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); \
192 cv_init(&(_sc)->sc_cv, "if_bridge_cv"); \
194 #define BRIDGE_LOCK_DESTROY(_sc) do { \
195 mtx_destroy(&(_sc)->sc_mtx); \
196 cv_destroy(&(_sc)->sc_cv); \
198 #define BRIDGE_LOCK(_sc) mtx_lock(&(_sc)->sc_mtx)
199 #define BRIDGE_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_mtx)
200 #define BRIDGE_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_mtx, MA_OWNED)
201 #define BRIDGE_UNLOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_mtx, MA_NOTOWNED)
202 #define BRIDGE_LOCK2REF(_sc, _err) do { \
203 mtx_assert(&(_sc)->sc_mtx, MA_OWNED); \
204 if ((_sc)->sc_iflist_xcnt > 0) \
207 (_sc)->sc_iflist_ref++; \
208 mtx_unlock(&(_sc)->sc_mtx); \
210 #define BRIDGE_UNREF(_sc) do { \
211 mtx_lock(&(_sc)->sc_mtx); \
212 (_sc)->sc_iflist_ref--; \
213 if (((_sc)->sc_iflist_xcnt > 0) && ((_sc)->sc_iflist_ref == 0)) \
214 cv_broadcast(&(_sc)->sc_cv); \
215 mtx_unlock(&(_sc)->sc_mtx); \
217 #define BRIDGE_XLOCK(_sc) do { \
218 mtx_assert(&(_sc)->sc_mtx, MA_OWNED); \
219 (_sc)->sc_iflist_xcnt++; \
220 while ((_sc)->sc_iflist_ref > 0) \
221 cv_wait(&(_sc)->sc_cv, &(_sc)->sc_mtx); \
223 #define BRIDGE_XDROP(_sc) do { \
224 mtx_assert(&(_sc)->sc_mtx, MA_OWNED); \
225 (_sc)->sc_iflist_xcnt--; \
229 * Bridge interface list entry.
231 struct bridge_iflist {
232 CK_LIST_ENTRY(bridge_iflist) bif_next;
233 struct ifnet *bif_ifp; /* member if */
234 struct bstp_port bif_stp; /* STP state */
235 uint32_t bif_flags; /* member if flags */
236 int bif_savedcaps; /* saved capabilities */
237 uint32_t bif_addrmax; /* max # of addresses */
238 uint32_t bif_addrcnt; /* cur. # of addresses */
239 uint32_t bif_addrexceeded;/* # of address violations */
241 struct epoch_context bif_epoch_ctx;
247 struct bridge_rtnode {
248 CK_LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
249 CK_LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
250 struct bridge_iflist *brt_dst; /* destination if */
251 unsigned long brt_expire; /* expiration time */
252 uint8_t brt_flags; /* address flags */
253 uint8_t brt_addr[ETHER_ADDR_LEN];
254 uint16_t brt_vlan; /* vlan id */
256 struct vnet *brt_vnet;
257 struct epoch_context brt_epoch_ctx;
259 #define brt_ifp brt_dst->bif_ifp
262 * Software state for each bridge.
264 struct bridge_softc {
265 struct ifnet *sc_ifp; /* make this an interface */
266 LIST_ENTRY(bridge_softc) sc_list;
269 uint32_t sc_brtmax; /* max # of addresses */
270 uint32_t sc_brtcnt; /* cur. # of addresses */
271 uint32_t sc_brttimeout; /* rt timeout in seconds */
272 struct callout sc_brcallout; /* bridge callout */
273 uint32_t sc_iflist_ref; /* refcount for sc_iflist */
274 uint32_t sc_iflist_xcnt; /* refcount for sc_iflist */
275 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
276 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
277 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
278 uint32_t sc_rthash_key; /* key for hash */
279 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
280 struct bstp_state sc_stp; /* STP state */
281 uint32_t sc_brtexceeded; /* # of cache drops */
282 struct ifnet *sc_ifaddr; /* member mac copied from */
283 struct ether_addr sc_defaddr; /* Default MAC address */
285 struct epoch_context sc_epoch_ctx;
288 VNET_DEFINE_STATIC(struct mtx, bridge_list_mtx);
289 #define V_bridge_list_mtx VNET(bridge_list_mtx)
290 static eventhandler_tag bridge_detach_cookie;
292 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
294 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
295 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone)
297 static int bridge_clone_create(struct if_clone *, int, caddr_t);
298 static void bridge_clone_destroy(struct ifnet *);
300 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
301 static void bridge_mutecaps(struct bridge_softc *);
302 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
304 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
305 static void bridge_init(void *);
306 static void bridge_dummynet(struct mbuf *, struct ifnet *);
307 static void bridge_stop(struct ifnet *, int);
308 static int bridge_transmit(struct ifnet *, struct mbuf *);
309 static void bridge_qflush(struct ifnet *);
310 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
311 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
313 static int bridge_enqueue(struct bridge_softc *, struct ifnet *,
315 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
317 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
320 static void bridge_timer(void *);
322 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
324 static void bridge_span(struct bridge_softc *, struct mbuf *);
326 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
327 uint16_t, struct bridge_iflist *, int, uint8_t);
328 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
330 static void bridge_rttrim(struct bridge_softc *);
331 static void bridge_rtage(struct bridge_softc *);
332 static void bridge_rtflush(struct bridge_softc *, int);
333 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
336 static void bridge_rtable_init(struct bridge_softc *);
337 static void bridge_rtable_fini(struct bridge_softc *);
339 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
340 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
341 const uint8_t *, uint16_t);
342 static int bridge_rtnode_insert(struct bridge_softc *,
343 struct bridge_rtnode *);
344 static void bridge_rtnode_destroy(struct bridge_softc *,
345 struct bridge_rtnode *);
346 static void bridge_rtable_expire(struct ifnet *, int);
347 static void bridge_state_change(struct ifnet *, int);
349 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
351 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
353 static void bridge_delete_member(struct bridge_softc *,
354 struct bridge_iflist *, int);
355 static void bridge_delete_span(struct bridge_softc *,
356 struct bridge_iflist *);
358 static int bridge_ioctl_add(struct bridge_softc *, void *);
359 static int bridge_ioctl_del(struct bridge_softc *, void *);
360 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
361 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
362 static int bridge_ioctl_scache(struct bridge_softc *, void *);
363 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
364 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
365 static int bridge_ioctl_rts(struct bridge_softc *, void *);
366 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
367 static int bridge_ioctl_sto(struct bridge_softc *, void *);
368 static int bridge_ioctl_gto(struct bridge_softc *, void *);
369 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
370 static int bridge_ioctl_flush(struct bridge_softc *, void *);
371 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
372 static int bridge_ioctl_spri(struct bridge_softc *, void *);
373 static int bridge_ioctl_ght(struct bridge_softc *, void *);
374 static int bridge_ioctl_sht(struct bridge_softc *, void *);
375 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
376 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
377 static int bridge_ioctl_gma(struct bridge_softc *, void *);
378 static int bridge_ioctl_sma(struct bridge_softc *, void *);
379 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
380 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
381 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
382 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
383 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
384 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
385 static int bridge_ioctl_grte(struct bridge_softc *, void *);
386 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
387 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
388 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
389 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
391 static int bridge_ip_checkbasic(struct mbuf **mp);
393 static int bridge_ip6_checkbasic(struct mbuf **mp);
395 static int bridge_fragment(struct ifnet *, struct mbuf **mp,
396 struct ether_header *, int, struct llc *);
397 static void bridge_linkstate(struct ifnet *ifp);
398 static void bridge_linkcheck(struct bridge_softc *sc);
401 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
402 #define VLANTAGOF(_m) \
403 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
405 static struct bstp_cb_ops bridge_ops = {
406 .bcb_state = bridge_state_change,
407 .bcb_rtage = bridge_rtable_expire
410 SYSCTL_DECL(_net_link);
411 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
413 /* only pass IP[46] packets when pfil is enabled */
414 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
415 #define V_pfil_onlyip VNET(pfil_onlyip)
416 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
417 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
418 "Only pass IP packets when pfil is enabled");
420 /* run pfil hooks on the bridge interface */
421 VNET_DEFINE_STATIC(int, pfil_bridge) = 1;
422 #define V_pfil_bridge VNET(pfil_bridge)
423 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
424 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
425 "Packet filter on the bridge interface");
427 /* layer2 filter with ipfw */
428 VNET_DEFINE_STATIC(int, pfil_ipfw);
429 #define V_pfil_ipfw VNET(pfil_ipfw)
431 /* layer2 ARP filter with ipfw */
432 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
