unsigned long hostid;
588
589	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
590	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
591	if (ifp == NULL) {
592		free(sc, M_DEVBUF);
593		return (ENOSPC);
594	}
595
596	BRIDGE_LOCK_INIT(sc);
597	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
598	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
599
600	/* Initialize our routing table. */
601	bridge_rtable_init(sc);
602
603	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
604
605	LIST_INIT(&sc->sc_iflist);
606	LIST_INIT(&sc->sc_spanlist);
607
608	ifp->if_softc = sc;
609	if_initname(ifp, bridge_name, unit);
610	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
611	ifp->if_ioctl = bridge_ioctl;
612	ifp->if_transmit = bridge_transmit;
613	ifp->if_qflush = bridge_qflush;
614	ifp->if_init = bridge_init;
615	ifp->if_type = IFT_BRIDGE;
616
617	/*
618	 * Generate an ethernet address with a locally administered address.
619	 *
620	 * Since we are using random ethernet addresses for the bridge, it is
621	 * possible that we might have address collisions, so make sure that
622	 * this hardware address isn't already in use on another bridge.
623	 * The first try uses the hostid and falls back to arc4rand().
624	 */
625	fb = 0;
626	getcredhostid(curthread->td_ucred, &hostid);
627	do {
628		if (fb || hostid == 0) {
629			arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
630			sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
631			sc->sc_defaddr[0] |= 2;	/* set the LAA bit */
632		} else {
633			sc->sc_defaddr[0] = 0x2;
634			sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
635			sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
636			sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
637			sc->sc_defaddr[4] =  hostid        & 0xff;
638			sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
639		}
640
641		fb = 1;
642		retry = 0;
643		mtx_lock(&bridge_list_mtx);
644		LIST_FOREACH(sc2, &bridge_list, sc_list) {
645			bifp = sc2->sc_ifp;
646			if (memcmp(sc->sc_defaddr,
647			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
648				retry = 1;
649				break;
650			}
651		}
652		mtx_unlock(&bridge_list_mtx);
653	} while (retry == 1);
654
655	bstp_attach(&sc->sc_stp, &bridge_ops);
656	ether_ifattach(ifp, sc->sc_defaddr);
657	/* Now undo some of the damage... */
658	ifp->if_baudrate = 0;
659	ifp->if_type = IFT_BRIDGE;
660
661	mtx_lock(&bridge_list_mtx);
662	LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
663	mtx_unlock(&bridge_list_mtx);
664
665	return (0);
666}
667
668/*
669 * bridge_clone_destroy:
670 *
671 *	Destroy a bridge instance.
672 */
673static void
674bridge_clone_destroy(struct ifnet *ifp)
675{
676	struct bridge_softc *sc = ifp->if_softc;
677	struct bridge_iflist *bif;
678
679	BRIDGE_LOCK(sc);
680
681	bridge_stop(ifp, 1);
682	ifp->if_flags &= ~IFF_UP;
683
684	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
685		bridge_delete_member(sc, bif, 0);
686
687	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
688		bridge_delete_span(sc, bif);
689	}
690
691	BRIDGE_UNLOCK(sc);
692
693	callout_drain(&sc->sc_brcallout);
694
695	mtx_lock(&bridge_list_mtx);
696	LIST_REMOVE(sc, sc_list);
697	mtx_unlock(&bridge_list_mtx);
698
699	bstp_detach(&sc->sc_stp);
700	ether_ifdetach(ifp);
701	if_free(ifp);
702
703	/* Tear down the routing table. */
704	bridge_rtable_fini(sc);
705
706	BRIDGE_LOCK_DESTROY(sc);
707	free(sc, M_DEVBUF);
708}
709
710/*
711 * bridge_ioctl:
712 *
713 *	Handle a control request from the operator.
714 */
715static int
716bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
717{
718	struct bridge_softc *sc = ifp->if_softc;
719	struct ifreq *ifr = (struct ifreq *)data;
720	struct bridge_iflist *bif;
721	struct thread *td = curthread;
722	union {
723		struct ifbreq ifbreq;
724		struct ifbifconf ifbifconf;
725		struct ifbareq ifbareq;
726		struct ifbaconf ifbaconf;
727		struct ifbrparam ifbrparam;
728		struct ifbropreq ifbropreq;
729	} args;
730	struct ifdrv *ifd = (struct ifdrv *) data;
731	const struct bridge_control *bc;
732	int error = 0;
733
734	switch (cmd) {
735
736	case SIOCADDMULTI:
737	case SIOCDELMULTI:
738		break;
739
740	case SIOCGDRVSPEC:
741	case SIOCSDRVSPEC:
742		if (ifd->ifd_cmd >= bridge_control_table_size) {
743			error = EINVAL;
744			break;
745		}
746		bc = &bridge_control_table[ifd->ifd_cmd];
747
748		if (cmd == SIOCGDRVSPEC &&
749		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
750			error = EINVAL;
751			break;
752		}
753		else if (cmd == SIOCSDRVSPEC &&
754		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
755			error = EINVAL;
756			break;
757		}
758
759		if (bc->bc_flags & BC_F_SUSER) {
760			error = priv_check(td, PRIV_NET_BRIDGE);
761			if (error)
762				break;
763		}
764
765		if (ifd->ifd_len != bc->bc_argsize ||
766		    ifd->ifd_len > sizeof(args)) {
767			error = EINVAL;
768			break;
769		}
770
771		bzero(&args, sizeof(args));
772		if (bc->bc_flags & BC_F_COPYIN) {
773			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
774			if (error)
775				break;
776		}
777
778		BRIDGE_LOCK(sc);
779		error = (*bc->bc_func)(sc, &args);
780		BRIDGE_UNLOCK(sc);
781		if (error)
782			break;
783
784		if (bc->bc_flags & BC_F_COPYOUT)
785			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
786
787		break;
788
789	case SIOCSIFFLAGS:
790		if (!(ifp->if_flags & IFF_UP) &&
791		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
792			/*
793			 * If interface is marked down and it is running,
794			 * then stop and disable it.
795			 */
796			BRIDGE_LOCK(sc);
797			bridge_stop(ifp, 1);
798			BRIDGE_UNLOCK(sc);
799		} else if ((ifp->if_flags & IFF_UP) &&
800		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
801			/*
802			 * If interface is marked up and it is stopped, then
803			 * start it.
804			 */
805			(*ifp->if_init)(sc);
806		}
807		break;
808
809	case SIOCSIFMTU:
810		if (ifr->ifr_mtu < 576) {
811			error = EINVAL;
812			break;
813		}
814		if (LIST_EMPTY(&sc->sc_iflist)) {
815			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
816			break;
817		}
818		BRIDGE_LOCK(sc);
819		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
820			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
821				log(LOG_NOTICE, "%s: invalid MTU: %lu(%s)"
822				    " != %d\n", sc->sc_ifp->if_xname,
823				    bif->bif_ifp->if_mtu,
824				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
825				error = EINVAL;
826				break;
827			}
828		}
829		if (!error)
830			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
831		BRIDGE_UNLOCK(sc);
832		break;
833	default:
834		/*
835		 * drop the lock as ether_ioctl() will call bridge_start() and
836		 * cause the lock to be recursed.
837		 */
838		error = ether_ioctl(ifp, cmd, data);
839		break;
840	}
841
842	return (error);
843}
844
845/*
846 * bridge_mutecaps:
847 *
848 *	Clear or restore unwanted capabilities on the member interface
849 */
850static void
851bridge_mutecaps(struct bridge_softc *sc)
852{
853	struct bridge_iflist *bif;
854	int enabled, mask;
855
856	/* Initial bitmask of capabilities to test */
857	mask = BRIDGE_IFCAPS_MASK;
858
859	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
860		/* Every member must support it or its disabled */
861		mask &= bif->bif_savedcaps;
862	}
863
864	BRIDGE_XLOCK(sc);
865	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
866		enabled = bif->bif_ifp->if_capenable;
867		enabled &= ~BRIDGE_IFCAPS_STRIP;
868		/* strip off mask bits and enable them again if allowed */
869		enabled &= ~BRIDGE_IFCAPS_MASK;
870		enabled |= mask;
871		BRIDGE_UNLOCK(sc);
872		bridge_set_ifcap(sc, bif, enabled);
873		BRIDGE_LOCK(sc);
874	}
875	BRIDGE_XDROP(sc);
876
877}
878
879static void
880bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
881{
882	struct ifnet *ifp = bif->bif_ifp;
883	struct ifreq ifr;
884	int error;
885
886	BRIDGE_UNLOCK_ASSERT(sc);
887
888	bzero(&ifr, sizeof(ifr));
889	ifr.ifr_reqcap = set;
890
891	if (ifp->if_capenable != set) {
892		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
893		if (error)
894			if_printf(sc->sc_ifp,
895			    "error setting interface capabilities on %s\n",
896			    ifp->if_xname);
897	}
898}
899
900/*
901 * bridge_lookup_member:
902 *
903 *	Lookup a bridge member interface.
904 */
905static struct bridge_iflist *
906bridge_lookup_member(struct bridge_softc *sc, const char *name)
907{
908	struct bridge_iflist *bif;
909	struct ifnet *ifp;
910
911	BRIDGE_LOCK_ASSERT(sc);
912
913	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
914		ifp = bif->bif_ifp;
915		if (strcmp(ifp->if_xname, name) == 0)
916			return (bif);
917	}
918
919	return (NULL);
920}
921
922/*
923 * bridge_lookup_member_if:
924 *
925 *	Lookup a bridge member interface by ifnet*.
926 */
927static struct bridge_iflist *
928bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
929{
930	struct bridge_iflist *bif;
931
932	BRIDGE_LOCK_ASSERT(sc);
933
934	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
935		if (bif->bif_ifp == member_ifp)
936			return (bif);
937	}
938
939	return (NULL);
940}
941
942/*
943 * bridge_delete_member:
944 *
945 *	Delete the specified member interface.
