1 /* $NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $ */
4 * SPDX-License-Identifier: BSD-4-Clause
6 * Copyright 2001 Wasabi Systems, Inc.
9 * Written by Jason R. Thorpe for Wasabi Systems, Inc.
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. All advertising materials mentioning features or use of this software
20 * must display the following acknowledgement:
21 * This product includes software developed for the NetBSD Project by
22 * Wasabi Systems, Inc.
23 * 4. The name of Wasabi Systems, Inc. may not be used to endorse
24 * or promote products derived from this software without specific prior
27 * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
30 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC
31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37 * POSSIBILITY OF SUCH DAMAGE.
41 * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
42 * All rights reserved.
44 * Redistribution and use in source and binary forms, with or without
45 * modification, are permitted provided that the following conditions
47 * 1. Redistributions of source code must retain the above copyright
48 * notice, this list of conditions and the following disclaimer.
49 * 2. Redistributions in binary form must reproduce the above copyright
50 * notice, this list of conditions and the following disclaimer in the
51 * documentation and/or other materials provided with the distribution.
53 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
54 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
55 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
56 * DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
57 * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
58 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
59 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
61 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
62 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
63 * POSSIBILITY OF SUCH DAMAGE.
65 * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
69 * Network interface bridge support.
73 * - Currently only supports Ethernet-like interfaces (Ethernet,
74 * 802.11, VLANs on Ethernet, etc.) Figure out a nice way
75 * to bridge other types of interfaces (maybe consider
76 * heterogeneous bridges).
79 #include <sys/cdefs.h>
80 __FBSDID("$FreeBSD$");
83 #include "opt_inet6.h"
85 #include <sys/param.h>
86 #include <sys/eventhandler.h>
88 #include <sys/malloc.h>
89 #include <sys/protosw.h>
90 #include <sys/systm.h>
93 #include <sys/socket.h> /* for net/if.h */
94 #include <sys/sockio.h>
95 #include <sys/ctype.h> /* string functions */
96 #include <sys/kernel.h>
97 #include <sys/random.h>
98 #include <sys/syslog.h>
99 #include <sys/sysctl.h>
101 #include <sys/module.h>
102 #include <sys/priv.h>
103 #include <sys/proc.h>
104 #include <sys/lock.h>
105 #include <sys/mutex.h>
109 #include <net/if_clone.h>
110 #include <net/if_dl.h>
111 #include <net/if_types.h>
112 #include <net/if_var.h>
113 #include <net/pfil.h>
114 #include <net/vnet.h>
116 #include <netinet/in.h>
117 #include <netinet/in_systm.h>
118 #include <netinet/in_var.h>
119 #include <netinet/ip.h>
120 #include <netinet/ip_var.h>
122 #include <netinet/ip6.h>
123 #include <netinet6/ip6_var.h>
124 #include <netinet6/in6_ifattach.h>
126 #if defined(INET) || defined(INET6)
127 #include <netinet/ip_carp.h>
129 #include <machine/in_cksum.h>
130 #include <netinet/if_ether.h>
131 #include <net/bridgestp.h>
132 #include <net/if_bridgevar.h>
133 #include <net/if_llc.h>
134 #include <net/if_vlan_var.h>
136 #include <net/route.h>
140 * XXX: declare here to avoid to include many inet6 related files..
141 * should be more generalized?
143 extern void nd6_setmtu(struct ifnet *);
147 * Size of the route hash table. Must be a power of two.
149 #ifndef BRIDGE_RTHASH_SIZE
150 #define BRIDGE_RTHASH_SIZE 1024
153 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1)
156 * Default maximum number of addresses to cache.
158 #ifndef BRIDGE_RTABLE_MAX
159 #define BRIDGE_RTABLE_MAX 2000
163 * Timeout (in seconds) for entries learned dynamically.
165 #ifndef BRIDGE_RTABLE_TIMEOUT
166 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */
170 * Number of seconds between walks of the route list.
172 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
173 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60)
177 * List of capabilities to possibly mask on the member interface.
179 #define BRIDGE_IFCAPS_MASK (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
183 * List of capabilities to strip
185 #define BRIDGE_IFCAPS_STRIP IFCAP_LRO
190 #define BRIDGE_LOCK_INIT(_sc) do { \
191 mtx_init(&(_sc)->sc_mtx, "if_bridge", NULL, MTX_DEF); \
193 #define BRIDGE_LOCK_DESTROY(_sc) do { \
194 mtx_destroy(&(_sc)->sc_mtx); \
196 #define BRIDGE_LOCK(_sc) mtx_lock(&(_sc)->sc_mtx)
197 #define BRIDGE_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_mtx)
198 #define BRIDGE_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_mtx, MA_OWNED)
199 #define BRIDGE_UNLOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_mtx, MA_NOTOWNED)
202 * Bridge interface list entry.
204 struct bridge_iflist {
205 CK_LIST_ENTRY(bridge_iflist) bif_next;
206 struct ifnet *bif_ifp; /* member if */
207 struct bstp_port bif_stp; /* STP state */
208 uint32_t bif_flags; /* member if flags */
209 int bif_savedcaps; /* saved capabilities */
210 uint32_t bif_addrmax; /* max # of addresses */
211 uint32_t bif_addrcnt; /* cur. # of addresses */
212 uint32_t bif_addrexceeded;/* # of address violations */
213 struct epoch_context bif_epoch_ctx;
219 struct bridge_rtnode {
220 CK_LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
221 CK_LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
222 struct bridge_iflist *brt_dst; /* destination if */
223 unsigned long brt_expire; /* expiration time */
224 uint8_t brt_flags; /* address flags */
225 uint8_t brt_addr[ETHER_ADDR_LEN];
226 uint16_t brt_vlan; /* vlan id */
227 struct vnet *brt_vnet;
228 struct epoch_context brt_epoch_ctx;
230 #define brt_ifp brt_dst->bif_ifp
233 * Software state for each bridge.
235 struct bridge_softc {
236 struct ifnet *sc_ifp; /* make this an interface */
237 LIST_ENTRY(bridge_softc) sc_list;
239 uint32_t sc_brtmax; /* max # of addresses */
240 uint32_t sc_brtcnt; /* cur. # of addresses */
241 uint32_t sc_brttimeout; /* rt timeout in seconds */
242 struct callout sc_brcallout; /* bridge callout */
243 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
244 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
245 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
246 uint32_t sc_rthash_key; /* key for hash */
247 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
248 struct bstp_state sc_stp; /* STP state */
249 uint32_t sc_brtexceeded; /* # of cache drops */
250 struct ifnet *sc_ifaddr; /* member mac copied from */
251 struct ether_addr sc_defaddr; /* Default MAC address */
252 struct epoch_context sc_epoch_ctx;
255 VNET_DEFINE_STATIC(struct mtx, bridge_list_mtx);
256 #define V_bridge_list_mtx VNET(bridge_list_mtx)
257 static eventhandler_tag bridge_detach_cookie;
259 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
261 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
262 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone)
264 static int bridge_clone_create(struct if_clone *, int, caddr_t);
265 static void bridge_clone_destroy(struct ifnet *);
267 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
268 static void bridge_mutecaps(struct bridge_softc *);
269 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
271 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
272 static void bridge_init(void *);
273 static void bridge_dummynet(struct mbuf *, struct ifnet *);
274 static void bridge_stop(struct ifnet *, int);
275 static int bridge_transmit(struct ifnet *, struct mbuf *);
276 static void bridge_qflush(struct ifnet *);
277 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
278 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
280 static int bridge_enqueue(struct bridge_softc *, struct ifnet *,
282 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
284 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
287 static void bridge_timer(void *);
289 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
291 static void bridge_span(struct bridge_softc *, struct mbuf *);
293 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
294 uint16_t, struct bridge_iflist *, int, uint8_t);
295 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
297 static void bridge_rttrim(struct bridge_softc *);
298 static void bridge_rtage(struct bridge_softc *);
299 static void bridge_rtflush(struct bridge_softc *, int);
300 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
303 static void bridge_rtable_init(struct bridge_softc *);
304 static void bridge_rtable_fini(struct bridge_softc *);
306 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
307 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
308 const uint8_t *, uint16_t);
309 static int bridge_rtnode_insert(struct bridge_softc *,
310 struct bridge_rtnode *);
311 static void bridge_rtnode_destroy(struct bridge_softc *,
312 struct bridge_rtnode *);
313 static void bridge_rtable_expire(struct ifnet *, int);
314 static void bridge_state_change(struct ifnet *, int);
316 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
318 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
320 static void bridge_delete_member(struct bridge_softc *,
321 struct bridge_iflist *, int);
322 static void bridge_delete_span(struct bridge_softc *,
323 struct bridge_iflist *);
325 static int bridge_ioctl_add(struct bridge_softc *, void *);
326 static int bridge_ioctl_del(struct bridge_softc *, void *);
327 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
328 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
329 static int bridge_ioctl_scache(struct bridge_softc *, void *);
330 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
331 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
332 static int bridge_ioctl_rts(struct bridge_softc *, void *);