433 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
434 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
435 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
436 "Filter ARP packets through IPFW layer2");
438 /* run pfil hooks on the member interface */
439 VNET_DEFINE_STATIC(int, pfil_member) = 1;
440 #define V_pfil_member VNET(pfil_member)
441 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
442 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
443 "Packet filter on the member interface");
445 /* run pfil hooks on the physical interface for locally destined packets */
446 VNET_DEFINE_STATIC(int, pfil_local_phys);
447 #define V_pfil_local_phys VNET(pfil_local_phys)
448 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
449 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
450 "Packet filter on the physical interface for locally destined packets");
452 /* log STP state changes */
453 VNET_DEFINE_STATIC(int, log_stp);
454 #define V_log_stp VNET(log_stp)
455 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
456 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
457 "Log STP state changes");
459 /* share MAC with first bridge member */
460 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
461 #define V_bridge_inherit_mac VNET(bridge_inherit_mac)
462 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
463 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
464 "Inherit MAC address from the first bridge member");
466 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
467 #define V_allow_llz_overlap VNET(allow_llz_overlap)
468 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
469 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
470 "Allow overlap of link-local scope "
471 "zones of a bridge interface and the member interfaces");
473 struct bridge_control {
474 int (*bc_func)(struct bridge_softc *, void *);
479 #define BC_F_COPYIN 0x01 /* copy arguments in */
480 #define BC_F_COPYOUT 0x02 /* copy arguments out */
481 #define BC_F_SUSER 0x04 /* do super-user check */
483 const struct bridge_control bridge_control_table[] = {
484 { bridge_ioctl_add, sizeof(struct ifbreq),
485 BC_F_COPYIN|BC_F_SUSER },
486 { bridge_ioctl_del, sizeof(struct ifbreq),
487 BC_F_COPYIN|BC_F_SUSER },
489 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
490 BC_F_COPYIN|BC_F_COPYOUT },
491 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
492 BC_F_COPYIN|BC_F_SUSER },
494 { bridge_ioctl_scache, sizeof(struct ifbrparam),
495 BC_F_COPYIN|BC_F_SUSER },
496 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
499 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
500 BC_F_COPYIN|BC_F_COPYOUT },
501 { bridge_ioctl_rts, sizeof(struct ifbaconf),
502 BC_F_COPYIN|BC_F_COPYOUT },
504 { bridge_ioctl_saddr, sizeof(struct ifbareq),
505 BC_F_COPYIN|BC_F_SUSER },
507 { bridge_ioctl_sto, sizeof(struct ifbrparam),
508 BC_F_COPYIN|BC_F_SUSER },
509 { bridge_ioctl_gto, sizeof(struct ifbrparam),
512 { bridge_ioctl_daddr, sizeof(struct ifbareq),
513 BC_F_COPYIN|BC_F_SUSER },
515 { bridge_ioctl_flush, sizeof(struct ifbreq),
516 BC_F_COPYIN|BC_F_SUSER },
518 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
520 { bridge_ioctl_spri, sizeof(struct ifbrparam),
521 BC_F_COPYIN|BC_F_SUSER },
523 { bridge_ioctl_ght, sizeof(struct ifbrparam),
525 { bridge_ioctl_sht, sizeof(struct ifbrparam),
526 BC_F_COPYIN|BC_F_SUSER },
528 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
530 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
531 BC_F_COPYIN|BC_F_SUSER },
533 { bridge_ioctl_gma, sizeof(struct ifbrparam),
535 { bridge_ioctl_sma, sizeof(struct ifbrparam),
536 BC_F_COPYIN|BC_F_SUSER },
538 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
539 BC_F_COPYIN|BC_F_SUSER },
541 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
542 BC_F_COPYIN|BC_F_SUSER },
544 { bridge_ioctl_addspan, sizeof(struct ifbreq),
545 BC_F_COPYIN|BC_F_SUSER },
546 { bridge_ioctl_delspan, sizeof(struct ifbreq),
547 BC_F_COPYIN|BC_F_SUSER },
549 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
552 { bridge_ioctl_grte, sizeof(struct ifbrparam),
555 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
556 BC_F_COPYIN|BC_F_COPYOUT },
558 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
559 BC_F_COPYIN|BC_F_SUSER },
561 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
562 BC_F_COPYIN|BC_F_SUSER },
564 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
565 BC_F_COPYIN|BC_F_SUSER },
568 const int bridge_control_table_size = nitems(bridge_control_table);
570 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
571 #define V_bridge_list VNET(bridge_list)
572 #define BRIDGE_LIST_LOCK_INIT(x) mtx_init(&V_bridge_list_mtx, \
573 "if_bridge list", NULL, MTX_DEF)
574 #define BRIDGE_LIST_LOCK_DESTROY(x) mtx_destroy(&V_bridge_list_mtx)
575 #define BRIDGE_LIST_LOCK(x) mtx_lock(&V_bridge_list_mtx)
576 #define BRIDGE_LIST_UNLOCK(x) mtx_unlock(&V_bridge_list_mtx)
578 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
579 #define V_bridge_cloner VNET(bridge_cloner)
581 static const char bridge_name[] = "bridge";
584 vnet_bridge_init(const void *unused __unused)
587 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
588 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
590 BRIDGE_LIST_LOCK_INIT();
591 LIST_INIT(&V_bridge_list);
592 V_bridge_cloner = if_clone_simple(bridge_name,
593 bridge_clone_create, bridge_clone_destroy, 0);
595 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
596 vnet_bridge_init, NULL);
599 vnet_bridge_uninit(const void *unused __unused)
602 if_clone_detach(V_bridge_cloner);
603 V_bridge_cloner = NULL;
604 BRIDGE_LIST_LOCK_DESTROY();
606 /* Before we can destroy the uma zone, because there are callbacks that
610 uma_zdestroy(V_bridge_rtnode_zone);
612 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
613 vnet_bridge_uninit, NULL);
616 bridge_modevent(module_t mod, int type, void *data)
621 bridge_dn_p = bridge_dummynet;
622 bridge_detach_cookie = EVENTHANDLER_REGISTER(
623 ifnet_departure_event, bridge_ifdetach, NULL,
624 EVENTHANDLER_PRI_ANY);
627 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
628 bridge_detach_cookie);
637 static moduledata_t bridge_mod = {
643 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
644 MODULE_VERSION(if_bridge, 1);
645 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
648 * handler for net.link.bridge.ipfw
651 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
653 int enable = V_pfil_ipfw;
656 error = sysctl_handle_int(oidp, &enable, 0, req);
659 if (enable != V_pfil_ipfw) {
660 V_pfil_ipfw = enable;
663 * Disable pfil so that ipfw doesnt run twice, if the user
664 * really wants both then they can re-enable pfil_bridge and/or
665 * pfil_member. Also allow non-ip packets as ipfw can filter by
677 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
678 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET,
679 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
680 "Layer2 filter with IPFW");
683 * bridge_clone_create:
685 * Create a new bridge instance.
688 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
690 struct bridge_softc *sc, *sc2;
691 struct ifnet *bifp, *ifp;
693 unsigned long hostid;
695 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
696 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
702 BRIDGE_LOCK_INIT(sc);
703 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
704 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
706 /* Initialize our routing table. */
707 bridge_rtable_init(sc);
709 callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
711 CK_LIST_INIT(&sc->sc_iflist);
712 CK_LIST_INIT(&sc->sc_spanlist);
715 if_initname(ifp, bridge_name, unit);
716 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
717 ifp->if_ioctl = bridge_ioctl;
718 ifp->if_transmit = bridge_transmit;
719 ifp->if_qflush = bridge_qflush;
720 ifp->if_init = bridge_init;
721 ifp->if_type = IFT_BRIDGE;
724 * Generate an ethernet address with a locally administered address.
726 * Since we are using random ethernet addresses for the bridge, it is
727 * possible that we might have address collisions, so make sure that
728 * this hardware address isn't already in use on another bridge.
729 * The first try uses the hostid and falls back to arc4rand().
732 getcredhostid(curthread->td_ucred, &hostid);
734 if (fb || hostid == 0) {
735 ether_gen_addr(ifp, &sc->sc_defaddr);
737 sc->sc_defaddr.octet[0] = 0x2;
738 sc->sc_defaddr.octet[1] = (hostid >> 24) & 0xff;
739 sc->sc_defaddr.octet[2] = (hostid >> 16) & 0xff;
740 sc->sc_defaddr.octet[3] = (hostid >> 8 ) & 0xff;
741 sc->sc_defaddr.octet[4] = hostid & 0xff;
742 sc->sc_defaddr.octet[5] = ifp->if_dunit & 0xff;
748 LIST_FOREACH(sc2, &V_bridge_list, sc_list) {
750 if (memcmp(sc->sc_defaddr.octet,
751 IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
756 BRIDGE_LIST_UNLOCK();
757 } while (retry == 1);
759 bstp_attach(&sc->sc_stp, &bridge_ops);
760 ether_ifattach(ifp, sc->sc_defaddr.octet);
761 /* Now undo some of the damage... */
762 ifp->if_baudrate = 0;
763 ifp->if_type = IFT_BRIDGE;
766 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
767 BRIDGE_LIST_UNLOCK();
773 bridge_clone_destroy_cb(struct epoch_context *ctx)
775 struct bridge_softc *sc;
777 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
779 BRIDGE_LOCK_DESTROY(sc);
784 * bridge_clone_destroy:
786 * Destroy a bridge instance.