946 */
947static void
948bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
949    int gone)
950{
951	struct ifnet *ifs = bif->bif_ifp;
952	struct ifnet *fif = NULL;
953
954	BRIDGE_LOCK_ASSERT(sc);
955
956	if (bif->bif_flags & IFBIF_STP)
957		bstp_disable(&bif->bif_stp);
958
959	ifs->if_bridge = NULL;
960	BRIDGE_XLOCK(sc);
961	LIST_REMOVE(bif, bif_next);
962	BRIDGE_XDROP(sc);
963
964	/*
965	 * If removing the interface that gave the bridge its mac address, set
966	 * the mac address of the bridge to the address of the next member, or
967	 * to its default address if no members are left.
968	 */
969	if (bridge_inherit_mac && sc->sc_ifaddr == ifs) {
970		if (LIST_EMPTY(&sc->sc_iflist)) {
971			bcopy(sc->sc_defaddr,
972			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
973			sc->sc_ifaddr = NULL;
974		} else {
975			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
976			bcopy(IF_LLADDR(fif),
977			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
978			sc->sc_ifaddr = fif;
979		}
980		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
981	}
982
983	bridge_linkcheck(sc);
984	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
985	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
986	KASSERT(bif->bif_addrcnt == 0,
987	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
988
989	BRIDGE_UNLOCK(sc);
990	if (!gone) {
991		switch (ifs->if_type) {
992		case IFT_ETHER:
993		case IFT_L2VLAN:
994			/*
995			 * Take the interface out of promiscuous mode, but only
996			 * if it was promiscuous in the first place. It might
997			 * not be if we're in the bridge_ioctl_add() error path.
998			 */
999			if (ifs->if_flags & IFF_PROMISC)
1000				(void) ifpromisc(ifs, 0);
1001			break;
1002
1003		case IFT_GIF:
1004			break;
1005
1006		default:
1007#ifdef DIAGNOSTIC
1008			panic("bridge_delete_member: impossible");
1009#endif
1010			break;
1011		}
1012		/* reneable any interface capabilities */
1013		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1014	}
1015	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1016	BRIDGE_LOCK(sc);
1017	free(bif, M_DEVBUF);
1018}
1019
1020/*
1021 * bridge_delete_span:
1022 *
1023 *	Delete the specified span interface.
1024 */
1025static void
1026bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1027{
1028	BRIDGE_LOCK_ASSERT(sc);
1029
1030	KASSERT(bif->bif_ifp->if_bridge == NULL,
1031	    ("%s: not a span interface", __func__));
1032
1033	LIST_REMOVE(bif, bif_next);
1034	free(bif, M_DEVBUF);
1035}
1036
1037static int
1038bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1039{
1040	struct ifbreq *req = arg;
1041	struct bridge_iflist *bif = NULL;
1042	struct ifnet *ifs;
1043	int error = 0;
1044
1045	ifs = ifunit(req->ifbr_ifsname);
1046	if (ifs == NULL)
1047		return (ENOENT);
1048	if (ifs->if_ioctl == NULL)	/* must be supported */
1049		return (EINVAL);
1050
1051	/* If it's in the span list, it can't be a member. */
1052	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1053		if (ifs == bif->bif_ifp)
1054			return (EBUSY);
1055
1056	if (ifs->if_bridge == sc)
1057		return (EEXIST);
1058
1059	if (ifs->if_bridge != NULL)
1060		return (EBUSY);
1061
1062	switch (ifs->if_type) {
1063	case IFT_ETHER:
1064	case IFT_L2VLAN:
1065	case IFT_GIF:
1066		/* permitted interface types */
1067		break;
1068	default:
1069		return (EINVAL);
1070	}
1071
1072#ifdef INET6
1073	/*
1074	 * Two valid inet6 addresses with link-local scope must not be
1075	 * on the parent interface and the member interfaces at the
1076	 * same time.  This restriction is needed to prevent violation
1077	 * of link-local scope zone.  Attempts to add a member
1078	 * interface which has inet6 addresses when the parent has
1079	 * inet6 triggers removal of all inet6 addresses on the member
1080	 * interface.
1081	 */
1082
1083	/* Check if the parent interface has a link-local scope addr. */
1084	if (V_allow_llz_overlap == 0 &&
1085	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1086		/*
1087		 * If any, remove all inet6 addresses from the member
1088		 * interfaces.
1089		 */
1090		BRIDGE_XLOCK(sc);
1091		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1092 			if (in6ifa_llaonifp(bif->bif_ifp)) {
1093				BRIDGE_UNLOCK(sc);
1094				in6_ifdetach(bif->bif_ifp);
1095				BRIDGE_LOCK(sc);
1096				if_printf(sc->sc_ifp,
1097				    "IPv6 addresses on %s have been removed "
1098				    "before adding it as a member to prevent "
1099				    "IPv6 address scope violation.\n",
1100				    bif->bif_ifp->if_xname);
1101			}
1102		}
1103		BRIDGE_XDROP(sc);
1104		if (in6ifa_llaonifp(ifs)) {
1105			BRIDGE_UNLOCK(sc);
1106			in6_ifdetach(ifs);
1107			BRIDGE_LOCK(sc);
1108			if_printf(sc->sc_ifp,
1109			    "IPv6 addresses on %s have been removed "
1110			    "before adding it as a member to prevent "
1111			    "IPv6 address scope violation.\n",
1112			    ifs->if_xname);
1113		}
1114	}
1115#endif
1116	/* Allow the first Ethernet member to define the MTU */
1117	if (LIST_EMPTY(&sc->sc_iflist))
1118		sc->sc_ifp->if_mtu = ifs->if_mtu;
1119	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1120		if_printf(sc->sc_ifp, "invalid MTU: %lu(%s) != %lu\n",
1121		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1122		return (EINVAL);
1123	}
1124
1125	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1126	if (bif == NULL)
1127		return (ENOMEM);
1128
1129	bif->bif_ifp = ifs;
1130	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1131	bif->bif_savedcaps = ifs->if_capenable;
1132
1133	/*
1134	 * Assign the interface's MAC address to the bridge if it's the first
1135	 * member and the MAC address of the bridge has not been changed from
1136	 * the default randomly generated one.
1137	 */
1138	if (bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1139	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1140		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1141		sc->sc_ifaddr = ifs;
1142		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1143	}
1144
1145	ifs->if_bridge = sc;
1146	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1147	/*
1148	 * XXX: XLOCK HERE!?!
1149	 *
1150	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1151	 */
1152	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1153
1154	/* Set interface capabilities to the intersection set of all members */
1155	bridge_mutecaps(sc);
1156	bridge_linkcheck(sc);
1157
1158	/* Place the interface into promiscuous mode */
1159	switch (ifs->if_type) {
1160		case IFT_ETHER:
1161		case IFT_L2VLAN:
1162			BRIDGE_UNLOCK(sc);
1163			error = ifpromisc(ifs, 1);
1164			BRIDGE_LOCK(sc);
1165			break;
1166	}
1167
1168	if (error)
1169		bridge_delete_member(sc, bif, 0);
1170	return (error);
1171}
1172
1173static int
1174bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1175{
1176	struct ifbreq *req = arg;
1177	struct bridge_iflist *bif;
1178
1179	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1180	if (bif == NULL)
1181		return (ENOENT);
1182
1183	bridge_delete_member(sc, bif, 0);
1184
1185	return (0);
1186}
1187
1188static int
1189bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1190{
1191	struct ifbreq *req = arg;
1192	struct bridge_iflist *bif;
1193	struct bstp_port *bp;
1194
1195	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1196	if (bif == NULL)
1197		return (ENOENT);
1198
1199	bp = &bif->bif_stp;
1200	req->ifbr_ifsflags = bif->bif_flags;
1201	req->ifbr_state = bp->bp_state;
1202	req->ifbr_priority = bp->bp_priority;
1203	req->ifbr_path_cost = bp->bp_path_cost;
1204	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1205	req->ifbr_proto = bp->bp_protover;
1206	req->ifbr_role = bp->bp_role;
1207	req->ifbr_stpflags = bp->bp_flags;
1208	req->ifbr_addrcnt = bif->bif_addrcnt;
1209	req->ifbr_addrmax = bif->bif_addrmax;
1210	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1211
1212	/* Copy STP state options as flags */
1213	if (bp->bp_operedge)
1214		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1215	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1216		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1217	if (bp->bp_ptp_link)
1218		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1219	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1220		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1221	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1222		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1223	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1224		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1225	return (0);
1226}
1227
1228static int
1229bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1230{
1231	struct ifbreq *req = arg;
1232	struct bridge_iflist *bif;
1233	struct bstp_port *bp;
1234	int error;
1235
1236	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1237	if (bif == NULL)
1238		return (ENOENT);
1239	bp = &bif->bif_stp;
1240
1241	if (req->ifbr_ifsflags & IFBIF_SPAN)
1242		/* SPAN is readonly */
1243		return (EINVAL);
1244
1245	if (req->ifbr_ifsflags & IFBIF_STP) {
1246		if ((bif->bif_flags & IFBIF_STP) == 0) {
1247			error = bstp_enable(&bif->bif_stp);
1248			if (error)
1249				return (error);
1250		}
1251	} else {
1252		if ((bif->bif_flags & IFBIF_STP) != 0)
1253			bstp_disable(&bif->bif_stp);
1254	}
1255
1256	/* Pass on STP flags */
1257	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1258	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1259	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1260	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1261
1262	/* Save the bits relating to the bridge */
1263	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1264
1265	return (0);
1266}
1267
1268static int
1269bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1270{
1271	struct ifbrparam *param = arg;
1272
1273	sc->sc_brtmax = param->ifbrp_csize;
1274	bridge_rttrim(sc);
1275
1276	return (0);
1277}
1278
1279static int
1280bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1281{
1282	struct ifbrparam *param = arg;
1283
1284	param->ifbrp_csize = sc->sc_brtmax;
1285
1286	return (0);
1287}
1288
1289static int
1290bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1291{
1292	struct ifbifconf *bifc = arg;
1293	struct bridge_iflist *bif;
1294	struct ifbreq breq;
1295	char *buf, *outbuf;
1296	int count, buflen, len, error = 0;
1297
1298	count = 0;
1299	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1300		count++;
1301	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1302		count++;
1303
1304	buflen = sizeof(breq) * count;
1305	if (bifc->ifbic_len == 0) {
1306		bifc->ifbic_len = buflen;
1307		return (0);
1308	}
1309	BRIDGE_UNLOCK(sc);
1310	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1311	BRIDGE_LOCK(sc);
1312
1313	count = 0;
1314	buf = outbuf;
1315	len = min(bifc->ifbic_len, buflen);
1316	bzero(&breq, sizeof(breq));
1317	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1318		if (len < sizeof(breq))
1319			break;
1320
1321		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1322		    sizeof(breq.ifbr_ifsname));
1323		/* Fill in the ifbreq structure */