333 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
334 static int bridge_ioctl_sto(struct bridge_softc *, void *);
335 static int bridge_ioctl_gto(struct bridge_softc *, void *);
336 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
337 static int bridge_ioctl_flush(struct bridge_softc *, void *);
338 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
339 static int bridge_ioctl_spri(struct bridge_softc *, void *);
340 static int bridge_ioctl_ght(struct bridge_softc *, void *);
341 static int bridge_ioctl_sht(struct bridge_softc *, void *);
342 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
343 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
344 static int bridge_ioctl_gma(struct bridge_softc *, void *);
345 static int bridge_ioctl_sma(struct bridge_softc *, void *);
346 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
347 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
348 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
349 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
350 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
351 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
352 static int bridge_ioctl_grte(struct bridge_softc *, void *);
353 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
354 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
355 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
356 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
358 static int bridge_ip_checkbasic(struct mbuf **mp);
360 static int bridge_ip6_checkbasic(struct mbuf **mp);
362 static int bridge_fragment(struct ifnet *, struct mbuf **mp,
363 struct ether_header *, int, struct llc *);
364 static void bridge_linkstate(struct ifnet *ifp);
365 static void bridge_linkcheck(struct bridge_softc *sc);
368 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
369 #define VLANTAGOF(_m) \
370 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
372 static struct bstp_cb_ops bridge_ops = {
373 .bcb_state = bridge_state_change,
374 .bcb_rtage = bridge_rtable_expire
377 SYSCTL_DECL(_net_link);
378 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
381 /* only pass IP[46] packets when pfil is enabled */
382 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
383 #define V_pfil_onlyip VNET(pfil_onlyip)
384 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
385 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
386 "Only pass IP packets when pfil is enabled");
388 /* run pfil hooks on the bridge interface */
389 VNET_DEFINE_STATIC(int, pfil_bridge) = 1;
390 #define V_pfil_bridge VNET(pfil_bridge)
391 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
392 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
393 "Packet filter on the bridge interface");
395 /* layer2 filter with ipfw */
396 VNET_DEFINE_STATIC(int, pfil_ipfw);
397 #define V_pfil_ipfw VNET(pfil_ipfw)
399 /* layer2 ARP filter with ipfw */
400 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
401 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
402 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
403 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
404 "Filter ARP packets through IPFW layer2");
406 /* run pfil hooks on the member interface */
407 VNET_DEFINE_STATIC(int, pfil_member) = 1;
408 #define V_pfil_member VNET(pfil_member)
409 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
410 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
411 "Packet filter on the member interface");
413 /* run pfil hooks on the physical interface for locally destined packets */
414 VNET_DEFINE_STATIC(int, pfil_local_phys);
415 #define V_pfil_local_phys VNET(pfil_local_phys)
416 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
417 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
418 "Packet filter on the physical interface for locally destined packets");
420 /* log STP state changes */
421 VNET_DEFINE_STATIC(int, log_stp);
422 #define V_log_stp VNET(log_stp)
423 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
424 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
425 "Log STP state changes");
427 /* share MAC with first bridge member */
428 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
429 #define V_bridge_inherit_mac VNET(bridge_inherit_mac)
430 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
431 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
432 "Inherit MAC address from the first bridge member");
434 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
435 #define V_allow_llz_overlap VNET(allow_llz_overlap)
436 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
437 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
438 "Allow overlap of link-local scope "
439 "zones of a bridge interface and the member interfaces");
441 struct bridge_control {
442 int (*bc_func)(struct bridge_softc *, void *);
447 #define BC_F_COPYIN 0x01 /* copy arguments in */
448 #define BC_F_COPYOUT 0x02 /* copy arguments out */
449 #define BC_F_SUSER 0x04 /* do super-user check */
451 const struct bridge_control bridge_control_table[] = {
452 { bridge_ioctl_add, sizeof(struct ifbreq),
453 BC_F_COPYIN|BC_F_SUSER },
454 { bridge_ioctl_del, sizeof(struct ifbreq),
455 BC_F_COPYIN|BC_F_SUSER },
457 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
458 BC_F_COPYIN|BC_F_COPYOUT },
459 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
460 BC_F_COPYIN|BC_F_SUSER },
462 { bridge_ioctl_scache, sizeof(struct ifbrparam),
463 BC_F_COPYIN|BC_F_SUSER },
464 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
467 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
468 BC_F_COPYIN|BC_F_COPYOUT },
469 { bridge_ioctl_rts, sizeof(struct ifbaconf),
470 BC_F_COPYIN|BC_F_COPYOUT },
472 { bridge_ioctl_saddr, sizeof(struct ifbareq),
473 BC_F_COPYIN|BC_F_SUSER },
475 { bridge_ioctl_sto, sizeof(struct ifbrparam),
476 BC_F_COPYIN|BC_F_SUSER },
477 { bridge_ioctl_gto, sizeof(struct ifbrparam),
480 { bridge_ioctl_daddr, sizeof(struct ifbareq),
481 BC_F_COPYIN|BC_F_SUSER },
483 { bridge_ioctl_flush, sizeof(struct ifbreq),
484 BC_F_COPYIN|BC_F_SUSER },
486 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
488 { bridge_ioctl_spri, sizeof(struct ifbrparam),
489 BC_F_COPYIN|BC_F_SUSER },
491 { bridge_ioctl_ght, sizeof(struct ifbrparam),
493 { bridge_ioctl_sht, sizeof(struct ifbrparam),
494 BC_F_COPYIN|BC_F_SUSER },
496 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
498 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
499 BC_F_COPYIN|BC_F_SUSER },
501 { bridge_ioctl_gma, sizeof(struct ifbrparam),
503 { bridge_ioctl_sma, sizeof(struct ifbrparam),
504 BC_F_COPYIN|BC_F_SUSER },
506 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
507 BC_F_COPYIN|BC_F_SUSER },
509 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
510 BC_F_COPYIN|BC_F_SUSER },
512 { bridge_ioctl_addspan, sizeof(struct ifbreq),
513 BC_F_COPYIN|BC_F_SUSER },
514 { bridge_ioctl_delspan, sizeof(struct ifbreq),
515 BC_F_COPYIN|BC_F_SUSER },
517 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
520 { bridge_ioctl_grte, sizeof(struct ifbrparam),
523 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
524 BC_F_COPYIN|BC_F_COPYOUT },
526 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
527 BC_F_COPYIN|BC_F_SUSER },
529 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
530 BC_F_COPYIN|BC_F_SUSER },
532 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
533 BC_F_COPYIN|BC_F_SUSER },
536 const int bridge_control_table_size = nitems(bridge_control_table);
538 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
539 #define V_bridge_list VNET(bridge_list)
540 #define BRIDGE_LIST_LOCK_INIT(x) mtx_init(&V_bridge_list_mtx, \
541 "if_bridge list", NULL, MTX_DEF)
542 #define BRIDGE_LIST_LOCK_DESTROY(x) mtx_destroy(&V_bridge_list_mtx)
543 #define BRIDGE_LIST_LOCK(x) mtx_lock(&V_bridge_list_mtx)
544 #define BRIDGE_LIST_UNLOCK(x) mtx_unlock(&V_bridge_list_mtx)
546 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
547 #define V_bridge_cloner VNET(bridge_cloner)
549 static const char bridge_name[] = "bridge";
552 vnet_bridge_init(const void *unused __unused)
555 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
556 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
558 BRIDGE_LIST_LOCK_INIT();
559 LIST_INIT(&V_bridge_list);
560 V_bridge_cloner = if_clone_simple(bridge_name,
561 bridge_clone_create, bridge_clone_destroy, 0);
563 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
564 vnet_bridge_init, NULL);
567 vnet_bridge_uninit(const void *unused __unused)
570 if_clone_detach(V_bridge_cloner);
571 V_bridge_cloner = NULL;
572 BRIDGE_LIST_LOCK_DESTROY();
574 /* Callbacks may use the UMA zone. */
575 epoch_drain_callbacks(net_epoch_preempt);
577 uma_zdestroy(V_bridge_rtnode_zone);
579 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
580 vnet_bridge_uninit, NULL);
583 bridge_modevent(module_t mod, int type, void *data)
588 bridge_dn_p = bridge_dummynet;
589 bridge_detach_cookie = EVENTHANDLER_REGISTER(
590 ifnet_departure_event, bridge_ifdetach, NULL,
591 EVENTHANDLER_PRI_ANY);
594 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
595 bridge_detach_cookie);
604 static moduledata_t bridge_mod = {
610 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
611 MODULE_VERSION(if_bridge, 1);
612 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
615 * handler for net.link.bridge.ipfw
618 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
620 int enable = V_pfil_ipfw;
623 error = sysctl_handle_int(oidp, &enable, 0, req);
626 if (enable != V_pfil_ipfw) {
627 V_pfil_ipfw = enable;
630 * Disable pfil so that ipfw doesnt run twice, if the user
631 * really wants both then they can re-enable pfil_bridge and/or
632 * pfil_member. Also allow non-ip packets as ipfw can filter by
644 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
645 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT,
646 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
647 "Layer2 filter with IPFW");
650 * bridge_clone_create:
652 * Create a new bridge instance.