789 bridge_clone_destroy(struct ifnet *ifp)
791 struct bridge_softc *sc = ifp->if_softc;
792 struct bridge_iflist *bif;
797 ifp->if_flags &= ~IFF_UP;
799 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
800 bridge_delete_member(sc, bif, 0);
802 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
803 bridge_delete_span(sc, bif);
806 /* Tear down the routing table. */
807 bridge_rtable_fini(sc);
811 callout_drain(&sc->sc_brcallout);
814 LIST_REMOVE(sc, sc_list);
815 BRIDGE_LIST_UNLOCK();
817 bstp_detach(&sc->sc_stp);
821 epoch_call(net_epoch_preempt, &sc->sc_epoch_ctx, bridge_clone_destroy_cb);
827 * Handle a control request from the operator.
830 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
832 struct bridge_softc *sc = ifp->if_softc;
833 struct ifreq *ifr = (struct ifreq *)data;
834 struct bridge_iflist *bif;
835 struct thread *td = curthread;
837 struct ifbreq ifbreq;
838 struct ifbifconf ifbifconf;
839 struct ifbareq ifbareq;
840 struct ifbaconf ifbaconf;
841 struct ifbrparam ifbrparam;
842 struct ifbropreq ifbropreq;
844 struct ifdrv *ifd = (struct ifdrv *) data;
845 const struct bridge_control *bc;
846 int error = 0, oldmtu;
847 struct epoch_tracker et;
849 NET_EPOCH_ENTER_ET(et);
859 if (ifd->ifd_cmd >= bridge_control_table_size) {
863 bc = &bridge_control_table[ifd->ifd_cmd];
865 if (cmd == SIOCGDRVSPEC &&
866 (bc->bc_flags & BC_F_COPYOUT) == 0) {
870 else if (cmd == SIOCSDRVSPEC &&
871 (bc->bc_flags & BC_F_COPYOUT) != 0) {
876 if (bc->bc_flags & BC_F_SUSER) {
877 error = priv_check(td, PRIV_NET_BRIDGE);
882 if (ifd->ifd_len != bc->bc_argsize ||
883 ifd->ifd_len > sizeof(args)) {
888 bzero(&args, sizeof(args));
889 if (bc->bc_flags & BC_F_COPYIN) {
890 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
895 oldmtu = ifp->if_mtu;
897 error = (*bc->bc_func)(sc, &args);
903 * Bridge MTU may change during addition of the first port.
904 * If it did, do network layer specific procedure.
906 if (ifp->if_mtu != oldmtu) {
913 if (bc->bc_flags & BC_F_COPYOUT)
914 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
919 if (!(ifp->if_flags & IFF_UP) &&
920 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
922 * If interface is marked down and it is running,
923 * then stop and disable it.
928 } else if ((ifp->if_flags & IFF_UP) &&
929 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
931 * If interface is marked up and it is stopped, then
939 if (ifr->ifr_mtu < 576) {
943 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
944 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
948 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
949 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
950 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
951 " != %d\n", sc->sc_ifp->if_xname,
952 bif->bif_ifp->if_mtu,
953 bif->bif_ifp->if_xname, ifr->ifr_mtu);
959 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
964 * drop the lock as ether_ioctl() will call bridge_start() and
965 * cause the lock to be recursed.
967 error = ether_ioctl(ifp, cmd, data);
971 NET_EPOCH_EXIT_ET(et);
979 * Clear or restore unwanted capabilities on the member interface
982 bridge_mutecaps(struct bridge_softc *sc)
984 struct bridge_iflist *bif;
987 BRIDGE_LOCK_ASSERT(sc);
989 /* Initial bitmask of capabilities to test */
990 mask = BRIDGE_IFCAPS_MASK;
992 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
993 /* Every member must support it or its disabled */
994 mask &= bif->bif_savedcaps;
998 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
999 enabled = bif->bif_ifp->if_capenable;
1000 enabled &= ~BRIDGE_IFCAPS_STRIP;
1001 /* strip off mask bits and enable them again if allowed */
1002 enabled &= ~BRIDGE_IFCAPS_MASK;
1005 bridge_set_ifcap(sc, bif, enabled);
1013 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1015 struct ifnet *ifp = bif->bif_ifp;
1017 int error, mask, stuck;
1019 BRIDGE_UNLOCK_ASSERT(sc);
1021 bzero(&ifr, sizeof(ifr));
1022 ifr.ifr_reqcap = set;
1024 if (ifp->if_capenable != set) {
1025 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1027 if_printf(sc->sc_ifp,
1028 "error setting capabilities on %s: %d\n",
1029 ifp->if_xname, error);
1030 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1031 stuck = ifp->if_capenable & mask & ~set;
1033 if_printf(sc->sc_ifp,
1034 "can't disable some capabilities on %s: 0x%x\n",
1035 ifp->if_xname, stuck);
1040 * bridge_lookup_member:
1042 * Lookup a bridge member interface.
1044 static struct bridge_iflist *
1045 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1047 struct bridge_iflist *bif;
1050 MPASS(in_epoch(net_epoch_preempt));
1052 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1054 if (strcmp(ifp->if_xname, name) == 0)
1062 * bridge_lookup_member_if:
1064 * Lookup a bridge member interface by ifnet*.
1066 static struct bridge_iflist *
1067 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1069 struct bridge_iflist *bif;
1071 MPASS(in_epoch(net_epoch_preempt));
1073 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1074 if (bif->bif_ifp == member_ifp)
1082 bridge_delete_member_cb(struct epoch_context *ctx)
1084 struct bridge_iflist *bif;
1086 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1088 free(bif, M_DEVBUF);
1092 * bridge_delete_member:
1094 * Delete the specified member interface.
1097 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1100 struct ifnet *ifs = bif->bif_ifp;
1101 struct ifnet *fif = NULL;
1102 struct bridge_iflist *bifl;
1104 BRIDGE_LOCK_ASSERT(sc);
1106 if (bif->bif_flags & IFBIF_STP)
1107 bstp_disable(&bif->bif_stp);
1109 ifs->if_bridge = NULL;
1111 CK_LIST_REMOVE(bif, bif_next);
1115 * If removing the interface that gave the bridge its mac address, set
1116 * the mac address of the bridge to the address of the next member, or
1117 * to its default address if no members are left.
1119 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1120 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1121 bcopy(&sc->sc_defaddr,
1122 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1123 sc->sc_ifaddr = NULL;
1125 bifl = CK_LIST_FIRST(&sc->sc_iflist);
1126 fif = bifl->bif_ifp;
1127 bcopy(IF_LLADDR(fif),
1128 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1129 sc->sc_ifaddr = fif;
1131 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1134 bridge_linkcheck(sc);
1135 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1136 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1137 KASSERT(bif->bif_addrcnt == 0,
1138 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1140 ifs->if_bridge_output = NULL;
1141 ifs->if_bridge_input = NULL;
1142 ifs->if_bridge_linkstate = NULL;
1145 switch (ifs->if_type) {
1149 * Take the interface out of promiscuous mode, but only
1150 * if it was promiscuous in the first place. It might
1151 * not be if we're in the bridge_ioctl_add() error path.
1153 if (ifs->if_flags & IFF_PROMISC)
1154 (void) ifpromisc(ifs, 0);
1162 panic("bridge_delete_member: impossible");
1166 /* reneable any interface capabilities */
1167 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1169 bstp_destroy(&bif->bif_stp); /* prepare to free */
1172 epoch_call(net_epoch_preempt, &bif->bif_epoch_ctx,
1173 bridge_delete_member_cb);
1177 * bridge_delete_span:
1179 * Delete the specified span interface.