1324		error = bridge_ioctl_gifflags(sc, &breq);
1325		if (error)
1326			break;
1327		memcpy(buf, &breq, sizeof(breq));
1328		count++;
1329		buf += sizeof(breq);
1330		len -= sizeof(breq);
1331	}
1332	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1333		if (len < sizeof(breq))
1334			break;
1335
1336		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1337		    sizeof(breq.ifbr_ifsname));
1338		breq.ifbr_ifsflags = bif->bif_flags;
1339		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1340		memcpy(buf, &breq, sizeof(breq));
1341		count++;
1342		buf += sizeof(breq);
1343		len -= sizeof(breq);
1344	}
1345
1346	BRIDGE_UNLOCK(sc);
1347	bifc->ifbic_len = sizeof(breq) * count;
1348	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1349	BRIDGE_LOCK(sc);
1350	free(outbuf, M_TEMP);
1351	return (error);
1352}
1353
1354static int
1355bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1356{
1357	struct ifbaconf *bac = arg;
1358	struct bridge_rtnode *brt;
1359	struct ifbareq bareq;
1360	char *buf, *outbuf;
1361	int count, buflen, len, error = 0;
1362
1363	if (bac->ifbac_len == 0)
1364		return (0);
1365
1366	count = 0;
1367	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1368		count++;
1369	buflen = sizeof(bareq) * count;
1370
1371	BRIDGE_UNLOCK(sc);
1372	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1373	BRIDGE_LOCK(sc);
1374
1375	count = 0;
1376	buf = outbuf;
1377	len = min(bac->ifbac_len, buflen);
1378	bzero(&bareq, sizeof(bareq));
1379	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1380		if (len < sizeof(bareq))
1381			goto out;
1382		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1383		    sizeof(bareq.ifba_ifsname));
1384		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1385		bareq.ifba_vlan = brt->brt_vlan;
1386		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1387				time_uptime < brt->brt_expire)
1388			bareq.ifba_expire = brt->brt_expire - time_uptime;
1389		else
1390			bareq.ifba_expire = 0;
1391		bareq.ifba_flags = brt->brt_flags;
1392
1393		memcpy(buf, &bareq, sizeof(bareq));
1394		count++;
1395		buf += sizeof(bareq);
1396		len -= sizeof(bareq);
1397	}
1398out:
1399	BRIDGE_UNLOCK(sc);
1400	bac->ifbac_len = sizeof(bareq) * count;
1401	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1402	BRIDGE_LOCK(sc);
1403	free(outbuf, M_TEMP);
1404	return (error);
1405}
1406
1407static int
1408bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1409{
1410	struct ifbareq *req = arg;
1411	struct bridge_iflist *bif;
1412	int error;
1413
1414	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1415	if (bif == NULL)
1416		return (ENOENT);
1417
1418	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1419	    req->ifba_flags);
1420
1421	return (error);
1422}
1423
1424static int
1425bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1426{
1427	struct ifbrparam *param = arg;
1428
1429	sc->sc_brttimeout = param->ifbrp_ctime;
1430	return (0);
1431}
1432
1433static int
1434bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1435{
1436	struct ifbrparam *param = arg;
1437
1438	param->ifbrp_ctime = sc->sc_brttimeout;
1439	return (0);
1440}
1441
1442static int
1443bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1444{
1445	struct ifbareq *req = arg;
1446
1447	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1448}
1449
1450static int
1451bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1452{
1453	struct ifbreq *req = arg;
1454
1455	bridge_rtflush(sc, req->ifbr_ifsflags);
1456	return (0);
1457}
1458
1459static int
1460bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1461{
1462	struct ifbrparam *param = arg;
1463	struct bstp_state *bs = &sc->sc_stp;
1464
1465	param->ifbrp_prio = bs->bs_bridge_priority;
1466	return (0);
1467}
1468
1469static int
1470bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1471{
1472	struct ifbrparam *param = arg;
1473
1474	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1475}
1476
1477static int
1478bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1479{
1480	struct ifbrparam *param = arg;
1481	struct bstp_state *bs = &sc->sc_stp;
1482
1483	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1484	return (0);
1485}
1486
1487static int
1488bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1489{
1490	struct ifbrparam *param = arg;
1491
1492	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1493}
1494
1495static int
1496bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1497{
1498	struct ifbrparam *param = arg;
1499	struct bstp_state *bs = &sc->sc_stp;
1500
1501	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1502	return (0);
1503}
1504
1505static int
1506bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1507{
1508	struct ifbrparam *param = arg;
1509
1510	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1511}
1512
1513static int
1514bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1515{
1516	struct ifbrparam *param = arg;
1517	struct bstp_state *bs = &sc->sc_stp;
1518
1519	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1520	return (0);
1521}
1522
1523static int
1524bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1525{
1526	struct ifbrparam *param = arg;
1527
1528	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1529}
1530
1531static int
1532bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1533{
1534	struct ifbreq *req = arg;
1535	struct bridge_iflist *bif;
1536
1537	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1538	if (bif == NULL)
1539		return (ENOENT);
1540
1541	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1542}
1543
1544static int
1545bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1546{
1547	struct ifbreq *req = arg;
1548	struct bridge_iflist *bif;
1549
1550	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1551	if (bif == NULL)
1552		return (ENOENT);
1553
1554	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1555}
1556
1557static int
1558bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1559{
1560	struct ifbreq *req = arg;
1561	struct bridge_iflist *bif;
1562
1563	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1564	if (bif == NULL)
1565		return (ENOENT);
1566
1567	bif->bif_addrmax = req->ifbr_addrmax;
1568	return (0);
1569}
1570
1571static int
1572bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1573{
1574	struct ifbreq *req = arg;
1575	struct bridge_iflist *bif = NULL;
1576	struct ifnet *ifs;
1577
1578	ifs = ifunit(req->ifbr_ifsname);
1579	if (ifs == NULL)
1580		return (ENOENT);
1581
1582	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1583		if (ifs == bif->bif_ifp)
1584			return (EBUSY);
1585
1586	if (ifs->if_bridge != NULL)
1587		return (EBUSY);
1588
1589	switch (ifs->if_type) {
1590		case IFT_ETHER:
1591		case IFT_GIF:
1592		case IFT_L2VLAN:
1593			break;
1594		default:
1595			return (EINVAL);
1596	}
1597
1598	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1599	if (bif == NULL)
1600		return (ENOMEM);
1601
1602	bif->bif_ifp = ifs;
1603	bif->bif_flags = IFBIF_SPAN;
1604
1605	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1606
1607	return (0);
1608}
1609
1610static int
1611bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1612{
1613	struct ifbreq *req = arg;
1614	struct bridge_iflist *bif;
1615	struct ifnet *ifs;
1616
1617	ifs = ifunit(req->ifbr_ifsname);
1618	if (ifs == NULL)
1619		return (ENOENT);
1620
1621	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1622		if (ifs == bif->bif_ifp)
1623			break;
1624
1625	if (bif == NULL)
1626		return (ENOENT);
1627
1628	bridge_delete_span(sc, bif);
1629
1630	return (0);
1631}
1632
1633static int
1634bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1635{
1636	struct ifbropreq *req = arg;
1637	struct bstp_state *bs = &sc->sc_stp;
1638	struct bstp_port *root_port;
1639
1640	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1641	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1642	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1643
1644	root_port = bs->bs_root_port;
1645	if (root_port == NULL)
1646		req->ifbop_root_port = 0;
1647	else
1648		req->ifbop_root_port = root_port->bp_ifp->if_index;
1649
1650	req->ifbop_holdcount = bs->bs_txholdcount;
1651	req->ifbop_priority = bs->bs_bridge_priority;
1652	req->ifbop_protocol = bs->bs_protover;
1653	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1654	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1655	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1656	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1657	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1658	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1659
1660	return (0);
1661}
1662
1663static int
1664bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1665{
1666	struct ifbrparam *param = arg;
1667
1668	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1669	return (0);
1670}
1671
1672static int
1673bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1674{
1675	struct ifbpstpconf *bifstp = arg;
1676	struct bridge_iflist *bif;
1677	struct bstp_port *bp;
1678	struct ifbpstpreq bpreq;
1679	char *buf, *outbuf;
1680	int count, buflen, len, error = 0;
1681
1682	count = 0;
1683	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1684		if ((bif->bif_flags & IFBIF_STP) != 0)
1685			count++;
1686	}
1687
1688	buflen = sizeof(bpreq) * count;
1689	if (bifstp->ifbpstp_len == 0) {
1690		bifstp->ifbpstp_len = buflen;
1691		return (0);
1692	}
1693
1694	BRIDGE_UNLOCK(sc);
1695	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1696	BRIDGE_LOCK(sc);
1697
1698	count = 0;
1699	buf = outbuf;
1700	len = min(bifstp->ifbpstp_len, buflen);
1701	bzero(&bpreq, sizeof(bpreq));
1702	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1703		if (len < sizeof(bpreq))
1704			break;
1705
1706		if ((bif->bif_flags & IFBIF_STP) == 0)
1707			continue;
1708
1709		bp = &bif->bif_stp;
1710		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1711		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1712		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1713		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1714		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1715		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1716
1717		memcpy(buf, &bpreq, sizeof(bpreq));
1718		count++;
1719		buf += sizeof(bpreq);
1720		len -= sizeof(bpreq);
1721	}
1722
1723	BRIDGE_UNLOCK(sc);
1724	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1725	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1726	BRIDGE_LOCK(sc);
1727	free(outbuf, M_TEMP);
1728	return (error);
1729}
1730
1731static int
1732bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1733{
1734	struct ifbrparam *param = arg;
1735
1736	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1737}
1738
1739static int
1740bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1741{
1742	struct ifbrparam *param = arg;
1743
1744	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1745}
1746
1747/*
1748 * bridge_ifdetach:
1749 *
1750 *	Detach an interface from a bridge.  Called when a member
1751 *	interface is detaching.