655 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
657 struct bridge_softc *sc;
660 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
661 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
667 BRIDGE_LOCK_INIT(sc);
668 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
669 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
671 /* Initialize our routing table. */
672 bridge_rtable_init(sc);
674 callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
676 CK_LIST_INIT(&sc->sc_iflist);
677 CK_LIST_INIT(&sc->sc_spanlist);
680 if_initname(ifp, bridge_name, unit);
681 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
682 ifp->if_ioctl = bridge_ioctl;
683 ifp->if_transmit = bridge_transmit;
684 ifp->if_qflush = bridge_qflush;
685 ifp->if_init = bridge_init;
686 ifp->if_type = IFT_BRIDGE;
688 ether_gen_addr(ifp, &sc->sc_defaddr);
690 bstp_attach(&sc->sc_stp, &bridge_ops);
691 ether_ifattach(ifp, sc->sc_defaddr.octet);
692 /* Now undo some of the damage... */
693 ifp->if_baudrate = 0;
694 ifp->if_type = IFT_BRIDGE;
697 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
698 BRIDGE_LIST_UNLOCK();
704 bridge_clone_destroy_cb(struct epoch_context *ctx)
706 struct bridge_softc *sc;
708 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
710 BRIDGE_LOCK_DESTROY(sc);
715 * bridge_clone_destroy:
717 * Destroy a bridge instance.
720 bridge_clone_destroy(struct ifnet *ifp)
722 struct bridge_softc *sc = ifp->if_softc;
723 struct bridge_iflist *bif;
724 struct epoch_tracker et;
730 ifp->if_flags &= ~IFF_UP;
732 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
733 bridge_delete_member(sc, bif, 0);
735 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
736 bridge_delete_span(sc, bif);
739 /* Tear down the routing table. */
740 bridge_rtable_fini(sc);
744 callout_drain(&sc->sc_brcallout);
747 LIST_REMOVE(sc, sc_list);
748 BRIDGE_LIST_UNLOCK();
750 bstp_detach(&sc->sc_stp);
756 NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx);
762 * Handle a control request from the operator.
765 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
767 struct bridge_softc *sc = ifp->if_softc;
768 struct ifreq *ifr = (struct ifreq *)data;
769 struct bridge_iflist *bif;
770 struct thread *td = curthread;
772 struct ifbreq ifbreq;
773 struct ifbifconf ifbifconf;
774 struct ifbareq ifbareq;
775 struct ifbaconf ifbaconf;
776 struct ifbrparam ifbrparam;
777 struct ifbropreq ifbropreq;
779 struct ifdrv *ifd = (struct ifdrv *) data;
780 const struct bridge_control *bc;
781 int error = 0, oldmtu;
782 struct epoch_tracker et;
794 if (ifd->ifd_cmd >= bridge_control_table_size) {
798 bc = &bridge_control_table[ifd->ifd_cmd];
800 if (cmd == SIOCGDRVSPEC &&
801 (bc->bc_flags & BC_F_COPYOUT) == 0) {
805 else if (cmd == SIOCSDRVSPEC &&
806 (bc->bc_flags & BC_F_COPYOUT) != 0) {
811 if (bc->bc_flags & BC_F_SUSER) {
812 error = priv_check(td, PRIV_NET_BRIDGE);
817 if (ifd->ifd_len != bc->bc_argsize ||
818 ifd->ifd_len > sizeof(args)) {
823 bzero(&args, sizeof(args));
824 if (bc->bc_flags & BC_F_COPYIN) {
825 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
830 oldmtu = ifp->if_mtu;
832 error = (*bc->bc_func)(sc, &args);
838 * Bridge MTU may change during addition of the first port.
839 * If it did, do network layer specific procedure.
841 if (ifp->if_mtu != oldmtu) {
848 if (bc->bc_flags & BC_F_COPYOUT)
849 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
854 if (!(ifp->if_flags & IFF_UP) &&
855 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
857 * If interface is marked down and it is running,
858 * then stop and disable it.
863 } else if ((ifp->if_flags & IFF_UP) &&
864 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
866 * If interface is marked up and it is stopped, then
874 if (ifr->ifr_mtu < 576) {
878 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
879 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
883 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
884 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
885 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
886 " != %d\n", sc->sc_ifp->if_xname,
887 bif->bif_ifp->if_mtu,
888 bif->bif_ifp->if_xname, ifr->ifr_mtu);
894 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
899 * drop the lock as ether_ioctl() will call bridge_start() and
900 * cause the lock to be recursed.
902 error = ether_ioctl(ifp, cmd, data);
914 * Clear or restore unwanted capabilities on the member interface
917 bridge_mutecaps(struct bridge_softc *sc)
919 struct bridge_iflist *bif;
922 BRIDGE_LOCK_ASSERT(sc);
924 /* Initial bitmask of capabilities to test */
925 mask = BRIDGE_IFCAPS_MASK;
927 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
928 /* Every member must support it or its disabled */
929 mask &= bif->bif_savedcaps;
932 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
933 enabled = bif->bif_ifp->if_capenable;
934 enabled &= ~BRIDGE_IFCAPS_STRIP;
935 /* strip off mask bits and enable them again if allowed */
936 enabled &= ~BRIDGE_IFCAPS_MASK;
939 bridge_set_ifcap(sc, bif, enabled);
945 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
947 struct ifnet *ifp = bif->bif_ifp;
949 int error, mask, stuck;
951 BRIDGE_UNLOCK_ASSERT(sc);
953 bzero(&ifr, sizeof(ifr));
954 ifr.ifr_reqcap = set;
956 if (ifp->if_capenable != set) {
957 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
959 if_printf(sc->sc_ifp,
960 "error setting capabilities on %s: %d\n",
961 ifp->if_xname, error);
962 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
963 stuck = ifp->if_capenable & mask & ~set;
965 if_printf(sc->sc_ifp,
966 "can't disable some capabilities on %s: 0x%x\n",
967 ifp->if_xname, stuck);
972 * bridge_lookup_member:
974 * Lookup a bridge member interface.
976 static struct bridge_iflist *
977 bridge_lookup_member(struct bridge_softc *sc, const char *name)
979 struct bridge_iflist *bif;
984 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
986 if (strcmp(ifp->if_xname, name) == 0)
994 * bridge_lookup_member_if:
996 * Lookup a bridge member interface by ifnet*.
998 static struct bridge_iflist *
999 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1001 struct bridge_iflist *bif;
1005 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1006 if (bif->bif_ifp == member_ifp)
1014 bridge_delete_member_cb(struct epoch_context *ctx)
1016 struct bridge_iflist *bif;
1018 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1020 free(bif, M_DEVBUF);
1024 * bridge_delete_member:
1026 * Delete the specified member interface.
1029 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1032 struct ifnet *ifs = bif->bif_ifp;
1033 struct ifnet *fif = NULL;
1034 struct bridge_iflist *bifl;
1036 BRIDGE_LOCK_ASSERT(sc);
1038 if (bif->bif_flags & IFBIF_STP)
1039 bstp_disable(&bif->bif_stp);
1041 ifs->if_bridge = NULL;
1042 CK_LIST_REMOVE(bif, bif_next);
1045 * If removing the interface that gave the bridge its mac address, set
1046 * the mac address of the bridge to the address of the next member, or
1047 * to its default address if no members are left.
1049 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1050 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1051 bcopy(&sc->sc_defaddr,
1052 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1053 sc->sc_ifaddr = NULL;
1055 bifl = CK_LIST_FIRST(&sc->sc_iflist);
1056 fif = bifl->bif_ifp;
1057 bcopy(IF_LLADDR(fif),
1058 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1059 sc->sc_ifaddr = fif;
1061 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1064 bridge_linkcheck(sc);
1065 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1066 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1067 KASSERT(bif->bif_addrcnt == 0,
1068 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1070 ifs->if_bridge_output = NULL;
1071 ifs->if_bridge_input = NULL;
1072 ifs->if_bridge_linkstate = NULL;
1075 switch (ifs->if_type) {
1079 * Take the interface out of promiscuous mode, but only
1080 * if it was promiscuous in the first place. It might
1081 * not be if we're in the bridge_ioctl_add() error path.