1182 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1184 BRIDGE_LOCK_ASSERT(sc);
1186 KASSERT(bif->bif_ifp->if_bridge == NULL,
1187 ("%s: not a span interface", __func__));
1189 CK_LIST_REMOVE(bif, bif_next);
1191 epoch_call(net_epoch_preempt, &bif->bif_epoch_ctx,
1192 bridge_delete_member_cb);
1196 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1198 struct ifbreq *req = arg;
1199 struct bridge_iflist *bif = NULL;
1203 ifs = ifunit(req->ifbr_ifsname);
1206 if (ifs->if_ioctl == NULL) /* must be supported */
1209 /* If it's in the span list, it can't be a member. */
1210 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1211 if (ifs == bif->bif_ifp)
1214 if (ifs->if_bridge == sc)
1217 if (ifs->if_bridge != NULL)
1220 switch (ifs->if_type) {
1224 /* permitted interface types */
1232 * Two valid inet6 addresses with link-local scope must not be
1233 * on the parent interface and the member interfaces at the
1234 * same time. This restriction is needed to prevent violation
1235 * of link-local scope zone. Attempts to add a member
1236 * interface which has inet6 addresses when the parent has
1237 * inet6 triggers removal of all inet6 addresses on the member
1241 /* Check if the parent interface has a link-local scope addr. */
1242 if (V_allow_llz_overlap == 0 &&
1243 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1245 * If any, remove all inet6 addresses from the member
1249 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1250 if (in6ifa_llaonifp(bif->bif_ifp)) {
1252 in6_ifdetach(bif->bif_ifp);
1254 if_printf(sc->sc_ifp,
1255 "IPv6 addresses on %s have been removed "
1256 "before adding it as a member to prevent "
1257 "IPv6 address scope violation.\n",
1258 bif->bif_ifp->if_xname);
1262 if (in6ifa_llaonifp(ifs)) {
1266 if_printf(sc->sc_ifp,
1267 "IPv6 addresses on %s have been removed "
1268 "before adding it as a member to prevent "
1269 "IPv6 address scope violation.\n",
1274 /* Allow the first Ethernet member to define the MTU */
1275 if (CK_LIST_EMPTY(&sc->sc_iflist))
1276 sc->sc_ifp->if_mtu = ifs->if_mtu;
1277 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1278 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1279 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1283 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1288 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1289 bif->bif_savedcaps = ifs->if_capenable;
1292 * Assign the interface's MAC address to the bridge if it's the first
1293 * member and the MAC address of the bridge has not been changed from
1294 * the default randomly generated one.
1296 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1297 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1298 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1299 sc->sc_ifaddr = ifs;
1300 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1303 ifs->if_bridge = sc;
1304 ifs->if_bridge_output = bridge_output;
1305 ifs->if_bridge_input = bridge_input;
1306 ifs->if_bridge_linkstate = bridge_linkstate;
1307 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1309 * XXX: XLOCK HERE!?!
1311 * NOTE: insert_***HEAD*** should be safe for the traversals.
1313 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1315 /* Set interface capabilities to the intersection set of all members */
1316 bridge_mutecaps(sc);
1317 bridge_linkcheck(sc);
1319 /* Place the interface into promiscuous mode */
1320 switch (ifs->if_type) {
1324 error = ifpromisc(ifs, 1);
1330 bridge_delete_member(sc, bif, 0);
1335 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1337 struct ifbreq *req = arg;
1338 struct bridge_iflist *bif;
1340 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1344 bridge_delete_member(sc, bif, 0);
1350 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1352 struct ifbreq *req = arg;
1353 struct bridge_iflist *bif;
1354 struct bstp_port *bp;
1356 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1361 req->ifbr_ifsflags = bif->bif_flags;
1362 req->ifbr_state = bp->bp_state;
1363 req->ifbr_priority = bp->bp_priority;
1364 req->ifbr_path_cost = bp->bp_path_cost;
1365 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1366 req->ifbr_proto = bp->bp_protover;
1367 req->ifbr_role = bp->bp_role;
1368 req->ifbr_stpflags = bp->bp_flags;
1369 req->ifbr_addrcnt = bif->bif_addrcnt;
1370 req->ifbr_addrmax = bif->bif_addrmax;
1371 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1373 /* Copy STP state options as flags */
1374 if (bp->bp_operedge)
1375 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1376 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1377 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1378 if (bp->bp_ptp_link)
1379 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1380 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1381 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1382 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1383 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1384 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1385 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1390 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1392 struct ifbreq *req = arg;
1393 struct bridge_iflist *bif;
1394 struct bstp_port *bp;
1397 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1402 if (req->ifbr_ifsflags & IFBIF_SPAN)
1403 /* SPAN is readonly */
1406 if (req->ifbr_ifsflags & IFBIF_STP) {
1407 if ((bif->bif_flags & IFBIF_STP) == 0) {
1408 error = bstp_enable(&bif->bif_stp);
1413 if ((bif->bif_flags & IFBIF_STP) != 0)
1414 bstp_disable(&bif->bif_stp);
1417 /* Pass on STP flags */
1418 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1419 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1420 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1421 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1423 /* Save the bits relating to the bridge */
1424 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1430 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1432 struct ifbrparam *param = arg;
1434 sc->sc_brtmax = param->ifbrp_csize;
1441 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1443 struct ifbrparam *param = arg;
1445 param->ifbrp_csize = sc->sc_brtmax;
1451 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1453 struct ifbifconf *bifc = arg;
1454 struct bridge_iflist *bif;
1457 int count, buflen, len, error = 0;
1460 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1462 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1465 buflen = sizeof(breq) * count;
1466 if (bifc->ifbic_len == 0) {
1467 bifc->ifbic_len = buflen;
1471 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1476 len = min(bifc->ifbic_len, buflen);
1477 bzero(&breq, sizeof(breq));
1478 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1479 if (len < sizeof(breq))
1482 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1483 sizeof(breq.ifbr_ifsname));
1484 /* Fill in the ifbreq structure */
1485 error = bridge_ioctl_gifflags(sc, &breq);
1488 memcpy(buf, &breq, sizeof(breq));
1490 buf += sizeof(breq);
1491 len -= sizeof(breq);
1493 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1494 if (len < sizeof(breq))
1497 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1498 sizeof(breq.ifbr_ifsname));
1499 breq.ifbr_ifsflags = bif->bif_flags;
1500 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1501 memcpy(buf, &breq, sizeof(breq));
1503 buf += sizeof(breq);
1504 len -= sizeof(breq);
1508 bifc->ifbic_len = sizeof(breq) * count;
1509 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1511 free(outbuf, M_TEMP);
1516 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1518 struct ifbaconf *bac = arg;
1519 struct bridge_rtnode *brt;
1520 struct ifbareq bareq;
1522 int count, buflen, len, error = 0;
1524 if (bac->ifbac_len == 0)
1528 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1530 buflen = sizeof(bareq) * count;
1533 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1538 len = min(bac->ifbac_len, buflen);
1539 bzero(&bareq, sizeof(bareq));
1540 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1541 if (len < sizeof(bareq))
1543 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1544 sizeof(bareq.ifba_ifsname));
1545 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1546 bareq.ifba_vlan = brt->brt_vlan;
1547 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1548 time_uptime < brt->brt_expire)
1549 bareq.ifba_expire = brt->brt_expire - time_uptime;
1551 bareq.ifba_expire = 0;
1552 bareq.ifba_flags = brt->brt_flags;
1554 memcpy(buf, &bareq, sizeof(bareq));
1556 buf += sizeof(bareq);
1557 len -= sizeof(bareq);
1561 bac->ifbac_len = sizeof(bareq) * count;
1562 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1564 free(outbuf, M_TEMP);
1569 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1571 struct ifbareq *req = arg;
1572 struct bridge_iflist *bif;
1575 MPASS(in_epoch(net_epoch_preempt));
1577 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1581 /* bridge_rtupdate() may acquire the lock. */
1583 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1591 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1593 struct ifbrparam *param = arg;
1595 sc->sc_brttimeout = param->ifbrp_ctime;
1600 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1602 struct ifbrparam *param = arg;
1604 param->ifbrp_ctime = sc->sc_brttimeout;
1609 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1611 struct ifbareq *req = arg;
1613 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1617 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1619 struct ifbreq *req = arg;
1621 bridge_rtflush(sc, req->ifbr_ifsflags);
1626 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1628 struct ifbrparam *param = arg;
1629 struct bstp_state *bs = &sc->sc_stp;
1631 param->ifbrp_prio = bs->bs_bridge_priority;
1636 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1638 struct ifbrparam *param = arg;
1640 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1644 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1646 struct ifbrparam *param = arg;
1647 struct bstp_state *bs = &sc->sc_stp;
1649 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1654 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1656 struct ifbrparam *param = arg;
1658 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1662 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1664 struct ifbrparam *param = arg;
1665 struct bstp_state *bs = &sc->sc_stp;
1667 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1672 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1674 struct ifbrparam *param = arg;
1676 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1680 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1682 struct ifbrparam *param = arg;
1683 struct bstp_state *bs = &sc->sc_stp;
1685 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1690 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1692 struct ifbrparam *param = arg;
1694 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1698 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1700 struct ifbreq *req = arg;
1701 struct bridge_iflist *bif;
1703 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1707 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1711 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1713 struct ifbreq *req = arg;
1714 struct bridge_iflist *bif;
1716 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1720 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1724 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1726 struct ifbreq *req = arg;
1727 struct bridge_iflist *bif;
1729 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1733 bif->bif_addrmax = req->ifbr_addrmax;
1738 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1740 struct ifbreq *req = arg;
1741 struct bridge_iflist *bif = NULL;
1744 ifs = ifunit(req->ifbr_ifsname);
1748 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1749 if (ifs == bif->bif_ifp)
1752 if (ifs->if_bridge != NULL)
1755 switch (ifs->if_type) {
1764 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1769 bif->bif_flags = IFBIF_SPAN;
1771 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1777 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1779 struct ifbreq *req = arg;
1780 struct bridge_iflist *bif;
1783 ifs = ifunit(req->ifbr_ifsname);
1787 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1788 if (ifs == bif->bif_ifp)
1794 bridge_delete_span(sc, bif);
1800 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1802 struct ifbropreq *req = arg;
1803 struct bstp_state *bs = &sc->sc_stp;
1804 struct bstp_port *root_port;
1806 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1807 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1808 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1810 root_port = bs->bs_root_port;
1811 if (root_port == NULL)
1812 req->ifbop_root_port = 0;
1814 req->ifbop_root_port = root_port->bp_ifp->if_index;
1816 req->ifbop_holdcount = bs->bs_txholdcount;
1817 req->ifbop_priority = bs->bs_bridge_priority;
1818 req->ifbop_protocol = bs->bs_protover;
1819 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1820 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1821 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1822 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1823 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1824 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1830 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1832 struct ifbrparam *param = arg;
1834 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1839 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1841 struct ifbpstpconf *bifstp = arg;
1842 struct bridge_iflist *bif;
1843 struct bstp_port *bp;
1844 struct ifbpstpreq bpreq;
1846 int count, buflen, len, error = 0;
1849 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1850 if ((bif->bif_flags & IFBIF_STP) != 0)
1854 buflen = sizeof(bpreq) * count;
1855 if (bifstp->ifbpstp_len == 0) {
1856 bifstp->ifbpstp_len = buflen;
1861 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1866 len = min(bifstp->ifbpstp_len, buflen);
1867 bzero(&bpreq, sizeof(bpreq));
1868 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1869 if (len < sizeof(bpreq))
1872 if ((bif->bif_flags & IFBIF_STP) == 0)
1876 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1877 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1878 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1879 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1880 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1881 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1883 memcpy(buf, &bpreq, sizeof(bpreq));
1885 buf += sizeof(bpreq);
1886 len -= sizeof(bpreq);
1890 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1891 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1893 free(outbuf, M_TEMP);
1898 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1900 struct ifbrparam *param = arg;
1902 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1906 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1908 struct ifbrparam *param = arg;
1910 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1916 * Detach an interface from a bridge. Called when a member
1917 * interface is detaching.