1752 */
1753static void
1754bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1755{
1756	struct bridge_softc *sc = ifp->if_bridge;
1757	struct bridge_iflist *bif;
1758
1759	if (ifp->if_flags & IFF_RENAMING)
1760		return;
1761
1762	/* Check if the interface is a bridge member */
1763	if (sc != NULL) {
1764		BRIDGE_LOCK(sc);
1765
1766		bif = bridge_lookup_member_if(sc, ifp);
1767		if (bif != NULL)
1768			bridge_delete_member(sc, bif, 1);
1769
1770		BRIDGE_UNLOCK(sc);
1771		return;
1772	}
1773
1774	/* Check if the interface is a span port */
1775	mtx_lock(&bridge_list_mtx);
1776	LIST_FOREACH(sc, &bridge_list, sc_list) {
1777		BRIDGE_LOCK(sc);
1778		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1779			if (ifp == bif->bif_ifp) {
1780				bridge_delete_span(sc, bif);
1781				break;
1782			}
1783
1784		BRIDGE_UNLOCK(sc);
1785	}
1786	mtx_unlock(&bridge_list_mtx);
1787}
1788
1789/*
1790 * bridge_init:
1791 *
1792 *	Initialize a bridge interface.
1793 */
1794static void
1795bridge_init(void *xsc)
1796{
1797	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1798	struct ifnet *ifp = sc->sc_ifp;
1799
1800	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1801		return;
1802
1803	BRIDGE_LOCK(sc);
1804	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1805	    bridge_timer, sc);
1806
1807	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1808	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1809
1810	BRIDGE_UNLOCK(sc);
1811}
1812
1813/*
1814 * bridge_stop:
1815 *
1816 *	Stop the bridge interface.
1817 */
1818static void
1819bridge_stop(struct ifnet *ifp, int disable)
1820{
1821	struct bridge_softc *sc = ifp->if_softc;
1822
1823	BRIDGE_LOCK_ASSERT(sc);
1824
1825	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1826		return;
1827
1828	callout_stop(&sc->sc_brcallout);
1829	bstp_stop(&sc->sc_stp);
1830
1831	bridge_rtflush(sc, IFBF_FLUSHDYN);
1832
1833	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1834}
1835
1836/*
1837 * bridge_enqueue:
1838 *
1839 *	Enqueue a packet on a bridge member interface.
1840 *
1841 */
1842static int
1843bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1844{
1845	int len, err = 0;
1846	short mflags;
1847	struct mbuf *m0;
1848
1849	/* We may be sending a fragment so traverse the mbuf */
1850	for (; m; m = m0) {
1851		m0 = m->m_nextpkt;
1852		m->m_nextpkt = NULL;
1853		len = m->m_pkthdr.len;
1854		mflags = m->m_flags;
1855
1856		/*
1857		 * If underlying interface can not do VLAN tag insertion itself
1858		 * then attach a packet tag that holds it.
1859		 */
1860		if ((m->m_flags & M_VLANTAG) &&
1861		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1862			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1863			if (m == NULL) {
1864				if_printf(dst_ifp,
1865				    "unable to prepend VLAN header\n");
1866				dst_ifp->if_oerrors++;
1867				continue;
1868			}
1869			m->m_flags &= ~M_VLANTAG;
1870		}
1871
1872		M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
1873		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1874			m_freem(m0);
1875			sc->sc_ifp->if_oerrors++;
1876			break;
1877		}
1878
1879		sc->sc_ifp->if_opackets++;
1880		sc->sc_ifp->if_obytes += len;
1881		if (mflags & M_MCAST)
1882			sc->sc_ifp->if_omcasts++;
1883	}
1884
1885	return (err);
1886}
1887
1888/*
1889 * bridge_dummynet:
1890 *
1891 * 	Receive a queued packet from dummynet and pass it on to the output
1892 * 	interface.
1893 *
1894 *	The mbuf has the Ethernet header already attached.
1895 */
1896static void
1897bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1898{
1899	struct bridge_softc *sc;
1900
1901	sc = ifp->if_bridge;
1902
1903	/*
1904	 * The packet didnt originate from a member interface. This should only
1905	 * ever happen if a member interface is removed while packets are
1906	 * queued for it.
1907	 */
1908	if (sc == NULL) {
1909		m_freem(m);
1910		return;
1911	}
1912
1913	if (PFIL_HOOKED(&V_inet_pfil_hook)
1914#ifdef INET6
1915	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1916#endif
1917	    ) {
1918		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1919			return;
1920		if (m == NULL)
1921			return;
1922	}
1923
1924	bridge_enqueue(sc, ifp, m);
1925}
1926
1927/*
1928 * bridge_output:
1929 *
1930 *	Send output from a bridge member interface.  This
1931 *	performs the bridging function for locally originated
1932 *	packets.
1933 *
1934 *	The mbuf has the Ethernet header already attached.  We must
1935 *	enqueue or free the mbuf before returning.
1936 */
1937static int
1938bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1939    struct rtentry *rt)
1940{
1941	struct ether_header *eh;
1942	struct ifnet *dst_if;
1943	struct bridge_softc *sc;
1944	uint16_t vlan;
1945
1946	if (m->m_len < ETHER_HDR_LEN) {
1947		m = m_pullup(m, ETHER_HDR_LEN);
1948		if (m == NULL)
1949			return (0);
1950	}
1951
1952	eh = mtod(m, struct ether_header *);
1953	sc = ifp->if_bridge;
1954	vlan = VLANTAGOF(m);
1955
1956	BRIDGE_LOCK(sc);
1957
1958	/*
1959	 * If bridge is down, but the original output interface is up,
1960	 * go ahead and send out that interface.  Otherwise, the packet
1961	 * is dropped below.
1962	 */
1963	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1964		dst_if = ifp;
1965		goto sendunicast;
1966	}
1967
1968	/*
1969	 * If the packet is a multicast, or we don't know a better way to
1970	 * get there, send to all interfaces.
1971	 */
1972	if (ETHER_IS_MULTICAST(eh->ether_dhost))
1973		dst_if = NULL;
1974	else
1975		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1976	if (dst_if == NULL) {
1977		struct bridge_iflist *bif;
1978		struct mbuf *mc;
1979		int error = 0, used = 0;
1980
1981		bridge_span(sc, m);
1982
1983		BRIDGE_LOCK2REF(sc, error);
1984		if (error) {
1985			m_freem(m);
1986			return (0);
1987		}
1988
1989		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1990			dst_if = bif->bif_ifp;
1991
1992			if (dst_if->if_type == IFT_GIF)
1993				continue;
1994			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1995				continue;
1996
1997			/*
1998			 * If this is not the original output interface,
1999			 * and the interface is participating in spanning
2000			 * tree, make sure the port is in a state that
2001			 * allows forwarding.
2002			 */
2003			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2004			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2005				continue;
2006
2007			if (LIST_NEXT(bif, bif_next) == NULL) {
2008				used = 1;
2009				mc = m;
2010			} else {
2011				mc = m_copypacket(m, M_NOWAIT);
2012				if (mc == NULL) {
2013					sc->sc_ifp->if_oerrors++;
2014					continue;
2015				}
2016			}
2017
2018			bridge_enqueue(sc, dst_if, mc);
2019		}
2020		if (used == 0)
2021			m_freem(m);
2022		BRIDGE_UNREF(sc);
2023		return (0);
2024	}
2025
2026sendunicast:
2027	/*
2028	 * XXX Spanning tree consideration here?
2029	 */
2030
2031	bridge_span(sc, m);
2032	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2033		m_freem(m);
2034		BRIDGE_UNLOCK(sc);
2035		return (0);
2036	}
2037
2038	BRIDGE_UNLOCK(sc);
2039	bridge_enqueue(sc, dst_if, m);
2040	return (0);
2041}
2042
2043/*
2044 * bridge_transmit:
2045 *
2046 *	Do output on a bridge.
2047 *
2048 */
2049static int
2050bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2051{
2052	struct bridge_softc *sc;
2053	struct ether_header *eh;
2054	struct ifnet *dst_if;
2055	int error = 0;
2056
2057	sc = ifp->if_softc;
2058
2059	ETHER_BPF_MTAP(ifp, m);
2060
2061	eh = mtod(m, struct ether_header *);
2062
2063	BRIDGE_LOCK(sc);
2064	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2065	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2066		BRIDGE_UNLOCK(sc);
2067		error = bridge_enqueue(sc, dst_if, m);
2068	} else
2069		bridge_broadcast(sc, ifp, m, 0);
2070
2071	return (error);
2072}
2073
2074/*
2075 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2076 */
2077static void
2078bridge_qflush(struct ifnet *ifp __unused)
2079{
2080}
2081
2082/*
2083 * bridge_forward:
2084 *
2085 *	The forwarding function of the bridge.
2086 *
2087 *	NOTE: Releases the lock on return.
2088 */
2089static void
2090bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2091    struct mbuf *m)
2092{
2093	struct bridge_iflist *dbif;
2094	struct ifnet *src_if, *dst_if, *ifp;
2095	struct ether_header *eh;
2096	uint16_t vlan;
2097	uint8_t *dst;
2098	int error;
2099
2100	src_if = m->m_pkthdr.rcvif;
2101	ifp = sc->sc_ifp;
2102
2103	ifp->if_ipackets++;
2104	ifp->if_ibytes += m->m_pkthdr.len;
2105	vlan = VLANTAGOF(m);
2106
2107	if ((sbif->bif_flags & IFBIF_STP) &&
2108	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2109		goto drop;
2110
2111	eh = mtod(m, struct ether_header *);
2112	dst = eh->ether_dhost;
2113
2114	/* If the interface is learning, record the address. */
2115	if (sbif->bif_flags & IFBIF_LEARNING) {
2116		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2117		    sbif, 0, IFBAF_DYNAMIC);
2118		/*
2119		 * If the interface has addresses limits then deny any source
2120		 * that is not in the cache.