1083 if (ifs->if_flags & IFF_PROMISC)
1084 (void) ifpromisc(ifs, 0);
1092 panic("bridge_delete_member: impossible");
1096 /* reneable any interface capabilities */
1097 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1099 bstp_destroy(&bif->bif_stp); /* prepare to free */
1102 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1106 * bridge_delete_span:
1108 * Delete the specified span interface.
1111 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1113 BRIDGE_LOCK_ASSERT(sc);
1115 KASSERT(bif->bif_ifp->if_bridge == NULL,
1116 ("%s: not a span interface", __func__));
1118 CK_LIST_REMOVE(bif, bif_next);
1120 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1124 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1126 struct ifbreq *req = arg;
1127 struct bridge_iflist *bif = NULL;
1131 ifs = ifunit(req->ifbr_ifsname);
1134 if (ifs->if_ioctl == NULL) /* must be supported */
1137 /* If it's in the span list, it can't be a member. */
1138 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1139 if (ifs == bif->bif_ifp)
1142 if (ifs->if_bridge == sc)
1145 if (ifs->if_bridge != NULL)
1148 switch (ifs->if_type) {
1152 /* permitted interface types */
1160 * Two valid inet6 addresses with link-local scope must not be
1161 * on the parent interface and the member interfaces at the
1162 * same time. This restriction is needed to prevent violation
1163 * of link-local scope zone. Attempts to add a member
1164 * interface which has inet6 addresses when the parent has
1165 * inet6 triggers removal of all inet6 addresses on the member
1169 /* Check if the parent interface has a link-local scope addr. */
1170 if (V_allow_llz_overlap == 0 &&
1171 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1173 * If any, remove all inet6 addresses from the member
1176 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1177 if (in6ifa_llaonifp(bif->bif_ifp)) {
1179 in6_ifdetach(bif->bif_ifp);
1181 if_printf(sc->sc_ifp,
1182 "IPv6 addresses on %s have been removed "
1183 "before adding it as a member to prevent "
1184 "IPv6 address scope violation.\n",
1185 bif->bif_ifp->if_xname);
1188 if (in6ifa_llaonifp(ifs)) {
1192 if_printf(sc->sc_ifp,
1193 "IPv6 addresses on %s have been removed "
1194 "before adding it as a member to prevent "
1195 "IPv6 address scope violation.\n",
1200 /* Allow the first Ethernet member to define the MTU */
1201 if (CK_LIST_EMPTY(&sc->sc_iflist))
1202 sc->sc_ifp->if_mtu = ifs->if_mtu;
1203 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1204 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1205 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1209 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1214 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1215 bif->bif_savedcaps = ifs->if_capenable;
1218 * Assign the interface's MAC address to the bridge if it's the first
1219 * member and the MAC address of the bridge has not been changed from
1220 * the default randomly generated one.
1222 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1223 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1224 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1225 sc->sc_ifaddr = ifs;
1226 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1229 ifs->if_bridge = sc;
1230 ifs->if_bridge_output = bridge_output;
1231 ifs->if_bridge_input = bridge_input;
1232 ifs->if_bridge_linkstate = bridge_linkstate;
1233 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1235 * XXX: XLOCK HERE!?!
1237 * NOTE: insert_***HEAD*** should be safe for the traversals.
1239 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1241 /* Set interface capabilities to the intersection set of all members */
1242 bridge_mutecaps(sc);
1243 bridge_linkcheck(sc);
1245 /* Place the interface into promiscuous mode */
1246 switch (ifs->if_type) {
1250 error = ifpromisc(ifs, 1);
1256 bridge_delete_member(sc, bif, 0);
1261 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1263 struct ifbreq *req = arg;
1264 struct bridge_iflist *bif;
1266 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1270 bridge_delete_member(sc, bif, 0);
1276 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1278 struct ifbreq *req = arg;
1279 struct bridge_iflist *bif;
1280 struct bstp_port *bp;
1282 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1287 req->ifbr_ifsflags = bif->bif_flags;
1288 req->ifbr_state = bp->bp_state;
1289 req->ifbr_priority = bp->bp_priority;
1290 req->ifbr_path_cost = bp->bp_path_cost;
1291 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1292 req->ifbr_proto = bp->bp_protover;
1293 req->ifbr_role = bp->bp_role;
1294 req->ifbr_stpflags = bp->bp_flags;
1295 req->ifbr_addrcnt = bif->bif_addrcnt;
1296 req->ifbr_addrmax = bif->bif_addrmax;
1297 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1299 /* Copy STP state options as flags */
1300 if (bp->bp_operedge)
1301 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1302 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1303 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1304 if (bp->bp_ptp_link)
1305 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1306 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1307 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1308 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1309 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1310 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1311 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1316 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1318 struct ifbreq *req = arg;
1319 struct bridge_iflist *bif;
1320 struct bstp_port *bp;
1323 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1328 if (req->ifbr_ifsflags & IFBIF_SPAN)
1329 /* SPAN is readonly */
1332 if (req->ifbr_ifsflags & IFBIF_STP) {
1333 if ((bif->bif_flags & IFBIF_STP) == 0) {
1334 error = bstp_enable(&bif->bif_stp);
1339 if ((bif->bif_flags & IFBIF_STP) != 0)
1340 bstp_disable(&bif->bif_stp);
1343 /* Pass on STP flags */
1344 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1345 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1346 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1347 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1349 /* Save the bits relating to the bridge */
1350 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1356 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1358 struct ifbrparam *param = arg;
1360 sc->sc_brtmax = param->ifbrp_csize;
1367 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1369 struct ifbrparam *param = arg;
1371 param->ifbrp_csize = sc->sc_brtmax;
1377 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1379 struct ifbifconf *bifc = arg;
1380 struct bridge_iflist *bif;
1383 int count, buflen, len, error = 0;
1386 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1388 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1391 buflen = sizeof(breq) * count;
1392 if (bifc->ifbic_len == 0) {
1393 bifc->ifbic_len = buflen;
1396 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1402 len = min(bifc->ifbic_len, buflen);
1403 bzero(&breq, sizeof(breq));
1404 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1405 if (len < sizeof(breq))
1408 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1409 sizeof(breq.ifbr_ifsname));
1410 /* Fill in the ifbreq structure */
1411 error = bridge_ioctl_gifflags(sc, &breq);
1414 memcpy(buf, &breq, sizeof(breq));
1416 buf += sizeof(breq);
1417 len -= sizeof(breq);
1419 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1420 if (len < sizeof(breq))
1423 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1424 sizeof(breq.ifbr_ifsname));
1425 breq.ifbr_ifsflags = bif->bif_flags;
1426 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1427 memcpy(buf, &breq, sizeof(breq));
1429 buf += sizeof(breq);
1430 len -= sizeof(breq);
1434 bifc->ifbic_len = sizeof(breq) * count;
1435 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1437 free(outbuf, M_TEMP);
1442 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1444 struct ifbaconf *bac = arg;
1445 struct bridge_rtnode *brt;
1446 struct ifbareq bareq;
1448 int count, buflen, len, error = 0;
1450 if (bac->ifbac_len == 0)
1454 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1456 buflen = sizeof(bareq) * count;
1458 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1464 len = min(bac->ifbac_len, buflen);
1465 bzero(&bareq, sizeof(bareq));
1466 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1467 if (len < sizeof(bareq))
1469 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1470 sizeof(bareq.ifba_ifsname));
1471 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1472 bareq.ifba_vlan = brt->brt_vlan;
1473 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1474 time_uptime < brt->brt_expire)
1475 bareq.ifba_expire = brt->brt_expire - time_uptime;
1477 bareq.ifba_expire = 0;
1478 bareq.ifba_flags = brt->brt_flags;
1480 memcpy(buf, &bareq, sizeof(bareq));
1482 buf += sizeof(bareq);
1483 len -= sizeof(bareq);
1487 bac->ifbac_len = sizeof(bareq) * count;
1488 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1490 free(outbuf, M_TEMP);
1495 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1497 struct ifbareq *req = arg;
1498 struct bridge_iflist *bif;
1503 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1507 /* bridge_rtupdate() may acquire the lock. */
1509 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1517 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1519 struct ifbrparam *param = arg;
1521 sc->sc_brttimeout = param->ifbrp_ctime;
1526 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1528 struct ifbrparam *param = arg;
1530 param->ifbrp_ctime = sc->sc_brttimeout;
1535 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1537 struct ifbareq *req = arg;
1539 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1543 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1545 struct ifbreq *req = arg;
1547 bridge_rtflush(sc, req->ifbr_ifsflags);
1552 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1554 struct ifbrparam *param = arg;
1555 struct bstp_state *bs = &sc->sc_stp;
1557 param->ifbrp_prio = bs->bs_bridge_priority;
1562 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1564 struct ifbrparam *param = arg;
1566 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1570 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1572 struct ifbrparam *param = arg;
1573 struct bstp_state *bs = &sc->sc_stp;
1575 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1580 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1582 struct ifbrparam *param = arg;
1584 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1588 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1590 struct ifbrparam *param = arg;
1591 struct bstp_state *bs = &sc->sc_stp;
1593 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1598 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1600 struct ifbrparam *param = arg;
1602 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1606 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1608 struct ifbrparam *param = arg;