1920 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1922 struct bridge_softc *sc = ifp->if_bridge;
1923 struct bridge_iflist *bif;
1924 struct epoch_tracker et;
1926 if (ifp->if_flags & IFF_RENAMING)
1928 if (V_bridge_cloner == NULL) {
1930 * This detach handler can be called after
1931 * vnet_bridge_uninit(). Just return in that case.
1935 NET_EPOCH_ENTER_ET(et);
1936 /* Check if the interface is a bridge member */
1940 bif = bridge_lookup_member_if(sc, ifp);
1942 bridge_delete_member(sc, bif, 1);
1945 NET_EPOCH_EXIT_ET(et);
1949 /* Check if the interface is a span port */
1951 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1953 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1954 if (ifp == bif->bif_ifp) {
1955 bridge_delete_span(sc, bif);
1961 BRIDGE_LIST_UNLOCK();
1962 NET_EPOCH_EXIT_ET(et);
1968 * Initialize a bridge interface.
1971 bridge_init(void *xsc)
1973 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1974 struct ifnet *ifp = sc->sc_ifp;
1976 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1980 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1983 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1984 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1992 * Stop the bridge interface.
1995 bridge_stop(struct ifnet *ifp, int disable)
1997 struct bridge_softc *sc = ifp->if_softc;
1999 BRIDGE_LOCK_ASSERT(sc);
2001 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2004 callout_stop(&sc->sc_brcallout);
2005 bstp_stop(&sc->sc_stp);
2007 bridge_rtflush(sc, IFBF_FLUSHDYN);
2009 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
2015 * Enqueue a packet on a bridge member interface.
2019 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
2025 /* We may be sending a fragment so traverse the mbuf */
2028 m->m_nextpkt = NULL;
2029 len = m->m_pkthdr.len;
2030 mflags = m->m_flags;
2033 * If underlying interface can not do VLAN tag insertion itself
2034 * then attach a packet tag that holds it.
2036 if ((m->m_flags & M_VLANTAG) &&
2037 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
2038 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
2041 "unable to prepend VLAN header\n");
2042 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2045 m->m_flags &= ~M_VLANTAG;
2048 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2049 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2051 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2055 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2056 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2057 if (mflags & M_MCAST)
2058 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2067 * Receive a queued packet from dummynet and pass it on to the output
2070 * The mbuf has the Ethernet header already attached.
2073 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2075 struct bridge_softc *sc;
2077 sc = ifp->if_bridge;
2080 * The packet didnt originate from a member interface. This should only
2081 * ever happen if a member interface is removed while packets are
2089 if (PFIL_HOOKED(&V_inet_pfil_hook)
2091 || PFIL_HOOKED(&V_inet6_pfil_hook)
2094 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2100 bridge_enqueue(sc, ifp, m);
2106 * Send output from a bridge member interface. This
2107 * performs the bridging function for locally originated
2110 * The mbuf has the Ethernet header already attached. We must
2111 * enqueue or free the mbuf before returning.
2114 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2117 struct ether_header *eh;
2118 struct ifnet *dst_if;
2119 struct bridge_softc *sc;
2122 MPASS(in_epoch(net_epoch_preempt));
2124 if (m->m_len < ETHER_HDR_LEN) {
2125 m = m_pullup(m, ETHER_HDR_LEN);
2130 eh = mtod(m, struct ether_header *);
2131 sc = ifp->if_bridge;
2132 vlan = VLANTAGOF(m);
2135 * If bridge is down, but the original output interface is up,
2136 * go ahead and send out that interface. Otherwise, the packet
2139 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2145 * If the packet is a multicast, or we don't know a better way to
2146 * get there, send to all interfaces.
2148 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2151 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2152 if (dst_if == NULL) {
2153 struct bridge_iflist *bif;
2159 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2160 dst_if = bif->bif_ifp;
2162 if (dst_if->if_type == IFT_GIF)
2164 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2168 * If this is not the original output interface,
2169 * and the interface is participating in spanning
2170 * tree, make sure the port is in a state that
2171 * allows forwarding.
2173 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2174 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2177 if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2181 mc = m_copypacket(m, M_NOWAIT);
2183 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2188 bridge_enqueue(sc, dst_if, mc);
2197 * XXX Spanning tree consideration here?
2201 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2206 bridge_enqueue(sc, dst_if, m);
2213 * Do output on a bridge.
2217 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2219 struct bridge_softc *sc;
2220 struct ether_header *eh;
2221 struct ifnet *dst_if;
2226 ETHER_BPF_MTAP(ifp, m);
2228 eh = mtod(m, struct ether_header *);
2230 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2231 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2232 error = bridge_enqueue(sc, dst_if, m);
2234 bridge_broadcast(sc, ifp, m, 0);
2240 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2243 bridge_qflush(struct ifnet *ifp __unused)
2250 * The forwarding function of the bridge.
2252 * NOTE: Releases the lock on return.
2255 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2258 struct bridge_iflist *dbif;
2259 struct ifnet *src_if, *dst_if, *ifp;
2260 struct ether_header *eh;
2265 MPASS(in_epoch(net_epoch_preempt));
2267 src_if = m->m_pkthdr.rcvif;
2270 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2271 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2272 vlan = VLANTAGOF(m);
2274 if ((sbif->bif_flags & IFBIF_STP) &&
2275 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2278 eh = mtod(m, struct ether_header *);
2279 dst = eh->ether_dhost;
2281 /* If the interface is learning, record the address. */
2282 if (sbif->bif_flags & IFBIF_LEARNING) {
2283 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2284 sbif, 0, IFBAF_DYNAMIC);
2286 * If the interface has addresses limits then deny any source
2287 * that is not in the cache.
2289 if (error && sbif->bif_addrmax)
2293 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2294 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2298 * At this point, the port either doesn't participate
2299 * in spanning tree or it is in the forwarding state.
2303 * If the packet is unicast, destined for someone on
2304 * "this" side of the bridge, drop it.
2306 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2307 dst_if = bridge_rtlookup(sc, dst, vlan);
2308 if (src_if == dst_if)
2312 * Check if its a reserved multicast address, any address
2313 * listed in 802.1D section 7.12.6 may not be forwarded by the
2315 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2317 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2318 dst[2] == 0xc2 && dst[3] == 0x00 &&
2319 dst[4] == 0x00 && dst[5] <= 0x0f)
2322 /* ...forward it to all interfaces. */
2323 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2328 * If we have a destination interface which is a member of our bridge,
2329 * OR this is a unicast packet, push it through the bpf(4) machinery.