2121		 */
2122		if (error && sbif->bif_addrmax)
2123			goto drop;
2124	}
2125
2126	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2127	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2128		goto drop;
2129
2130	/*
2131	 * At this point, the port either doesn't participate
2132	 * in spanning tree or it is in the forwarding state.
2133	 */
2134
2135	/*
2136	 * If the packet is unicast, destined for someone on
2137	 * "this" side of the bridge, drop it.
2138	 */
2139	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2140		dst_if = bridge_rtlookup(sc, dst, vlan);
2141		if (src_if == dst_if)
2142			goto drop;
2143	} else {
2144		/*
2145		 * Check if its a reserved multicast address, any address
2146		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2147		 * bridge.
2148		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2149		 */
2150		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2151		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2152		    dst[4] == 0x00 && dst[5] <= 0x0f)
2153			goto drop;
2154
2155		/* ...forward it to all interfaces. */
2156		ifp->if_imcasts++;
2157		dst_if = NULL;
2158	}
2159
2160	/*
2161	 * If we have a destination interface which is a member of our bridge,
2162	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2163	 * For broadcast or multicast packets, don't bother because it will
2164	 * be reinjected into ether_input. We do this before we pass the packets
2165	 * through the pfil(9) framework, as it is possible that pfil(9) will
2166	 * drop the packet, or possibly modify it, making it difficult to debug
2167	 * firewall issues on the bridge.
2168	 */
2169	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2170		ETHER_BPF_MTAP(ifp, m);
2171
2172	/* run the packet filter */
2173	if (PFIL_HOOKED(&V_inet_pfil_hook)
2174#ifdef INET6
2175	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2176#endif
2177	    ) {
2178		BRIDGE_UNLOCK(sc);
2179		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2180			return;
2181		if (m == NULL)
2182			return;
2183		BRIDGE_LOCK(sc);
2184	}
2185
2186	if (dst_if == NULL) {
2187		bridge_broadcast(sc, src_if, m, 1);
2188		return;
2189	}
2190
2191	/*
2192	 * At this point, we're dealing with a unicast frame
2193	 * going to a different interface.
2194	 */
2195	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2196		goto drop;
2197
2198	dbif = bridge_lookup_member_if(sc, dst_if);
2199	if (dbif == NULL)
2200		/* Not a member of the bridge (anymore?) */
2201		goto drop;
2202
2203	/* Private segments can not talk to each other */
2204	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2205		goto drop;
2206
2207	if ((dbif->bif_flags & IFBIF_STP) &&
2208	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2209		goto drop;
2210
2211	BRIDGE_UNLOCK(sc);
2212
2213	if (PFIL_HOOKED(&V_inet_pfil_hook)
2214#ifdef INET6
2215	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2216#endif
2217	    ) {
2218		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2219			return;
2220		if (m == NULL)
2221			return;
2222	}
2223
2224	bridge_enqueue(sc, dst_if, m);
2225	return;
2226
2227drop:
2228	BRIDGE_UNLOCK(sc);
2229	m_freem(m);
2230}
2231
2232/*
2233 * bridge_input:
2234 *
2235 *	Receive input from a member interface.  Queue the packet for
2236 *	bridging if it is not for us.
2237 */
2238static struct mbuf *
2239bridge_input(struct ifnet *ifp, struct mbuf *m)
2240{
2241	struct bridge_softc *sc = ifp->if_bridge;
2242	struct bridge_iflist *bif, *bif2;
2243	struct ifnet *bifp;
2244	struct ether_header *eh;
2245	struct mbuf *mc, *mc2;
2246	uint16_t vlan;
2247	int error;
2248
2249	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2250		return (m);
2251
2252	bifp = sc->sc_ifp;
2253	vlan = VLANTAGOF(m);
2254
2255	/*
2256	 * Implement support for bridge monitoring. If this flag has been
2257	 * set on this interface, discard the packet once we push it through
2258	 * the bpf(4) machinery, but before we do, increment the byte and
2259	 * packet counters associated with this interface.
2260	 */
2261	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2262		m->m_pkthdr.rcvif  = bifp;
2263		ETHER_BPF_MTAP(bifp, m);
2264		bifp->if_ipackets++;
2265		bifp->if_ibytes += m->m_pkthdr.len;
2266		m_freem(m);
2267		return (NULL);
2268	}
2269	BRIDGE_LOCK(sc);
2270	bif = bridge_lookup_member_if(sc, ifp);
2271	if (bif == NULL) {
2272		BRIDGE_UNLOCK(sc);
2273		return (m);
2274	}
2275
2276	eh = mtod(m, struct ether_header *);
2277
2278	bridge_span(sc, m);
2279
2280	if (m->m_flags & (M_BCAST|M_MCAST)) {
2281		/* Tap off 802.1D packets; they do not get forwarded. */
2282		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2283		    ETHER_ADDR_LEN) == 0) {
2284			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2285			BRIDGE_UNLOCK(sc);
2286			return (NULL);
2287		}
2288
2289		if ((bif->bif_flags & IFBIF_STP) &&
2290		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2291			BRIDGE_UNLOCK(sc);
2292			return (m);
2293		}
2294
2295		/*
2296		 * Make a deep copy of the packet and enqueue the copy
2297		 * for bridge processing; return the original packet for
2298		 * local processing.
2299		 */
2300		mc = m_dup(m, M_NOWAIT);
2301		if (mc == NULL) {
2302			BRIDGE_UNLOCK(sc);
2303			return (m);
2304		}
2305
2306		/* Perform the bridge forwarding function with the copy. */
2307		bridge_forward(sc, bif, mc);
2308
2309		/*
2310		 * Reinject the mbuf as arriving on the bridge so we have a
2311		 * chance at claiming multicast packets. We can not loop back
2312		 * here from ether_input as a bridge is never a member of a
2313		 * bridge.
2314		 */
2315		KASSERT(bifp->if_bridge == NULL,
2316		    ("loop created in bridge_input"));
2317		mc2 = m_dup(m, M_NOWAIT);
2318		if (mc2 != NULL) {
2319			/* Keep the layer3 header aligned */
2320			int i = min(mc2->m_pkthdr.len, max_protohdr);
2321			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2322		}
2323		if (mc2 != NULL) {
2324			mc2->m_pkthdr.rcvif = bifp;
2325			(*bifp->if_input)(bifp, mc2);
2326		}
2327
2328		/* Return the original packet for local processing. */
2329		return (m);
2330	}
2331
2332	if ((bif->bif_flags & IFBIF_STP) &&
2333	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2334		BRIDGE_UNLOCK(sc);
2335		return (m);
2336	}
2337
2338#if (defined(INET) || defined(INET6))
2339#   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2340	|| ((iface)->if_carp \
2341	    && (*carp_forus_p)((iface), eh->ether_dhost))
2342#   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2343	|| ((iface)->if_carp \
2344	    && (*carp_forus_p)((iface), eh->ether_shost))
2345#else
2346#   define OR_CARP_CHECK_WE_ARE_DST(iface)
2347#   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2348#endif
2349
2350#ifdef INET6
2351#   define OR_PFIL_HOOKED_INET6 \
2352	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2353#else
2354#   define OR_PFIL_HOOKED_INET6
2355#endif
2356
2357#define GRAB_OUR_PACKETS(iface) \
2358	if ((iface)->if_type == IFT_GIF) \
2359		continue; \
2360	/* It is destined for us. */ \
2361	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2362	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2363	    ) {								\
2364		if ((iface)->if_type == IFT_BRIDGE) {			\
2365			ETHER_BPF_MTAP(iface, m);			\
2366			iface->if_ipackets++;				\
2367			iface->if_ibytes += m->m_pkthdr.len;		\
2368			/* Filter on the physical interface. */		\
2369			if (pfil_local_phys &&				\
2370			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2371			     OR_PFIL_HOOKED_INET6)) {			\
2372				if (bridge_pfil(&m, NULL, ifp,		\
2373				    PFIL_IN) != 0 || m == NULL) {	\
2374					BRIDGE_UNLOCK(sc);		\
2375					return (NULL);			\
2376				}					\
2377				eh = mtod(m, struct ether_header *);	\
2378			}						\
2379		}							\
2380		if (bif->bif_flags & IFBIF_LEARNING) {			\
2381			error = bridge_rtupdate(sc, eh->ether_shost,	\
2382			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2383			if (error && bif->bif_addrmax) {		\
2384				BRIDGE_UNLOCK(sc);			\
2385				m_freem(m);				\
2386				return (NULL);				\
2387			}						\
2388		}							\
2389		m->m_pkthdr.rcvif = iface;				\
2390		BRIDGE_UNLOCK(sc);					\
2391		return (m);						\
2392	}								\
2393									\
2394	/* We just received a packet that we sent out. */		\
2395	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2396	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2397	    ) {								\
2398		BRIDGE_UNLOCK(sc);					\
2399		m_freem(m);						\
2400		return (NULL);						\
2401	}
2402
2403	/*
2404	 * Unicast.  Make sure it's not for the bridge.
2405	 */
2406	do { GRAB_OUR_PACKETS(bifp) } while (0);
2407
2408	/*
2409	 * Give a chance for ifp at first priority. This will help when	the
2410	 * packet comes through the interface like VLAN's with the same MACs
2411	 * on several interfaces from the same bridge. This also will save
2412	 * some CPU cycles in case the destination interface and the input
2413	 * interface (eq ifp) are the same.
2414	 */
2415	do { GRAB_OUR_PACKETS(ifp) } while (0);
2416
2417	/* Now check the all bridge members. */
2418	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2419		GRAB_OUR_PACKETS(bif2->bif_ifp)
2420	}
2421
2422#undef OR_CARP_CHECK_WE_ARE_DST
2423#undef OR_CARP_CHECK_WE_ARE_SRC
2424#undef OR_PFIL_HOOKED_INET6
2425#undef GRAB_OUR_PACKETS
2426
2427	/* Perform the bridge forwarding function. */
2428	bridge_forward(sc, bif, m);
2429
2430	return (NULL);
2431}
2432
2433/*
2434 * bridge_broadcast:
2435 *
2436 *	Send a frame to all interfaces that are members of
2437 *	the bridge, except for the one on which the packet
2438 *	arrived.