1609 struct bstp_state *bs = &sc->sc_stp;
1611 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1616 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1618 struct ifbrparam *param = arg;
1620 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1624 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1626 struct ifbreq *req = arg;
1627 struct bridge_iflist *bif;
1629 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1633 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1637 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1639 struct ifbreq *req = arg;
1640 struct bridge_iflist *bif;
1642 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1646 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1650 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1652 struct ifbreq *req = arg;
1653 struct bridge_iflist *bif;
1655 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1659 bif->bif_addrmax = req->ifbr_addrmax;
1664 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1666 struct ifbreq *req = arg;
1667 struct bridge_iflist *bif = NULL;
1670 ifs = ifunit(req->ifbr_ifsname);
1674 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1675 if (ifs == bif->bif_ifp)
1678 if (ifs->if_bridge != NULL)
1681 switch (ifs->if_type) {
1690 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1695 bif->bif_flags = IFBIF_SPAN;
1697 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1703 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1705 struct ifbreq *req = arg;
1706 struct bridge_iflist *bif;
1709 ifs = ifunit(req->ifbr_ifsname);
1713 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1714 if (ifs == bif->bif_ifp)
1720 bridge_delete_span(sc, bif);
1726 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1728 struct ifbropreq *req = arg;
1729 struct bstp_state *bs = &sc->sc_stp;
1730 struct bstp_port *root_port;
1732 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1733 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1734 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1736 root_port = bs->bs_root_port;
1737 if (root_port == NULL)
1738 req->ifbop_root_port = 0;
1740 req->ifbop_root_port = root_port->bp_ifp->if_index;
1742 req->ifbop_holdcount = bs->bs_txholdcount;
1743 req->ifbop_priority = bs->bs_bridge_priority;
1744 req->ifbop_protocol = bs->bs_protover;
1745 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1746 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1747 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1748 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1749 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1750 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1756 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1758 struct ifbrparam *param = arg;
1760 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1765 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1767 struct ifbpstpconf *bifstp = arg;
1768 struct bridge_iflist *bif;
1769 struct bstp_port *bp;
1770 struct ifbpstpreq bpreq;
1772 int count, buflen, len, error = 0;
1775 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1776 if ((bif->bif_flags & IFBIF_STP) != 0)
1780 buflen = sizeof(bpreq) * count;
1781 if (bifstp->ifbpstp_len == 0) {
1782 bifstp->ifbpstp_len = buflen;
1786 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1792 len = min(bifstp->ifbpstp_len, buflen);
1793 bzero(&bpreq, sizeof(bpreq));
1794 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1795 if (len < sizeof(bpreq))
1798 if ((bif->bif_flags & IFBIF_STP) == 0)
1802 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1803 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1804 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1805 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1806 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1807 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1809 memcpy(buf, &bpreq, sizeof(bpreq));
1811 buf += sizeof(bpreq);
1812 len -= sizeof(bpreq);
1816 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1817 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1819 free(outbuf, M_TEMP);
1824 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1826 struct ifbrparam *param = arg;
1828 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1832 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1834 struct ifbrparam *param = arg;
1836 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1842 * Detach an interface from a bridge. Called when a member
1843 * interface is detaching.
1846 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1848 struct bridge_softc *sc = ifp->if_bridge;
1849 struct bridge_iflist *bif;
1850 struct epoch_tracker et;
1852 if (ifp->if_flags & IFF_RENAMING)
1854 if (V_bridge_cloner == NULL) {
1856 * This detach handler can be called after
1857 * vnet_bridge_uninit(). Just return in that case.
1861 NET_EPOCH_ENTER(et);
1862 /* Check if the interface is a bridge member */
1866 bif = bridge_lookup_member_if(sc, ifp);
1868 bridge_delete_member(sc, bif, 1);
1875 /* Check if the interface is a span port */
1877 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1879 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1880 if (ifp == bif->bif_ifp) {
1881 bridge_delete_span(sc, bif);
1887 BRIDGE_LIST_UNLOCK();
1894 * Initialize a bridge interface.
1897 bridge_init(void *xsc)
1899 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1900 struct ifnet *ifp = sc->sc_ifp;
1902 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1906 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1909 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1910 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1918 * Stop the bridge interface.
1921 bridge_stop(struct ifnet *ifp, int disable)
1923 struct bridge_softc *sc = ifp->if_softc;
1926 BRIDGE_LOCK_ASSERT(sc);
1928 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1931 callout_stop(&sc->sc_brcallout);
1932 bstp_stop(&sc->sc_stp);
1934 bridge_rtflush(sc, IFBF_FLUSHDYN);
1936 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1942 * Enqueue a packet on a bridge member interface.
1946 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1952 /* We may be sending a fragment so traverse the mbuf */
1955 m->m_nextpkt = NULL;
1956 len = m->m_pkthdr.len;
1957 mflags = m->m_flags;
1960 * If underlying interface can not do VLAN tag insertion itself
1961 * then attach a packet tag that holds it.
1963 if ((m->m_flags & M_VLANTAG) &&
1964 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1965 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1968 "unable to prepend VLAN header\n");
1969 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
1972 m->m_flags &= ~M_VLANTAG;
1975 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
1976 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1978 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
1982 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
1983 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
1984 if (mflags & M_MCAST)
1985 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
1994 * Receive a queued packet from dummynet and pass it on to the output
1997 * The mbuf has the Ethernet header already attached.
2000 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2002 struct bridge_softc *sc;
2004 sc = ifp->if_bridge;
2007 * The packet didnt originate from a member interface. This should only
2008 * ever happen if a member interface is removed while packets are
2016 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2018 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2021 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2027 bridge_enqueue(sc, ifp, m);
2033 * Send output from a bridge member interface. This
2034 * performs the bridging function for locally originated
2037 * The mbuf has the Ethernet header already attached. We must
2038 * enqueue or free the mbuf before returning.
2041 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2044 struct ether_header *eh;
2045 struct ifnet *bifp, *dst_if;
2046 struct bridge_softc *sc;
2051 if (m->m_len < ETHER_HDR_LEN) {
2052 m = m_pullup(m, ETHER_HDR_LEN);
2057 eh = mtod(m, struct ether_header *);
2058 sc = ifp->if_bridge;
2059 vlan = VLANTAGOF(m);
2064 * If bridge is down, but the original output interface is up,
2065 * go ahead and send out that interface. Otherwise, the packet
2068 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2074 * If the packet is a multicast, or we don't know a better way to
2075 * get there, send to all interfaces.
2077 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2080 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2081 /* Tap any traffic not passing back out the originating interface */
2083 ETHER_BPF_MTAP(bifp, m);
2084 if (dst_if == NULL) {
2085 struct bridge_iflist *bif;
2091 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2092 dst_if = bif->bif_ifp;
2094 if (dst_if->if_type == IFT_GIF)
2096 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2100 * If this is not the original output interface,
2101 * and the interface is participating in spanning
2102 * tree, make sure the port is in a state that
2103 * allows forwarding.
2105 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2106 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2109 if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2113 mc = m_copypacket(m, M_NOWAIT);
2115 if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2120 bridge_enqueue(sc, dst_if, mc);
2129 * XXX Spanning tree consideration here?
2133 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2138 bridge_enqueue(sc, dst_if, m);
2145 * Do output on a bridge.
2149 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2151 struct bridge_softc *sc;
2152 struct ether_header *eh;
2153 struct ifnet *dst_if;
2158 ETHER_BPF_MTAP(ifp, m);
2160 eh = mtod(m, struct ether_header *);
2162 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2163 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2164 error = bridge_enqueue(sc, dst_if, m);
2166 bridge_broadcast(sc, ifp, m, 0);
2172 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2175 bridge_qflush(struct ifnet *ifp __unused)
2182 * The forwarding function of the bridge.
2184 * NOTE: Releases the lock on return.
2187 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2190 struct bridge_iflist *dbif;
2191 struct ifnet *src_if, *dst_if, *ifp;
2192 struct ether_header *eh;
2199 src_if = m->m_pkthdr.rcvif;
2202 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2203 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2204 vlan = VLANTAGOF(m);
2206 if ((sbif->bif_flags & IFBIF_STP) &&
2207 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2210 eh = mtod(m, struct ether_header *);
2211 dst = eh->ether_dhost;
2213 /* If the interface is learning, record the address. */
2214 if (sbif->bif_flags & IFBIF_LEARNING) {
2215 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2216 sbif, 0, IFBAF_DYNAMIC);
2218 * If the interface has addresses limits then deny any source
2219 * that is not in the cache.