2330 * For broadcast or multicast packets, don't bother because it will
2331 * be reinjected into ether_input. We do this before we pass the packets
2332 * through the pfil(9) framework, as it is possible that pfil(9) will
2333 * drop the packet, or possibly modify it, making it difficult to debug
2334 * firewall issues on the bridge.
2336 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2337 ETHER_BPF_MTAP(ifp, m);
2339 /* run the packet filter */
2340 if (PFIL_HOOKED(&V_inet_pfil_hook)
2342 || PFIL_HOOKED(&V_inet6_pfil_hook)
2345 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2351 if (dst_if == NULL) {
2352 bridge_broadcast(sc, src_if, m, 1);
2357 * At this point, we're dealing with a unicast frame
2358 * going to a different interface.
2360 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2363 dbif = bridge_lookup_member_if(sc, dst_if);
2365 /* Not a member of the bridge (anymore?) */
2368 /* Private segments can not talk to each other */
2369 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2372 if ((dbif->bif_flags & IFBIF_STP) &&
2373 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2376 if (PFIL_HOOKED(&V_inet_pfil_hook)
2378 || PFIL_HOOKED(&V_inet6_pfil_hook)
2381 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2387 bridge_enqueue(sc, dst_if, m);
2397 * Receive input from a member interface. Queue the packet for
2398 * bridging if it is not for us.
2400 static struct mbuf *
2401 bridge_input(struct ifnet *ifp, struct mbuf *m)
2403 struct bridge_softc *sc = ifp->if_bridge;
2404 struct bridge_iflist *bif, *bif2;
2406 struct ether_header *eh;
2407 struct mbuf *mc, *mc2;
2411 MPASS(in_epoch(net_epoch_preempt));
2413 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2417 vlan = VLANTAGOF(m);
2420 * Implement support for bridge monitoring. If this flag has been
2421 * set on this interface, discard the packet once we push it through
2422 * the bpf(4) machinery, but before we do, increment the byte and
2423 * packet counters associated with this interface.
2425 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2426 m->m_pkthdr.rcvif = bifp;
2427 ETHER_BPF_MTAP(bifp, m);
2428 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2429 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2433 bif = bridge_lookup_member_if(sc, ifp);
2438 eh = mtod(m, struct ether_header *);
2442 if (m->m_flags & (M_BCAST|M_MCAST)) {
2443 /* Tap off 802.1D packets; they do not get forwarded. */
2444 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2445 ETHER_ADDR_LEN) == 0) {
2446 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2450 if ((bif->bif_flags & IFBIF_STP) &&
2451 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2456 * Make a deep copy of the packet and enqueue the copy
2457 * for bridge processing; return the original packet for
2460 mc = m_dup(m, M_NOWAIT);
2465 /* Perform the bridge forwarding function with the copy. */
2466 bridge_forward(sc, bif, mc);
2469 * Reinject the mbuf as arriving on the bridge so we have a
2470 * chance at claiming multicast packets. We can not loop back
2471 * here from ether_input as a bridge is never a member of a
2474 KASSERT(bifp->if_bridge == NULL,
2475 ("loop created in bridge_input"));
2476 mc2 = m_dup(m, M_NOWAIT);
2478 /* Keep the layer3 header aligned */
2479 int i = min(mc2->m_pkthdr.len, max_protohdr);
2480 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2483 mc2->m_pkthdr.rcvif = bifp;
2484 (*bifp->if_input)(bifp, mc2);
2487 /* Return the original packet for local processing. */
2491 if ((bif->bif_flags & IFBIF_STP) &&
2492 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2496 #if (defined(INET) || defined(INET6))
2497 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2498 || ((iface)->if_carp \
2499 && (*carp_forus_p)((iface), eh->ether_dhost))
2500 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2501 || ((iface)->if_carp \
2502 && (*carp_forus_p)((iface), eh->ether_shost))
2504 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2505 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2509 # define OR_PFIL_HOOKED_INET6 \
2510 || PFIL_HOOKED(&V_inet6_pfil_hook)
2512 # define OR_PFIL_HOOKED_INET6
2515 #define GRAB_OUR_PACKETS(iface) \
2516 if ((iface)->if_type == IFT_GIF) \
2518 /* It is destined for us. */ \
2519 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2520 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2522 if ((iface)->if_type == IFT_BRIDGE) { \
2523 ETHER_BPF_MTAP(iface, m); \
2524 if_inc_counter(iface, IFCOUNTER_IPACKETS, 1); \
2525 if_inc_counter(iface, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2526 /* Filter on the physical interface. */ \
2527 if (V_pfil_local_phys && \
2528 (PFIL_HOOKED(&V_inet_pfil_hook) \
2529 OR_PFIL_HOOKED_INET6)) { \
2530 if (bridge_pfil(&m, NULL, ifp, \
2531 PFIL_IN) != 0 || m == NULL) { \
2534 eh = mtod(m, struct ether_header *); \
2537 if (bif->bif_flags & IFBIF_LEARNING) { \
2538 error = bridge_rtupdate(sc, eh->ether_shost, \
2539 vlan, bif, 0, IFBAF_DYNAMIC); \
2540 if (error && bif->bif_addrmax) { \
2545 m->m_pkthdr.rcvif = iface; \
2549 /* We just received a packet that we sent out. */ \
2550 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2551 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2558 * Unicast. Make sure it's not for the bridge.
2560 do { GRAB_OUR_PACKETS(bifp) } while (0);
2563 * Give a chance for ifp at first priority. This will help when the
2564 * packet comes through the interface like VLAN's with the same MACs
2565 * on several interfaces from the same bridge. This also will save
2566 * some CPU cycles in case the destination interface and the input
2567 * interface (eq ifp) are the same.
2569 do { GRAB_OUR_PACKETS(ifp) } while (0);
2571 /* Now check the all bridge members. */
2572 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2573 GRAB_OUR_PACKETS(bif2->bif_ifp)
2576 #undef OR_CARP_CHECK_WE_ARE_DST
2577 #undef OR_CARP_CHECK_WE_ARE_SRC
2578 #undef OR_PFIL_HOOKED_INET6
2579 #undef GRAB_OUR_PACKETS
2581 /* Perform the bridge forwarding function. */
2582 bridge_forward(sc, bif, m);
2590 * Send a frame to all interfaces that are members of
2591 * the bridge, except for the one on which the packet
2594 * NOTE: Releases the lock on return.
2597 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2598 struct mbuf *m, int runfilt)
2600 struct bridge_iflist *dbif, *sbif;
2602 struct ifnet *dst_if;
2605 MPASS(in_epoch(net_epoch_preempt));
2607 sbif = bridge_lookup_member_if(sc, src_if);
2609 /* Filter on the bridge interface before broadcasting */
2610 if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2612 || PFIL_HOOKED(&V_inet6_pfil_hook)
2615 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2621 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2622 dst_if = dbif->bif_ifp;
2623 if (dst_if == src_if)
2626 /* Private segments can not talk to each other */
2627 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2630 if ((dbif->bif_flags & IFBIF_STP) &&
2631 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2634 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2635 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2638 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2641 if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2645 mc = m_dup(m, M_NOWAIT);
2647 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2653 * Filter on the output interface. Pass a NULL bridge interface
2654 * pointer so we do not redundantly filter on the bridge for
2655 * each interface we broadcast on.
2657 if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2659 || PFIL_HOOKED(&V_inet6_pfil_hook)
2663 /* Keep the layer3 header aligned */
2664 i = min(mc->m_pkthdr.len, max_protohdr);
2665 mc = m_copyup(mc, i, ETHER_ALIGN);
2667 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2671 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2677 bridge_enqueue(sc, dst_if, mc);
2686 * Duplicate a packet out one or more interfaces that are in span mode,
2687 * the original mbuf is unmodified.
2690 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2692 struct bridge_iflist *bif;
2693 struct ifnet *dst_if;
2696 MPASS(in_epoch(net_epoch_preempt));
2698 if (CK_LIST_EMPTY(&sc->sc_spanlist))
2701 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2702 dst_if = bif->bif_ifp;
2704 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2707 mc = m_copypacket(m, M_NOWAIT);
2709 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2713 bridge_enqueue(sc, dst_if, mc);
2720 * Add a bridge routing entry.
2723 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2724 struct bridge_iflist *bif, int setflags, uint8_t flags)
2726 struct bridge_rtnode *brt;
2729 MPASS(in_epoch(net_epoch_preempt));
2730 BRIDGE_UNLOCK_ASSERT(sc);
2732 /* Check the source address is valid and not multicast. */
2733 if (ETHER_IS_MULTICAST(dst) ||
2734 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2735 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2738 /* 802.1p frames map to vlan 1 */
2743 * A route for this destination might already exist. If so,
2744 * update it, otherwise create a new one.