2439 *
2440 *	NOTE: Releases the lock on return.
2441 */
2442static void
2443bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2444    struct mbuf *m, int runfilt)
2445{
2446	struct bridge_iflist *dbif, *sbif;
2447	struct mbuf *mc;
2448	struct ifnet *dst_if;
2449	int error = 0, used = 0, i;
2450
2451	sbif = bridge_lookup_member_if(sc, src_if);
2452
2453	BRIDGE_LOCK2REF(sc, error);
2454	if (error) {
2455		m_freem(m);
2456		return;
2457	}
2458
2459	/* Filter on the bridge interface before broadcasting */
2460	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2461#ifdef INET6
2462	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2463#endif
2464	    )) {
2465		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2466			goto out;
2467		if (m == NULL)
2468			goto out;
2469	}
2470
2471	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2472		dst_if = dbif->bif_ifp;
2473		if (dst_if == src_if)
2474			continue;
2475
2476		/* Private segments can not talk to each other */
2477		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2478			continue;
2479
2480		if ((dbif->bif_flags & IFBIF_STP) &&
2481		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2482			continue;
2483
2484		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2485		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2486			continue;
2487
2488		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2489			continue;
2490
2491		if (LIST_NEXT(dbif, bif_next) == NULL) {
2492			mc = m;
2493			used = 1;
2494		} else {
2495			mc = m_dup(m, M_NOWAIT);
2496			if (mc == NULL) {
2497				sc->sc_ifp->if_oerrors++;
2498				continue;
2499			}
2500		}
2501
2502		/*
2503		 * Filter on the output interface. Pass a NULL bridge interface
2504		 * pointer so we do not redundantly filter on the bridge for
2505		 * each interface we broadcast on.
2506		 */
2507		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2508#ifdef INET6
2509		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2510#endif
2511		    )) {
2512			if (used == 0) {
2513				/* Keep the layer3 header aligned */
2514				i = min(mc->m_pkthdr.len, max_protohdr);
2515				mc = m_copyup(mc, i, ETHER_ALIGN);
2516				if (mc == NULL) {
2517					sc->sc_ifp->if_oerrors++;
2518					continue;
2519				}
2520			}
2521			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2522				continue;
2523			if (mc == NULL)
2524				continue;
2525		}
2526
2527		bridge_enqueue(sc, dst_if, mc);
2528	}
2529	if (used == 0)
2530		m_freem(m);
2531
2532out:
2533	BRIDGE_UNREF(sc);
2534}
2535
2536/*
2537 * bridge_span:
2538 *
2539 *	Duplicate a packet out one or more interfaces that are in span mode,
2540 *	the original mbuf is unmodified.
2541 */
2542static void
2543bridge_span(struct bridge_softc *sc, struct mbuf *m)
2544{
2545	struct bridge_iflist *bif;
2546	struct ifnet *dst_if;
2547	struct mbuf *mc;
2548
2549	if (LIST_EMPTY(&sc->sc_spanlist))
2550		return;
2551
2552	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2553		dst_if = bif->bif_ifp;
2554
2555		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2556			continue;
2557
2558		mc = m_copypacket(m, M_NOWAIT);
2559		if (mc == NULL) {
2560			sc->sc_ifp->if_oerrors++;
2561			continue;
2562		}
2563
2564		bridge_enqueue(sc, dst_if, mc);
2565	}
2566}
2567
2568/*
2569 * bridge_rtupdate:
2570 *
2571 *	Add a bridge routing entry.
2572 */
2573static int
2574bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2575    struct bridge_iflist *bif, int setflags, uint8_t flags)
2576{
2577	struct bridge_rtnode *brt;
2578	int error;
2579
2580	BRIDGE_LOCK_ASSERT(sc);
2581
2582	/* Check the source address is valid and not multicast. */
2583	if (ETHER_IS_MULTICAST(dst) ||
2584	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2585	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2586		return (EINVAL);
2587
2588	/* 802.1p frames map to vlan 1 */
2589	if (vlan == 0)
2590		vlan = 1;
2591
2592	/*
2593	 * A route for this destination might already exist.  If so,
2594	 * update it, otherwise create a new one.
2595	 */
2596	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2597		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2598			sc->sc_brtexceeded++;
2599			return (ENOSPC);
2600		}
2601		/* Check per interface address limits (if enabled) */
2602		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2603			bif->bif_addrexceeded++;
2604			return (ENOSPC);
2605		}
2606
2607		/*
2608		 * Allocate a new bridge forwarding node, and
2609		 * initialize the expiration time and Ethernet
2610		 * address.
2611		 */
2612		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2613		if (brt == NULL)
2614			return (ENOMEM);
2615
2616		if (bif->bif_flags & IFBIF_STICKY)
2617			brt->brt_flags = IFBAF_STICKY;
2618		else
2619			brt->brt_flags = IFBAF_DYNAMIC;
2620
2621		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2622		brt->brt_vlan = vlan;
2623
2624		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2625			uma_zfree(bridge_rtnode_zone, brt);
2626			return (error);
2627		}
2628		brt->brt_dst = bif;
2629		bif->bif_addrcnt++;
2630	}
2631
2632	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2633	    brt->brt_dst != bif) {
2634		brt->brt_dst->bif_addrcnt--;
2635		brt->brt_dst = bif;
2636		brt->brt_dst->bif_addrcnt++;
2637	}
2638
2639	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2640		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2641	if (setflags)
2642		brt->brt_flags = flags;
2643
2644	return (0);
2645}
2646
2647/*
2648 * bridge_rtlookup:
2649 *
2650 *	Lookup the destination interface for an address.
2651 */
2652static struct ifnet *
2653bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2654{
2655	struct bridge_rtnode *brt;
2656
2657	BRIDGE_LOCK_ASSERT(sc);
2658
2659	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2660		return (NULL);
2661
2662	return (brt->brt_ifp);
2663}
2664
2665/*
2666 * bridge_rttrim:
2667 *
2668 *	Trim the routine table so that we have a number
2669 *	of routing entries less than or equal to the
2670 *	maximum number.
2671 */
2672static void
2673bridge_rttrim(struct bridge_softc *sc)
2674{
2675	struct bridge_rtnode *brt, *nbrt;
2676
2677	BRIDGE_LOCK_ASSERT(sc);
2678
2679	/* Make sure we actually need to do this. */
2680	if (sc->sc_brtcnt <= sc->sc_brtmax)
2681		return;
2682
2683	/* Force an aging cycle; this might trim enough addresses. */
2684	bridge_rtage(sc);
2685	if (sc->sc_brtcnt <= sc->sc_brtmax)
2686		return;
2687
2688	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2689		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2690			bridge_rtnode_destroy(sc, brt);
2691			if (sc->sc_brtcnt <= sc->sc_brtmax)
2692				return;
2693		}
2694	}
2695}
2696
2697/*
2698 * bridge_timer:
2699 *
2700 *	Aging timer for the bridge.
2701 */
2702static void
2703bridge_timer(void *arg)
2704{
2705	struct bridge_softc *sc = arg;
2706
2707	BRIDGE_LOCK_ASSERT(sc);
2708
2709	bridge_rtage(sc);
2710
2711	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2712		callout_reset(&sc->sc_brcallout,
2713		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2714}
2715
2716/*
2717 * bridge_rtage:
2718 *
2719 *	Perform an aging cycle.
2720 */
2721static void
2722bridge_rtage(struct bridge_softc *sc)
2723{
2724	struct bridge_rtnode *brt, *nbrt;
2725
2726	BRIDGE_LOCK_ASSERT(sc);
2727
2728	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2729		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2730			if (time_uptime >= brt->brt_expire)
2731				bridge_rtnode_destroy(sc, brt);
2732		}
2733	}
2734}
2735
2736/*
2737 * bridge_rtflush:
2738 *
2739 *	Remove all dynamic addresses from the bridge.
2740 */
2741static void
2742bridge_rtflush(struct bridge_softc *sc, int full)
2743{
2744	struct bridge_rtnode *brt, *nbrt;
2745
2746	BRIDGE_LOCK_ASSERT(sc);
2747
2748	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2749		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2750			bridge_rtnode_destroy(sc, brt);
2751	}
2752}
2753
2754/*
2755 * bridge_rtdaddr:
2756 *
2757 *	Remove an address from the table.
2758 */
2759static int
2760bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2761{
2762	struct bridge_rtnode *brt;
2763	int found = 0;
2764
2765	BRIDGE_LOCK_ASSERT(sc);
2766
2767	/*
2768	 * If vlan is zero then we want to delete for all vlans so the lookup
2769	 * may return more than one.
2770	 */
2771	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2772		bridge_rtnode_destroy(sc, brt);
2773		found = 1;
2774	}
2775
2776	return (found ? 0 : ENOENT);
2777}
2778
2779/*
2780 * bridge_rtdelete:
2781 *
2782 *	Delete routes to a speicifc member interface.
2783 */
2784static void
2785bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2786{
2787	struct bridge_rtnode *brt, *nbrt;
2788
2789	BRIDGE_LOCK_ASSERT(sc);
2790
2791	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2792		if (brt->brt_ifp == ifp && (full ||
2793			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2794			bridge_rtnode_destroy(sc, brt);
2795	}
2796}
2797
2798/*
2799 * bridge_rtable_init:
2800 *
2801 *	Initialize the route table for this bridge.
2802 */
2803static void
2804bridge_rtable_init(struct bridge_softc *sc)
2805{
2806	int i;
2807
2808	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2809	    M_DEVBUF, M_WAITOK);
2810
2811	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2812		LIST_INIT(&sc->sc_rthash[i]);
2813
2814	sc->sc_rthash_key = arc4random();
2815	LIST_INIT(&sc->sc_rtlist);
2816}
2817
2818/*
2819 * bridge_rtable_fini:
2820 *
2821 *	Deconstruct the route table for this bridge.