2221 if (error && sbif->bif_addrmax)
2225 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2226 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2230 * At this point, the port either doesn't participate
2231 * in spanning tree or it is in the forwarding state.
2235 * If the packet is unicast, destined for someone on
2236 * "this" side of the bridge, drop it.
2238 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2239 dst_if = bridge_rtlookup(sc, dst, vlan);
2240 if (src_if == dst_if)
2244 * Check if its a reserved multicast address, any address
2245 * listed in 802.1D section 7.12.6 may not be forwarded by the
2247 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2249 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2250 dst[2] == 0xc2 && dst[3] == 0x00 &&
2251 dst[4] == 0x00 && dst[5] <= 0x0f)
2254 /* ...forward it to all interfaces. */
2255 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2260 * If we have a destination interface which is a member of our bridge,
2261 * OR this is a unicast packet, push it through the bpf(4) machinery.
2262 * For broadcast or multicast packets, don't bother because it will
2263 * be reinjected into ether_input. We do this before we pass the packets
2264 * through the pfil(9) framework, as it is possible that pfil(9) will
2265 * drop the packet, or possibly modify it, making it difficult to debug
2266 * firewall issues on the bridge.
2268 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2269 ETHER_BPF_MTAP(ifp, m);
2271 /* run the packet filter */
2272 if (PFIL_HOOKED_IN(V_inet_pfil_head)
2274 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2277 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2283 if (dst_if == NULL) {
2284 bridge_broadcast(sc, src_if, m, 1);
2289 * At this point, we're dealing with a unicast frame
2290 * going to a different interface.
2292 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2295 dbif = bridge_lookup_member_if(sc, dst_if);
2297 /* Not a member of the bridge (anymore?) */
2300 /* Private segments can not talk to each other */
2301 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2304 if ((dbif->bif_flags & IFBIF_STP) &&
2305 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2308 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2310 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2313 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2319 bridge_enqueue(sc, dst_if, m);
2329 * Receive input from a member interface. Queue the packet for
2330 * bridging if it is not for us.
2332 static struct mbuf *
2333 bridge_input(struct ifnet *ifp, struct mbuf *m)
2335 struct bridge_softc *sc = ifp->if_bridge;
2336 struct bridge_iflist *bif, *bif2;
2338 struct ether_header *eh;
2339 struct mbuf *mc, *mc2;
2345 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2349 vlan = VLANTAGOF(m);
2352 * Implement support for bridge monitoring. If this flag has been
2353 * set on this interface, discard the packet once we push it through
2354 * the bpf(4) machinery, but before we do, increment the byte and
2355 * packet counters associated with this interface.
2357 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2358 m->m_pkthdr.rcvif = bifp;
2359 ETHER_BPF_MTAP(bifp, m);
2360 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2361 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2365 bif = bridge_lookup_member_if(sc, ifp);
2370 eh = mtod(m, struct ether_header *);
2374 if (m->m_flags & (M_BCAST|M_MCAST)) {
2375 /* Tap off 802.1D packets; they do not get forwarded. */
2376 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2377 ETHER_ADDR_LEN) == 0) {
2378 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2382 if ((bif->bif_flags & IFBIF_STP) &&
2383 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2388 * Make a deep copy of the packet and enqueue the copy
2389 * for bridge processing; return the original packet for
2392 mc = m_dup(m, M_NOWAIT);
2397 /* Perform the bridge forwarding function with the copy. */
2398 bridge_forward(sc, bif, mc);
2401 * Reinject the mbuf as arriving on the bridge so we have a
2402 * chance at claiming multicast packets. We can not loop back
2403 * here from ether_input as a bridge is never a member of a
2406 KASSERT(bifp->if_bridge == NULL,
2407 ("loop created in bridge_input"));
2408 mc2 = m_dup(m, M_NOWAIT);
2410 /* Keep the layer3 header aligned */
2411 int i = min(mc2->m_pkthdr.len, max_protohdr);
2412 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2415 mc2->m_pkthdr.rcvif = bifp;
2416 (*bifp->if_input)(bifp, mc2);
2419 /* Return the original packet for local processing. */
2423 if ((bif->bif_flags & IFBIF_STP) &&
2424 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2428 #if (defined(INET) || defined(INET6))
2429 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2430 || ((iface)->if_carp \
2431 && (*carp_forus_p)((iface), eh->ether_dhost))
2432 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2433 || ((iface)->if_carp \
2434 && (*carp_forus_p)((iface), eh->ether_shost))
2436 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2437 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2441 # define OR_PFIL_HOOKED_INET6 \
2442 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2444 # define OR_PFIL_HOOKED_INET6
2447 #define GRAB_OUR_PACKETS(iface) \
2448 if ((iface)->if_type == IFT_GIF) \
2450 /* It is destined for us. */ \
2451 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2452 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2454 if (bif->bif_flags & IFBIF_LEARNING) { \
2455 error = bridge_rtupdate(sc, eh->ether_shost, \
2456 vlan, bif, 0, IFBAF_DYNAMIC); \
2457 if (error && bif->bif_addrmax) { \
2462 m->m_pkthdr.rcvif = iface; \
2463 if ((iface) == ifp) { \
2464 /* Skip bridge processing... src == dest */ \
2467 /* It's passing over or to the bridge, locally. */ \
2468 ETHER_BPF_MTAP(bifp, m); \
2469 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \
2470 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2471 /* Filter on the physical interface. */ \
2472 if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \
2473 OR_PFIL_HOOKED_INET6)) { \
2474 if (bridge_pfil(&m, NULL, ifp, \
2475 PFIL_IN) != 0 || m == NULL) { \
2479 if ((iface) != bifp) \
2480 ETHER_BPF_MTAP(iface, m); \
2484 /* We just received a packet that we sent out. */ \
2485 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2486 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2493 * Unicast. Make sure it's not for the bridge.
2495 do { GRAB_OUR_PACKETS(bifp) } while (0);
2498 * Give a chance for ifp at first priority. This will help when the
2499 * packet comes through the interface like VLAN's with the same MACs
2500 * on several interfaces from the same bridge. This also will save
2501 * some CPU cycles in case the destination interface and the input
2502 * interface (eq ifp) are the same.
2504 do { GRAB_OUR_PACKETS(ifp) } while (0);
2506 /* Now check the all bridge members. */
2507 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2508 GRAB_OUR_PACKETS(bif2->bif_ifp)
2511 #undef OR_CARP_CHECK_WE_ARE_DST
2512 #undef OR_CARP_CHECK_WE_ARE_SRC
2513 #undef OR_PFIL_HOOKED_INET6
2514 #undef GRAB_OUR_PACKETS
2516 /* Perform the bridge forwarding function. */
2517 bridge_forward(sc, bif, m);
2525 * Send a frame to all interfaces that are members of
2526 * the bridge, except for the one on which the packet
2529 * NOTE: Releases the lock on return.
2532 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2533 struct mbuf *m, int runfilt)
2535 struct bridge_iflist *dbif, *sbif;
2537 struct ifnet *dst_if;
2542 sbif = bridge_lookup_member_if(sc, src_if);
2544 /* Filter on the bridge interface before broadcasting */
2545 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2547 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2550 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2556 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2557 dst_if = dbif->bif_ifp;
2558 if (dst_if == src_if)
2561 /* Private segments can not talk to each other */
2562 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2565 if ((dbif->bif_flags & IFBIF_STP) &&
2566 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2569 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2570 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2573 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2576 if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2580 mc = m_dup(m, M_NOWAIT);
2582 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2588 * Filter on the output interface. Pass a NULL bridge interface
2589 * pointer so we do not redundantly filter on the bridge for
2590 * each interface we broadcast on.
2592 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2594 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2598 /* Keep the layer3 header aligned */
2599 i = min(mc->m_pkthdr.len, max_protohdr);
2600 mc = m_copyup(mc, i, ETHER_ALIGN);
2602 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2606 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2612 bridge_enqueue(sc, dst_if, mc);
2621 * Duplicate a packet out one or more interfaces that are in span mode,
2622 * the original mbuf is unmodified.
2625 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2627 struct bridge_iflist *bif;
2628 struct ifnet *dst_if;
2633 if (CK_LIST_EMPTY(&sc->sc_spanlist))
2636 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2637 dst_if = bif->bif_ifp;
2639 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2642 mc = m_copypacket(m, M_NOWAIT);
2644 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2648 bridge_enqueue(sc, dst_if, mc);
2655 * Add a bridge routing entry.