2746 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2749 /* Check again, now that we have the lock. There could have
2750 * been a race and we only want to insert this once. */
2751 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) {
2756 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2757 sc->sc_brtexceeded++;
2761 /* Check per interface address limits (if enabled) */
2762 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2763 bif->bif_addrexceeded++;
2769 * Allocate a new bridge forwarding node, and
2770 * initialize the expiration time and Ethernet
2773 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2778 brt->brt_vnet = curvnet;
2780 if (bif->bif_flags & IFBIF_STICKY)
2781 brt->brt_flags = IFBAF_STICKY;
2783 brt->brt_flags = IFBAF_DYNAMIC;
2785 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2786 brt->brt_vlan = vlan;
2788 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2789 uma_zfree(V_bridge_rtnode_zone, brt);
2799 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2800 brt->brt_dst != bif) {
2802 brt->brt_dst->bif_addrcnt--;
2804 brt->brt_dst->bif_addrcnt++;
2808 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2809 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2811 brt->brt_flags = flags;
2819 * Lookup the destination interface for an address.
2821 static struct ifnet *
2822 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2824 struct bridge_rtnode *brt;
2826 MPASS(in_epoch(net_epoch_preempt));
2828 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2831 return (brt->brt_ifp);
2837 * Trim the routine table so that we have a number
2838 * of routing entries less than or equal to the
2842 bridge_rttrim(struct bridge_softc *sc)
2844 struct bridge_rtnode *brt, *nbrt;
2846 BRIDGE_LOCK_ASSERT(sc);
2848 /* Make sure we actually need to do this. */
2849 if (sc->sc_brtcnt <= sc->sc_brtmax)
2852 /* Force an aging cycle; this might trim enough addresses. */
2854 if (sc->sc_brtcnt <= sc->sc_brtmax)
2857 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2858 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2859 bridge_rtnode_destroy(sc, brt);
2860 if (sc->sc_brtcnt <= sc->sc_brtmax)
2869 * Aging timer for the bridge.
2872 bridge_timer(void *arg)
2874 struct bridge_softc *sc = arg;
2876 BRIDGE_LOCK_ASSERT(sc);
2878 /* Destruction of rtnodes requires a proper vnet context */
2879 CURVNET_SET(sc->sc_ifp->if_vnet);
2882 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2883 callout_reset(&sc->sc_brcallout,
2884 bridge_rtable_prune_period * hz, bridge_timer, sc);
2891 * Perform an aging cycle.
2894 bridge_rtage(struct bridge_softc *sc)
2896 struct bridge_rtnode *brt, *nbrt;
2898 BRIDGE_LOCK_ASSERT(sc);
2900 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2901 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2902 if (time_uptime >= brt->brt_expire)
2903 bridge_rtnode_destroy(sc, brt);
2911 * Remove all dynamic addresses from the bridge.
2914 bridge_rtflush(struct bridge_softc *sc, int full)
2916 struct bridge_rtnode *brt, *nbrt;
2918 BRIDGE_LOCK_ASSERT(sc);
2920 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2921 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2922 bridge_rtnode_destroy(sc, brt);
2929 * Remove an address from the table.
2932 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2934 struct bridge_rtnode *brt;
2937 BRIDGE_LOCK_ASSERT(sc);
2940 * If vlan is zero then we want to delete for all vlans so the lookup
2941 * may return more than one.
2943 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2944 bridge_rtnode_destroy(sc, brt);
2948 return (found ? 0 : ENOENT);
2954 * Delete routes to a speicifc member interface.
2957 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2959 struct bridge_rtnode *brt, *nbrt;
2961 BRIDGE_LOCK_ASSERT(sc);
2963 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2964 if (brt->brt_ifp == ifp && (full ||
2965 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2966 bridge_rtnode_destroy(sc, brt);
2971 * bridge_rtable_init:
2973 * Initialize the route table for this bridge.
2976 bridge_rtable_init(struct bridge_softc *sc)
2980 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2981 M_DEVBUF, M_WAITOK);
2983 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2984 CK_LIST_INIT(&sc->sc_rthash[i]);
2986 sc->sc_rthash_key = arc4random();
2987 CK_LIST_INIT(&sc->sc_rtlist);
2991 * bridge_rtable_fini:
2993 * Deconstruct the route table for this bridge.
2996 bridge_rtable_fini(struct bridge_softc *sc)
2999 KASSERT(sc->sc_brtcnt == 0,
3000 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3001 free(sc->sc_rthash, M_DEVBUF);
3005 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3006 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3008 #define mix(a, b, c) \
3010 a -= b; a -= c; a ^= (c >> 13); \
3011 b -= c; b -= a; b ^= (a << 8); \
3012 c -= a; c -= b; c ^= (b >> 13); \
3013 a -= b; a -= c; a ^= (c >> 12); \
3014 b -= c; b -= a; b ^= (a << 16); \
3015 c -= a; c -= b; c ^= (b >> 5); \
3016 a -= b; a -= c; a ^= (c >> 3); \
3017 b -= c; b -= a; b ^= (a << 10); \
3018 c -= a; c -= b; c ^= (b >> 15); \
3019 } while (/*CONSTCOND*/0)
3021 static __inline uint32_t
3022 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3024 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3035 return (c & BRIDGE_RTHASH_MASK);
3041 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3045 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3046 d = ((int)a[i]) - ((int)b[i]);
3053 * bridge_rtnode_lookup:
3055 * Look up a bridge route node for the specified destination. Compare the
3056 * vlan id or if zero then just return the first match.
3058 static struct bridge_rtnode *
3059 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3061 struct bridge_rtnode *brt;
3065 MPASS(in_epoch(net_epoch_preempt));
3067 hash = bridge_rthash(sc, addr);
3068 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3069 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3070 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3080 * bridge_rtnode_insert:
3082 * Insert the specified bridge node into the route table. We
3083 * assume the entry is not already in the table.
3086 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3088 struct bridge_rtnode *lbrt;
3092 BRIDGE_LOCK_ASSERT(sc);
3094 hash = bridge_rthash(sc, brt->brt_addr);
3096 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3098 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3103 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3104 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3107 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3110 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3111 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3114 lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3115 } while (lbrt != NULL);
3118 panic("bridge_rtnode_insert: impossible");
3122 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3129 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3131 struct bridge_rtnode *brt;
3133 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3135 CURVNET_SET(brt->brt_vnet);
3136 uma_zfree(V_bridge_rtnode_zone, brt);
3141 * bridge_rtnode_destroy:
3143 * Destroy a bridge rtnode.
3146 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3148 BRIDGE_LOCK_ASSERT(sc);
3150 CK_LIST_REMOVE(brt, brt_hash);
3152 CK_LIST_REMOVE(brt, brt_list);
3154 brt->brt_dst->bif_addrcnt--;
3156 epoch_call(net_epoch_preempt, &brt->brt_epoch_ctx,
3157 bridge_rtnode_destroy_cb);
3161 * bridge_rtable_expire:
3163 * Set the expiry time for all routes on an interface.
3166 bridge_rtable_expire(struct ifnet *ifp, int age)
3168 struct bridge_softc *sc = ifp->if_bridge;
3169 struct bridge_rtnode *brt;
3171 CURVNET_SET(ifp->if_vnet);
3175 * If the age is zero then flush, otherwise set all the expiry times to
3176 * age for the interface
3179 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3181 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3182 /* Cap the expiry time to 'age' */
3183 if (brt->brt_ifp == ifp &&
3184 brt->brt_expire > time_uptime + age &&
3185 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3186 brt->brt_expire = time_uptime + age;
3194 * bridge_state_change:
3196 * Callback from the bridgestp code when a port changes states.
3199 bridge_state_change(struct ifnet *ifp, int state)
3201 struct bridge_softc *sc = ifp->if_bridge;
3202 static const char *stpstates[] = {
3211 CURVNET_SET(ifp->if_vnet);
3213 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3214 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3219 * Send bridge packets through pfil if they are one of the types pfil can deal
3220 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3221 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3225 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3227 int snap, error, i, hlen;
3228 struct ether_header *eh1, eh2;
3231 u_int16_t ether_type;
3234 error = -1; /* Default error if not error == 0 */
3237 /* we may return with the IP fields swapped, ensure its not shared */
3238 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3241 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3242 return (0); /* filtering is disabled */
3244 i = min((*mp)->m_pkthdr.len, max_protohdr);
3245 if ((*mp)->m_len < i) {
3246 *mp = m_pullup(*mp, i);
3248 printf("%s: m_pullup failed\n", __func__);
3253 eh1 = mtod(*mp, struct ether_header *);
3254 ether_type = ntohs(eh1->ether_type);
3257 * Check for SNAP/LLC.