2822 */
2823static void
2824bridge_rtable_fini(struct bridge_softc *sc)
2825{
2826
2827	KASSERT(sc->sc_brtcnt == 0,
2828	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2829	free(sc->sc_rthash, M_DEVBUF);
2830}
2831
2832/*
2833 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2834 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2835 */
2836#define	mix(a, b, c)							\
2837do {									\
2838	a -= b; a -= c; a ^= (c >> 13);					\
2839	b -= c; b -= a; b ^= (a << 8);					\
2840	c -= a; c -= b; c ^= (b >> 13);					\
2841	a -= b; a -= c; a ^= (c >> 12);					\
2842	b -= c; b -= a; b ^= (a << 16);					\
2843	c -= a; c -= b; c ^= (b >> 5);					\
2844	a -= b; a -= c; a ^= (c >> 3);					\
2845	b -= c; b -= a; b ^= (a << 10);					\
2846	c -= a; c -= b; c ^= (b >> 15);					\
2847} while (/*CONSTCOND*/0)
2848
2849static __inline uint32_t
2850bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2851{
2852	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2853
2854	b += addr[5] << 8;
2855	b += addr[4];
2856	a += addr[3] << 24;
2857	a += addr[2] << 16;
2858	a += addr[1] << 8;
2859	a += addr[0];
2860
2861	mix(a, b, c);
2862
2863	return (c & BRIDGE_RTHASH_MASK);
2864}
2865
2866#undef mix
2867
2868static int
2869bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2870{
2871	int i, d;
2872
2873	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2874		d = ((int)a[i]) - ((int)b[i]);
2875	}
2876
2877	return (d);
2878}
2879
2880/*
2881 * bridge_rtnode_lookup:
2882 *
2883 *	Look up a bridge route node for the specified destination. Compare the
2884 *	vlan id or if zero then just return the first match.
2885 */
2886static struct bridge_rtnode *
2887bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2888{
2889	struct bridge_rtnode *brt;
2890	uint32_t hash;
2891	int dir;
2892
2893	BRIDGE_LOCK_ASSERT(sc);
2894
2895	hash = bridge_rthash(sc, addr);
2896	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2897		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2898		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2899			return (brt);
2900		if (dir > 0)
2901			return (NULL);
2902	}
2903
2904	return (NULL);
2905}
2906
2907/*
2908 * bridge_rtnode_insert:
2909 *
2910 *	Insert the specified bridge node into the route table.  We
2911 *	assume the entry is not already in the table.
2912 */
2913static int
2914bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2915{
2916	struct bridge_rtnode *lbrt;
2917	uint32_t hash;
2918	int dir;
2919
2920	BRIDGE_LOCK_ASSERT(sc);
2921
2922	hash = bridge_rthash(sc, brt->brt_addr);
2923
2924	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2925	if (lbrt == NULL) {
2926		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2927		goto out;
2928	}
2929
2930	do {
2931		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2932		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2933			return (EEXIST);
2934		if (dir > 0) {
2935			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2936			goto out;
2937		}
2938		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2939			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2940			goto out;
2941		}
2942		lbrt = LIST_NEXT(lbrt, brt_hash);
2943	} while (lbrt != NULL);
2944
2945#ifdef DIAGNOSTIC
2946	panic("bridge_rtnode_insert: impossible");
2947#endif
2948
2949out:
2950	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2951	sc->sc_brtcnt++;
2952
2953	return (0);
2954}
2955
2956/*
2957 * bridge_rtnode_destroy:
2958 *
2959 *	Destroy a bridge rtnode.
2960 */
2961static void
2962bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2963{
2964	BRIDGE_LOCK_ASSERT(sc);
2965
2966	LIST_REMOVE(brt, brt_hash);
2967
2968	LIST_REMOVE(brt, brt_list);
2969	sc->sc_brtcnt--;
2970	brt->brt_dst->bif_addrcnt--;
2971	uma_zfree(bridge_rtnode_zone, brt);
2972}
2973
2974/*
2975 * bridge_rtable_expire:
2976 *
2977 *	Set the expiry time for all routes on an interface.
2978 */
2979static void
2980bridge_rtable_expire(struct ifnet *ifp, int age)
2981{
2982	struct bridge_softc *sc = ifp->if_bridge;
2983	struct bridge_rtnode *brt;
2984
2985	BRIDGE_LOCK(sc);
2986
2987	/*
2988	 * If the age is zero then flush, otherwise set all the expiry times to
2989	 * age for the interface
2990	 */
2991	if (age == 0)
2992		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2993	else {
2994		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2995			/* Cap the expiry time to 'age' */
2996			if (brt->brt_ifp == ifp &&
2997			    brt->brt_expire > time_uptime + age &&
2998			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2999				brt->brt_expire = time_uptime + age;
3000		}
3001	}
3002	BRIDGE_UNLOCK(sc);
3003}
3004
3005/*
3006 * bridge_state_change:
3007 *
3008 *	Callback from the bridgestp code when a port changes states.
3009 */
3010static void
3011bridge_state_change(struct ifnet *ifp, int state)
3012{
3013	struct bridge_softc *sc = ifp->if_bridge;
3014	static const char *stpstates[] = {
3015		"disabled",
3016		"listening",
3017		"learning",
3018		"forwarding",
3019		"blocking",
3020		"discarding"
3021	};
3022
3023	if (log_stp)
3024		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3025		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3026}
3027
3028/*
3029 * Send bridge packets through pfil if they are one of the types pfil can deal
3030 * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3031 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3032 * that interface.
3033 */
3034static int
3035bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3036{
3037	int snap, error, i, hlen;
3038	struct ether_header *eh1, eh2;
3039	struct ip *ip;
3040	struct llc llc1;
3041	u_int16_t ether_type;
3042
3043	snap = 0;
3044	error = -1;	/* Default error if not error == 0 */
3045
3046#if 0
3047	/* we may return with the IP fields swapped, ensure its not shared */
3048	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3049#endif
3050
3051	if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
3052		return (0); /* filtering is disabled */
3053
3054	i = min((*mp)->m_pkthdr.len, max_protohdr);
3055	if ((*mp)->m_len < i) {
3056	    *mp = m_pullup(*mp, i);
3057	    if (*mp == NULL) {
3058		printf("%s: m_pullup failed\n", __func__);
3059		return (-1);
3060	    }
3061	}
3062
3063	eh1 = mtod(*mp, struct ether_header *);
3064	ether_type = ntohs(eh1->ether_type);
3065
3066	/*
3067	 * Check for SNAP/LLC.
3068	 */
3069	if (ether_type < ETHERMTU) {
3070		struct llc *llc2 = (struct llc *)(eh1 + 1);
3071
3072		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3073		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3074		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3075		    llc2->llc_control == LLC_UI) {
3076			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3077			snap = 1;
3078		}
3079	}
3080
3081	/*
3082	 * If we're trying to filter bridge traffic, don't look at anything
3083	 * other than IP and ARP traffic.  If the filter doesn't understand
3084	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3085	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3086	 * but of course we don't have an AppleTalk filter to begin with.
3087	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3088	 * ARP traffic.)
3089	 */
3090	switch (ether_type) {
3091		case ETHERTYPE_ARP:
3092		case ETHERTYPE_REVARP:
3093			if (pfil_ipfw_arp == 0)
3094				return (0); /* Automatically pass */
3095			break;
3096
3097		case ETHERTYPE_IP:
3098#ifdef INET6
3099		case ETHERTYPE_IPV6:
3100#endif /* INET6 */
3101			break;
3102		default:
3103			/*
3104			 * Check to see if the user wants to pass non-ip
3105			 * packets, these will not be checked by pfil(9) and
3106			 * passed unconditionally so the default is to drop.
3107			 */
3108			if (pfil_onlyip)
3109				goto bad;
3110	}
3111
3112	/* Run the packet through pfil before stripping link headers */
3113	if (PFIL_HOOKED(&V_link_pfil_hook) && pfil_ipfw != 0 &&
3114			dir == PFIL_OUT && ifp != NULL) {
3115
3116		error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, NULL);
3117
3118		if (*mp == NULL || error != 0) /* packet consumed by filter */
3119			return (error);
3120	}
3121
3122	/* Strip off the Ethernet header and keep a copy. */
3123	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3124	m_adj(*mp, ETHER_HDR_LEN);
3125
3126	/* Strip off snap header, if present */
3127	if (snap) {
3128		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3129		m_adj(*mp, sizeof(struct llc));
3130	}
3131
3132	/*
3133	 * Check the IP header for alignment and errors
3134	 */
3135	if (dir == PFIL_IN) {
3136		switch (ether_type) {
3137			case ETHERTYPE_IP:
3138				error = bridge_ip_checkbasic(mp);
3139				break;
3140#ifdef INET6
3141			case ETHERTYPE_IPV6:
3142				error = bridge_ip6_checkbasic(mp);
3143				break;
3144#endif /* INET6 */
3145			default:
3146				error = 0;
3147		}
3148		if (error)
3149			goto bad;
3150	}
3151
3152	error = 0;
3153
3154	/*
3155	 * Run the packet through pfil
3156	 */
3157	switch (ether_type) {
3158	case ETHERTYPE_IP:
3159		/*
3160		 * Run pfil on the member interface and the bridge, both can
3161		 * be skipped by clearing pfil_member or pfil_bridge.