2658 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2659 struct bridge_iflist *bif, int setflags, uint8_t flags)
2661 struct bridge_rtnode *brt;
2665 BRIDGE_UNLOCK_ASSERT(sc);
2667 /* Check the source address is valid and not multicast. */
2668 if (ETHER_IS_MULTICAST(dst) ||
2669 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2670 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2673 /* 802.1p frames map to vlan 1 */
2678 * A route for this destination might already exist. If so,
2679 * update it, otherwise create a new one.
2681 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2684 /* Check again, now that we have the lock. There could have
2685 * been a race and we only want to insert this once. */
2686 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) {
2691 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2692 sc->sc_brtexceeded++;
2696 /* Check per interface address limits (if enabled) */
2697 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2698 bif->bif_addrexceeded++;
2704 * Allocate a new bridge forwarding node, and
2705 * initialize the expiration time and Ethernet
2708 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2713 brt->brt_vnet = curvnet;
2715 if (bif->bif_flags & IFBIF_STICKY)
2716 brt->brt_flags = IFBAF_STICKY;
2718 brt->brt_flags = IFBAF_DYNAMIC;
2720 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2721 brt->brt_vlan = vlan;
2723 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2724 uma_zfree(V_bridge_rtnode_zone, brt);
2734 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2735 brt->brt_dst != bif) {
2737 brt->brt_dst->bif_addrcnt--;
2739 brt->brt_dst->bif_addrcnt++;
2743 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2744 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2746 brt->brt_flags = flags;
2754 * Lookup the destination interface for an address.
2756 static struct ifnet *
2757 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2759 struct bridge_rtnode *brt;
2763 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2766 return (brt->brt_ifp);
2772 * Trim the routine table so that we have a number
2773 * of routing entries less than or equal to the
2777 bridge_rttrim(struct bridge_softc *sc)
2779 struct bridge_rtnode *brt, *nbrt;
2782 BRIDGE_LOCK_ASSERT(sc);
2784 /* Make sure we actually need to do this. */
2785 if (sc->sc_brtcnt <= sc->sc_brtmax)
2788 /* Force an aging cycle; this might trim enough addresses. */
2790 if (sc->sc_brtcnt <= sc->sc_brtmax)
2793 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2794 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2795 bridge_rtnode_destroy(sc, brt);
2796 if (sc->sc_brtcnt <= sc->sc_brtmax)
2805 * Aging timer for the bridge.
2808 bridge_timer(void *arg)
2810 struct bridge_softc *sc = arg;
2811 struct epoch_tracker et;
2813 NET_EPOCH_ENTER(et);
2814 BRIDGE_LOCK_ASSERT(sc);
2816 /* Destruction of rtnodes requires a proper vnet context */
2817 CURVNET_SET(sc->sc_ifp->if_vnet);
2820 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2821 callout_reset(&sc->sc_brcallout,
2822 bridge_rtable_prune_period * hz, bridge_timer, sc);
2830 * Perform an aging cycle.
2833 bridge_rtage(struct bridge_softc *sc)
2835 struct bridge_rtnode *brt, *nbrt;
2838 BRIDGE_LOCK_ASSERT(sc);
2840 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2841 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2842 if (time_uptime >= brt->brt_expire)
2843 bridge_rtnode_destroy(sc, brt);
2851 * Remove all dynamic addresses from the bridge.
2854 bridge_rtflush(struct bridge_softc *sc, int full)
2856 struct bridge_rtnode *brt, *nbrt;
2859 BRIDGE_LOCK_ASSERT(sc);
2861 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2862 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2863 bridge_rtnode_destroy(sc, brt);
2870 * Remove an address from the table.
2873 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2875 struct bridge_rtnode *brt;
2879 BRIDGE_LOCK_ASSERT(sc);
2882 * If vlan is zero then we want to delete for all vlans so the lookup
2883 * may return more than one.
2885 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2886 bridge_rtnode_destroy(sc, brt);
2890 return (found ? 0 : ENOENT);
2896 * Delete routes to a speicifc member interface.
2899 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2901 struct bridge_rtnode *brt, *nbrt;
2904 BRIDGE_LOCK_ASSERT(sc);
2906 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2907 if (brt->brt_ifp == ifp && (full ||
2908 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2909 bridge_rtnode_destroy(sc, brt);
2914 * bridge_rtable_init:
2916 * Initialize the route table for this bridge.
2919 bridge_rtable_init(struct bridge_softc *sc)
2923 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2924 M_DEVBUF, M_WAITOK);
2926 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2927 CK_LIST_INIT(&sc->sc_rthash[i]);
2929 sc->sc_rthash_key = arc4random();
2930 CK_LIST_INIT(&sc->sc_rtlist);
2934 * bridge_rtable_fini:
2936 * Deconstruct the route table for this bridge.
2939 bridge_rtable_fini(struct bridge_softc *sc)
2942 KASSERT(sc->sc_brtcnt == 0,
2943 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2944 free(sc->sc_rthash, M_DEVBUF);
2948 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2949 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2951 #define mix(a, b, c) \
2953 a -= b; a -= c; a ^= (c >> 13); \
2954 b -= c; b -= a; b ^= (a << 8); \
2955 c -= a; c -= b; c ^= (b >> 13); \
2956 a -= b; a -= c; a ^= (c >> 12); \
2957 b -= c; b -= a; b ^= (a << 16); \
2958 c -= a; c -= b; c ^= (b >> 5); \
2959 a -= b; a -= c; a ^= (c >> 3); \
2960 b -= c; b -= a; b ^= (a << 10); \
2961 c -= a; c -= b; c ^= (b >> 15); \
2962 } while (/*CONSTCOND*/0)
2964 static __inline uint32_t
2965 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2967 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2978 return (c & BRIDGE_RTHASH_MASK);
2984 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2988 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2989 d = ((int)a[i]) - ((int)b[i]);
2996 * bridge_rtnode_lookup:
2998 * Look up a bridge route node for the specified destination. Compare the
2999 * vlan id or if zero then just return the first match.
3001 static struct bridge_rtnode *
3002 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3004 struct bridge_rtnode *brt;
3010 hash = bridge_rthash(sc, addr);
3011 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3012 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3013 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3023 * bridge_rtnode_insert:
3025 * Insert the specified bridge node into the route table. We
3026 * assume the entry is not already in the table.
3029 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3031 struct bridge_rtnode *lbrt;
3035 BRIDGE_LOCK_ASSERT(sc);
3037 hash = bridge_rthash(sc, brt->brt_addr);
3039 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3041 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3046 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3047 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3050 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3053 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3054 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3057 lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3058 } while (lbrt != NULL);
3061 panic("bridge_rtnode_insert: impossible");
3065 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3072 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3074 struct bridge_rtnode *brt;
3076 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3078 CURVNET_SET(brt->brt_vnet);
3079 uma_zfree(V_bridge_rtnode_zone, brt);
3084 * bridge_rtnode_destroy:
3086 * Destroy a bridge rtnode.
3089 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3092 BRIDGE_LOCK_ASSERT(sc);
3094 CK_LIST_REMOVE(brt, brt_hash);
3096 CK_LIST_REMOVE(brt, brt_list);
3098 brt->brt_dst->bif_addrcnt--;
3100 NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx);
3104 * bridge_rtable_expire:
3106 * Set the expiry time for all routes on an interface.
3109 bridge_rtable_expire(struct ifnet *ifp, int age)
3111 struct bridge_softc *sc = ifp->if_bridge;
3112 struct bridge_rtnode *brt;
3113 struct epoch_tracker et;
3115 NET_EPOCH_ENTER(et);
3116 CURVNET_SET(ifp->if_vnet);
3120 * If the age is zero then flush, otherwise set all the expiry times to
3121 * age for the interface
3124 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3126 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3127 /* Cap the expiry time to 'age' */
3128 if (brt->brt_ifp == ifp &&
3129 brt->brt_expire > time_uptime + age &&
3130 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3131 brt->brt_expire = time_uptime + age;
3140 * bridge_state_change:
3142 * Callback from the bridgestp code when a port changes states.
3145 bridge_state_change(struct ifnet *ifp, int state)
3147 struct bridge_softc *sc = ifp->if_bridge;
3148 static const char *stpstates[] = {
3157 CURVNET_SET(ifp->if_vnet);
3159 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3160 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3165 * Send bridge packets through pfil if they are one of the types pfil can deal
3166 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3167 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3171 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3173 int snap, error, i, hlen;
3174 struct ether_header *eh1, eh2;
3177 u_int16_t ether_type;
3181 error = -1; /* Default error if not error == 0 */
3184 /* we may return with the IP fields swapped, ensure its not shared */
3185 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3188 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3189 return (0); /* filtering is disabled */
3191 i = min((*mp)->m_pkthdr.len, max_protohdr);
3192 if ((*mp)->m_len < i) {
3193 *mp = m_pullup(*mp, i);
3195 printf("%s: m_pullup failed\n", __func__);
3200 eh1 = mtod(*mp, struct ether_header *);
3201 ether_type = ntohs(eh1->ether_type);
3204 * Check for SNAP/LLC.