3259 if (ether_type < ETHERMTU) {
3260 struct llc *llc2 = (struct llc *)(eh1 + 1);
3262 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3263 llc2->llc_dsap == LLC_SNAP_LSAP &&
3264 llc2->llc_ssap == LLC_SNAP_LSAP &&
3265 llc2->llc_control == LLC_UI) {
3266 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3272 * If we're trying to filter bridge traffic, don't look at anything
3273 * other than IP and ARP traffic. If the filter doesn't understand
3274 * IPv6, don't allow IPv6 through the bridge either. This is lame
3275 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3276 * but of course we don't have an AppleTalk filter to begin with.
3277 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3280 switch (ether_type) {
3282 case ETHERTYPE_REVARP:
3283 if (V_pfil_ipfw_arp == 0)
3284 return (0); /* Automatically pass */
3289 case ETHERTYPE_IPV6:
3294 * Check to see if the user wants to pass non-ip
3295 * packets, these will not be checked by pfil(9) and
3296 * passed unconditionally so the default is to drop.
3302 /* Run the packet through pfil before stripping link headers */
3303 if (PFIL_HOOKED(&V_link_pfil_hook) && V_pfil_ipfw != 0 &&
3304 dir == PFIL_OUT && ifp != NULL) {
3306 error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, 0,
3309 if (*mp == NULL || error != 0) /* packet consumed by filter */
3313 /* Strip off the Ethernet header and keep a copy. */
3314 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3315 m_adj(*mp, ETHER_HDR_LEN);
3317 /* Strip off snap header, if present */
3319 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3320 m_adj(*mp, sizeof(struct llc));
3324 * Check the IP header for alignment and errors
3326 if (dir == PFIL_IN) {
3327 switch (ether_type) {
3329 error = bridge_ip_checkbasic(mp);
3332 case ETHERTYPE_IPV6:
3333 error = bridge_ip6_checkbasic(mp);
3346 * Run the packet through pfil
3348 switch (ether_type) {
3351 * Run pfil on the member interface and the bridge, both can
3352 * be skipped by clearing pfil_member or pfil_bridge.
3355 * in_if -> bridge_if -> out_if
3357 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3358 error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3361 if (*mp == NULL || error != 0) /* filter may consume */
3364 if (V_pfil_member && ifp != NULL)
3365 error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3368 if (*mp == NULL || error != 0) /* filter may consume */
3371 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3372 error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3375 if (*mp == NULL || error != 0) /* filter may consume */
3378 /* check if we need to fragment the packet */
3379 /* bridge_fragment generates a mbuf chain of packets */
3380 /* that already include eth headers */
3381 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3382 i = (*mp)->m_pkthdr.len;
3383 if (i > ifp->if_mtu) {
3384 error = bridge_fragment(ifp, mp, &eh2, snap,
3390 /* Recalculate the ip checksum. */
3391 ip = mtod(*mp, struct ip *);
3392 hlen = ip->ip_hl << 2;
3393 if (hlen < sizeof(struct ip))
3395 if (hlen > (*mp)->m_len) {
3396 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3398 ip = mtod(*mp, struct ip *);
3403 if (hlen == sizeof(struct ip))
3404 ip->ip_sum = in_cksum_hdr(ip);
3406 ip->ip_sum = in_cksum(*mp, hlen);
3410 case ETHERTYPE_IPV6:
3411 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3412 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3415 if (*mp == NULL || error != 0) /* filter may consume */
3418 if (V_pfil_member && ifp != NULL)
3419 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3422 if (*mp == NULL || error != 0) /* filter may consume */
3425 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3426 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3443 * Finally, put everything back the way it was and return
3446 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3449 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3452 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3455 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3466 * Perform basic checks on header size since
3467 * pfil assumes ip_input has already processed
3468 * it for it. Cut-and-pasted from ip_input.c.
3469 * Given how simple the IPv6 version is,
3470 * does the IPv4 version really need to be
3473 * XXX Should we update ipstat here, or not?
3474 * XXX Right now we update ipstat but not
3478 bridge_ip_checkbasic(struct mbuf **mp)
3480 struct mbuf *m = *mp;
3488 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3489 if ((m = m_copyup(m, sizeof(struct ip),
3490 (max_linkhdr + 3) & ~3)) == NULL) {
3491 /* XXXJRT new stat, please */
3492 KMOD_IPSTAT_INC(ips_toosmall);
3495 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3496 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3497 KMOD_IPSTAT_INC(ips_toosmall);
3501 ip = mtod(m, struct ip *);
3502 if (ip == NULL) goto bad;
3504 if (ip->ip_v != IPVERSION) {
3505 KMOD_IPSTAT_INC(ips_badvers);
3508 hlen = ip->ip_hl << 2;
3509 if (hlen < sizeof(struct ip)) { /* minimum header length */
3510 KMOD_IPSTAT_INC(ips_badhlen);
3513 if (hlen > m->m_len) {
3514 if ((m = m_pullup(m, hlen)) == NULL) {
3515 KMOD_IPSTAT_INC(ips_badhlen);
3518 ip = mtod(m, struct ip *);
3519 if (ip == NULL) goto bad;
3522 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3523 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3525 if (hlen == sizeof(struct ip)) {
3526 sum = in_cksum_hdr(ip);
3528 sum = in_cksum(m, hlen);
3532 KMOD_IPSTAT_INC(ips_badsum);
3536 /* Retrieve the packet length. */
3537 len = ntohs(ip->ip_len);
3540 * Check for additional length bogosity
3543 KMOD_IPSTAT_INC(ips_badlen);
3548 * Check that the amount of data in the buffers
3549 * is as at least much as the IP header would have us expect.
3550 * Drop packet if shorter than we expect.
3552 if (m->m_pkthdr.len < len) {
3553 KMOD_IPSTAT_INC(ips_tooshort);
3557 /* Checks out, proceed */
3568 * Same as above, but for IPv6.
3569 * Cut-and-pasted from ip6_input.c.
3570 * XXX Should we update ip6stat, or not?
3573 bridge_ip6_checkbasic(struct mbuf **mp)
3575 struct mbuf *m = *mp;
3576 struct ip6_hdr *ip6;
3579 * If the IPv6 header is not aligned, slurp it up into a new
3580 * mbuf with space for link headers, in the event we forward
3581 * it. Otherwise, if it is aligned, make sure the entire base
3582 * IPv6 header is in the first mbuf of the chain.
3584 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3585 struct ifnet *inifp = m->m_pkthdr.rcvif;
3586 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3587 (max_linkhdr + 3) & ~3)) == NULL) {
3588 /* XXXJRT new stat, please */
3589 IP6STAT_INC(ip6s_toosmall);
3590 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3593 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3594 struct ifnet *inifp = m->m_pkthdr.rcvif;
3595 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3596 IP6STAT_INC(ip6s_toosmall);
3597 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3602 ip6 = mtod(m, struct ip6_hdr *);
3604 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3605 IP6STAT_INC(ip6s_badvers);
3606 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3610 /* Checks out, proceed */
3623 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3626 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3627 int snap, struct llc *llc)
3629 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3633 if (m->m_len < sizeof(struct ip) &&
3634 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3636 ip = mtod(m, struct ip *);
3638 m->m_pkthdr.csum_flags |= CSUM_IP;
3639 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3644 * Walk the chain and re-add the Ethernet header for
3647 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3648 nextpkt = mcur->m_nextpkt;
3649 mcur->m_nextpkt = NULL;
3651 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3655 mprev->m_nextpkt = nextpkt;
3658 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3661 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3665 mprev->m_nextpkt = nextpkt;
3668 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3671 * The previous two M_PREPEND could have inserted one or two
3672 * mbufs in front so we have to update the previous packet's
3675 mcur->m_nextpkt = nextpkt;
3677 mprev->m_nextpkt = mcur;
3679 /* The first mbuf in the original chain needs to be
3686 KMOD_IPSTAT_INC(ips_fragmented);
3690 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3691 m = mcur->m_nextpkt;
3698 bridge_linkstate(struct ifnet *ifp)
3700 struct bridge_softc *sc = ifp->if_bridge;
3701 struct bridge_iflist *bif;
3702 struct epoch_tracker et;
3704 NET_EPOCH_ENTER_ET(et);
3706 bif = bridge_lookup_member_if(sc, ifp);
3708 NET_EPOCH_EXIT_ET(et);
3711 bridge_linkcheck(sc);
3713 bstp_linkstate(&bif->bif_stp);
3715 NET_EPOCH_EXIT_ET(et);
3719 bridge_linkcheck(struct bridge_softc *sc)
3721 struct bridge_iflist *bif;
3722 int new_link, hasls;
3724 MPASS(in_epoch(net_epoch_preempt));
3726 new_link = LINK_STATE_DOWN;
3728 /* Our link is considered up if at least one of our ports is active */
3729 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3730 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3732 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3733 new_link = LINK_STATE_UP;
3737 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3738 /* If no interfaces support link-state then we default to up */
3739 new_link = LINK_STATE_UP;
3741 if_link_state_change(sc->sc_ifp, new_link);