3162		 *
3163		 * Keep the order:
3164		 *   in_if -> bridge_if -> out_if
3165		 */
3166		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3167			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3168					dir, NULL);
3169
3170		if (*mp == NULL || error != 0) /* filter may consume */
3171			break;
3172
3173		if (pfil_member && ifp != NULL)
3174			error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3175					dir, NULL);
3176
3177		if (*mp == NULL || error != 0) /* filter may consume */
3178			break;
3179
3180		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3181			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3182					dir, NULL);
3183
3184		if (*mp == NULL || error != 0) /* filter may consume */
3185			break;
3186
3187		/* check if we need to fragment the packet */
3188		/* bridge_fragment generates a mbuf chain of packets */
3189		/* that already include eth headers */
3190		if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3191			i = (*mp)->m_pkthdr.len;
3192			if (i > ifp->if_mtu) {
3193				error = bridge_fragment(ifp, mp, &eh2, snap,
3194					    &llc1);
3195				return (error);
3196			}
3197		}
3198
3199		/* Recalculate the ip checksum. */
3200		ip = mtod(*mp, struct ip *);
3201		hlen = ip->ip_hl << 2;
3202		if (hlen < sizeof(struct ip))
3203			goto bad;
3204		if (hlen > (*mp)->m_len) {
3205			if ((*mp = m_pullup(*mp, hlen)) == 0)
3206				goto bad;
3207			ip = mtod(*mp, struct ip *);
3208			if (ip == NULL)
3209				goto bad;
3210		}
3211		ip->ip_sum = 0;
3212		if (hlen == sizeof(struct ip))
3213			ip->ip_sum = in_cksum_hdr(ip);
3214		else
3215			ip->ip_sum = in_cksum(*mp, hlen);
3216
3217		break;
3218#ifdef INET6
3219	case ETHERTYPE_IPV6:
3220		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3221			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3222					dir, NULL);
3223
3224		if (*mp == NULL || error != 0) /* filter may consume */
3225			break;
3226
3227		if (pfil_member && ifp != NULL)
3228			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3229					dir, NULL);
3230
3231		if (*mp == NULL || error != 0) /* filter may consume */
3232			break;
3233
3234		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3235			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3236					dir, NULL);
3237		break;
3238#endif
3239	default:
3240		error = 0;
3241		break;
3242	}
3243
3244	if (*mp == NULL)
3245		return (error);
3246	if (error != 0)
3247		goto bad;
3248
3249	error = -1;
3250
3251	/*
3252	 * Finally, put everything back the way it was and return
3253	 */
3254	if (snap) {
3255		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3256		if (*mp == NULL)
3257			return (error);
3258		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3259	}
3260
3261	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3262	if (*mp == NULL)
3263		return (error);
3264	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3265
3266	return (0);
3267
3268bad:
3269	m_freem(*mp);
3270	*mp = NULL;
3271	return (error);
3272}
3273
3274/*
3275 * Perform basic checks on header size since
3276 * pfil assumes ip_input has already processed
3277 * it for it.  Cut-and-pasted from ip_input.c.
3278 * Given how simple the IPv6 version is,
3279 * does the IPv4 version really need to be
3280 * this complicated?
3281 *
3282 * XXX Should we update ipstat here, or not?
3283 * XXX Right now we update ipstat but not
3284 * XXX csum_counter.
3285 */
3286static int
3287bridge_ip_checkbasic(struct mbuf **mp)
3288{
3289	struct mbuf *m = *mp;
3290	struct ip *ip;
3291	int len, hlen;
3292	u_short sum;
3293
3294	if (*mp == NULL)
3295		return (-1);
3296
3297	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3298		if ((m = m_copyup(m, sizeof(struct ip),
3299			(max_linkhdr + 3) & ~3)) == NULL) {
3300			/* XXXJRT new stat, please */
3301			KMOD_IPSTAT_INC(ips_toosmall);
3302			goto bad;
3303		}
3304	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3305		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3306			KMOD_IPSTAT_INC(ips_toosmall);
3307			goto bad;
3308		}
3309	}
3310	ip = mtod(m, struct ip *);
3311	if (ip == NULL) goto bad;
3312
3313	if (ip->ip_v != IPVERSION) {
3314		KMOD_IPSTAT_INC(ips_badvers);
3315		goto bad;
3316	}
3317	hlen = ip->ip_hl << 2;
3318	if (hlen < sizeof(struct ip)) { /* minimum header length */
3319		KMOD_IPSTAT_INC(ips_badhlen);
3320		goto bad;
3321	}
3322	if (hlen > m->m_len) {
3323		if ((m = m_pullup(m, hlen)) == 0) {
3324			KMOD_IPSTAT_INC(ips_badhlen);
3325			goto bad;
3326		}
3327		ip = mtod(m, struct ip *);
3328		if (ip == NULL) goto bad;
3329	}
3330
3331	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3332		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3333	} else {
3334		if (hlen == sizeof(struct ip)) {
3335			sum = in_cksum_hdr(ip);
3336		} else {
3337			sum = in_cksum(m, hlen);
3338		}
3339	}
3340	if (sum) {
3341		KMOD_IPSTAT_INC(ips_badsum);
3342		goto bad;
3343	}
3344
3345	/* Retrieve the packet length. */
3346	len = ntohs(ip->ip_len);
3347
3348	/*
3349	 * Check for additional length bogosity
3350	 */
3351	if (len < hlen) {
3352		KMOD_IPSTAT_INC(ips_badlen);
3353		goto bad;
3354	}
3355
3356	/*
3357	 * Check that the amount of data in the buffers
3358	 * is as at least much as the IP header would have us expect.
3359	 * Drop packet if shorter than we expect.
3360	 */
3361	if (m->m_pkthdr.len < len) {
3362		KMOD_IPSTAT_INC(ips_tooshort);
3363		goto bad;
3364	}
3365
3366	/* Checks out, proceed */
3367	*mp = m;
3368	return (0);
3369
3370bad:
3371	*mp = m;
3372	return (-1);
3373}
3374
3375#ifdef INET6
3376/*
3377 * Same as above, but for IPv6.
3378 * Cut-and-pasted from ip6_input.c.
3379 * XXX Should we update ip6stat, or not?
3380 */
3381static int
3382bridge_ip6_checkbasic(struct mbuf **mp)
3383{
3384	struct mbuf *m = *mp;
3385	struct ip6_hdr *ip6;
3386
3387	/*
3388	 * If the IPv6 header is not aligned, slurp it up into a new
3389	 * mbuf with space for link headers, in the event we forward
3390	 * it.  Otherwise, if it is aligned, make sure the entire base
3391	 * IPv6 header is in the first mbuf of the chain.
3392	 */
3393	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3394		struct ifnet *inifp = m->m_pkthdr.rcvif;
3395		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3396			    (max_linkhdr + 3) & ~3)) == NULL) {
3397			/* XXXJRT new stat, please */
3398			IP6STAT_INC(ip6s_toosmall);
3399			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3400			goto bad;
3401		}
3402	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3403		struct ifnet *inifp = m->m_pkthdr.rcvif;
3404		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3405			IP6STAT_INC(ip6s_toosmall);
3406			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3407			goto bad;
3408		}
3409	}
3410
3411	ip6 = mtod(m, struct ip6_hdr *);
3412
3413	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3414		IP6STAT_INC(ip6s_badvers);
3415		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3416		goto bad;
3417	}
3418
3419	/* Checks out, proceed */
3420	*mp = m;
3421	return (0);
3422
3423bad:
3424	*mp = m;
3425	return (-1);
3426}
3427#endif /* INET6 */
3428
3429/*
3430 * bridge_fragment:
3431 *
3432 *	Fragment mbuf chain in multiple packets and prepend ethernet header.
3433 */
3434static int
3435bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3436    int snap, struct llc *llc)
3437{
3438	struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3439	struct ip *ip;
3440	int error = -1;
3441
3442	if (m->m_len < sizeof(struct ip) &&
3443	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3444		goto dropit;
3445	ip = mtod(m, struct ip *);
3446
3447	m->m_pkthdr.csum_flags |= CSUM_IP;
3448	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3449	if (error)
3450		goto dropit;
3451
3452	/*
3453	 * Walk the chain and re-add the Ethernet header for
3454	 * each mbuf packet.
3455	 */
3456	for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3457		nextpkt = mcur->m_nextpkt;
3458		mcur->m_nextpkt = NULL;
3459		if (snap) {
3460			M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3461			if (mcur == NULL) {
3462				error = ENOBUFS;
3463				if (mprev != NULL)
3464					mprev->m_nextpkt = nextpkt;
3465				goto dropit;
3466			}
3467			bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3468		}
3469
3470		M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3471		if (mcur == NULL) {
3472			error = ENOBUFS;
3473			if (mprev != NULL)
3474				mprev->m_nextpkt = nextpkt;
3475			goto dropit;
3476		}
3477		bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3478
3479		/*
3480		 * The previous two M_PREPEND could have inserted one or two
3481		 * mbufs in front so we have to update the previous packet's
3482		 * m_nextpkt.
3483		 */
3484		mcur->m_nextpkt = nextpkt;
3485		if (mprev != NULL)
3486			mprev->m_nextpkt = mcur;
3487		else {
3488			/* The first mbuf in the original chain needs to be
3489			 * updated. */
3490			*mp = mcur;
3491		}
3492		mprev = mcur;
3493	}
3494
3495	KMOD_IPSTAT_INC(ips_fragmented);
3496	return (error);
3497
3498dropit:
3499	for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3500		m = mcur->m_nextpkt;
3501		m_freem(mcur);
3502	}
3503	return (error);
3504}
3505
3506static void
3507bridge_linkstate(struct ifnet *ifp)
3508{
3509	struct bridge_softc *sc = ifp->if_bridge;
3510	struct bridge_iflist *bif;
3511
3512	BRIDGE_LOCK(sc);
3513	bif = bridge_lookup_member_if(sc, ifp);
3514	if (bif == NULL) {
3515		BRIDGE_UNLOCK(sc);
3516		return;
3517	}
3518	bridge_linkcheck(sc);
3519	BRIDGE_UNLOCK(sc);
3520
3521	bstp_linkstate(&bif->bif_stp);
3522}
3523
3524static void
3525bridge_linkcheck(struct bridge_softc *sc)
3526{
3527	struct bridge_iflist *bif;
3528	int new_link, hasls;
3529
3530	BRIDGE_LOCK_ASSERT(sc);
3531	new_link = LINK_STATE_DOWN;
3532	hasls = 0;
3533	/* Our link is considered up if at least one of our ports is active */
3534	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3535		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3536			hasls++;
3537		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3538			new_link = LINK_STATE_UP;
3539			break;
3540		}
3541	}
3542	if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3543		/* If no interfaces support link-state then we default to up */
3544		new_link = LINK_STATE_UP;
3545	}
3546	if_link_state_change(sc->sc_ifp, new_link);
3547}
3548