3206 if (ether_type < ETHERMTU) {
3207 struct llc *llc2 = (struct llc *)(eh1 + 1);
3209 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3210 llc2->llc_dsap == LLC_SNAP_LSAP &&
3211 llc2->llc_ssap == LLC_SNAP_LSAP &&
3212 llc2->llc_control == LLC_UI) {
3213 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3219 * If we're trying to filter bridge traffic, don't look at anything
3220 * other than IP and ARP traffic. If the filter doesn't understand
3221 * IPv6, don't allow IPv6 through the bridge either. This is lame
3222 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3223 * but of course we don't have an AppleTalk filter to begin with.
3224 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3227 switch (ether_type) {
3229 case ETHERTYPE_REVARP:
3230 if (V_pfil_ipfw_arp == 0)
3231 return (0); /* Automatically pass */
3236 case ETHERTYPE_IPV6:
3241 * Check to see if the user wants to pass non-ip
3242 * packets, these will not be checked by pfil(9) and
3243 * passed unconditionally so the default is to drop.
3249 /* Run the packet through pfil before stripping link headers */
3250 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3251 dir == PFIL_OUT && ifp != NULL) {
3252 switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) {
3260 /* Strip off the Ethernet header and keep a copy. */
3261 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3262 m_adj(*mp, ETHER_HDR_LEN);
3264 /* Strip off snap header, if present */
3266 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3267 m_adj(*mp, sizeof(struct llc));
3271 * Check the IP header for alignment and errors
3273 if (dir == PFIL_IN) {
3274 switch (ether_type) {
3276 error = bridge_ip_checkbasic(mp);
3279 case ETHERTYPE_IPV6:
3280 error = bridge_ip6_checkbasic(mp);
3293 * Run the packet through pfil
3296 switch (ether_type) {
3299 * Run pfil on the member interface and the bridge, both can
3300 * be skipped by clearing pfil_member or pfil_bridge.
3303 * in_if -> bridge_if -> out_if
3305 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3306 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3310 if (V_pfil_member && ifp != NULL && (rv =
3311 pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) !=
3315 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3316 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3320 /* check if we need to fragment the packet */
3321 /* bridge_fragment generates a mbuf chain of packets */
3322 /* that already include eth headers */
3323 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3324 i = (*mp)->m_pkthdr.len;
3325 if (i > ifp->if_mtu) {
3326 error = bridge_fragment(ifp, mp, &eh2, snap,
3332 /* Recalculate the ip checksum. */
3333 ip = mtod(*mp, struct ip *);
3334 hlen = ip->ip_hl << 2;
3335 if (hlen < sizeof(struct ip))
3337 if (hlen > (*mp)->m_len) {
3338 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3340 ip = mtod(*mp, struct ip *);
3345 if (hlen == sizeof(struct ip))
3346 ip->ip_sum = in_cksum_hdr(ip);
3348 ip->ip_sum = in_cksum(*mp, hlen);
3352 case ETHERTYPE_IPV6:
3353 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3354 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3358 if (V_pfil_member && ifp != NULL && (rv =
3359 pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) !=
3363 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3364 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3383 * Finally, put everything back the way it was and return
3386 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3389 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3392 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3395 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3406 * Perform basic checks on header size since
3407 * pfil assumes ip_input has already processed
3408 * it for it. Cut-and-pasted from ip_input.c.
3409 * Given how simple the IPv6 version is,
3410 * does the IPv4 version really need to be
3413 * XXX Should we update ipstat here, or not?
3414 * XXX Right now we update ipstat but not
3418 bridge_ip_checkbasic(struct mbuf **mp)
3420 struct mbuf *m = *mp;
3428 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3429 if ((m = m_copyup(m, sizeof(struct ip),
3430 (max_linkhdr + 3) & ~3)) == NULL) {
3431 /* XXXJRT new stat, please */
3432 KMOD_IPSTAT_INC(ips_toosmall);
3435 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3436 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3437 KMOD_IPSTAT_INC(ips_toosmall);
3441 ip = mtod(m, struct ip *);
3442 if (ip == NULL) goto bad;
3444 if (ip->ip_v != IPVERSION) {
3445 KMOD_IPSTAT_INC(ips_badvers);
3448 hlen = ip->ip_hl << 2;
3449 if (hlen < sizeof(struct ip)) { /* minimum header length */
3450 KMOD_IPSTAT_INC(ips_badhlen);
3453 if (hlen > m->m_len) {
3454 if ((m = m_pullup(m, hlen)) == NULL) {
3455 KMOD_IPSTAT_INC(ips_badhlen);
3458 ip = mtod(m, struct ip *);
3459 if (ip == NULL) goto bad;
3462 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3463 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3465 if (hlen == sizeof(struct ip)) {
3466 sum = in_cksum_hdr(ip);
3468 sum = in_cksum(m, hlen);
3472 KMOD_IPSTAT_INC(ips_badsum);
3476 /* Retrieve the packet length. */
3477 len = ntohs(ip->ip_len);
3480 * Check for additional length bogosity
3483 KMOD_IPSTAT_INC(ips_badlen);
3488 * Check that the amount of data in the buffers
3489 * is as at least much as the IP header would have us expect.
3490 * Drop packet if shorter than we expect.
3492 if (m->m_pkthdr.len < len) {
3493 KMOD_IPSTAT_INC(ips_tooshort);
3497 /* Checks out, proceed */
3508 * Same as above, but for IPv6.
3509 * Cut-and-pasted from ip6_input.c.
3510 * XXX Should we update ip6stat, or not?
3513 bridge_ip6_checkbasic(struct mbuf **mp)
3515 struct mbuf *m = *mp;
3516 struct ip6_hdr *ip6;
3519 * If the IPv6 header is not aligned, slurp it up into a new
3520 * mbuf with space for link headers, in the event we forward
3521 * it. Otherwise, if it is aligned, make sure the entire base
3522 * IPv6 header is in the first mbuf of the chain.
3524 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3525 struct ifnet *inifp = m->m_pkthdr.rcvif;
3526 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3527 (max_linkhdr + 3) & ~3)) == NULL) {
3528 /* XXXJRT new stat, please */
3529 IP6STAT_INC(ip6s_toosmall);
3530 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3533 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3534 struct ifnet *inifp = m->m_pkthdr.rcvif;
3535 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3536 IP6STAT_INC(ip6s_toosmall);
3537 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3542 ip6 = mtod(m, struct ip6_hdr *);
3544 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3545 IP6STAT_INC(ip6s_badvers);
3546 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3550 /* Checks out, proceed */
3563 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3566 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3567 int snap, struct llc *llc)
3569 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3573 if (m->m_len < sizeof(struct ip) &&
3574 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3576 ip = mtod(m, struct ip *);
3578 m->m_pkthdr.csum_flags |= CSUM_IP;
3579 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3584 * Walk the chain and re-add the Ethernet header for
3587 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3588 nextpkt = mcur->m_nextpkt;
3589 mcur->m_nextpkt = NULL;
3591 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3595 mprev->m_nextpkt = nextpkt;
3598 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3601 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3605 mprev->m_nextpkt = nextpkt;
3608 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3611 * The previous two M_PREPEND could have inserted one or two
3612 * mbufs in front so we have to update the previous packet's
3615 mcur->m_nextpkt = nextpkt;
3617 mprev->m_nextpkt = mcur;
3619 /* The first mbuf in the original chain needs to be
3626 KMOD_IPSTAT_INC(ips_fragmented);
3630 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3631 m = mcur->m_nextpkt;
3638 bridge_linkstate(struct ifnet *ifp)
3640 struct bridge_softc *sc = ifp->if_bridge;
3641 struct bridge_iflist *bif;
3642 struct epoch_tracker et;
3644 NET_EPOCH_ENTER(et);
3646 bif = bridge_lookup_member_if(sc, ifp);
3651 bridge_linkcheck(sc);
3653 bstp_linkstate(&bif->bif_stp);
3659 bridge_linkcheck(struct bridge_softc *sc)
3661 struct bridge_iflist *bif;
3662 int new_link, hasls;
3666 new_link = LINK_STATE_DOWN;
3668 /* Our link is considered up if at least one of our ports is active */
3669 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3670 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3672 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3673 new_link = LINK_STATE_UP;
3677 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3678 /* If no interfaces support link-state then we default to up */
3679 new_link = LINK_STATE_UP;
3681 if_link_state_change(sc->sc_ifp, new_link);