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 * The bridge relies heavily on the epoch(9) system to protect its data
191 * structures. This means we can safely use CK_LISTs while in NET_EPOCH, but we
192 * must ensure there is only one writer at a time.
194 * That is: for read accesses we only need to be in NET_EPOCH, but for write
195 * accesses we must hold:
197 * - BRIDGE_RT_LOCK, for any change to bridge_rtnodes
198 * - BRIDGE_LOCK, for any other change
200 * The BRIDGE_LOCK is a sleepable lock, because it is held accross ioctl()
201 * calls to bridge member interfaces and these ioctl()s can sleep.
202 * The BRIDGE_RT_LOCK is a non-sleepable mutex, because it is sometimes
203 * required while we're in NET_EPOCH and then we're not allowed to sleep.
205 #define BRIDGE_LOCK_INIT(_sc) do { \
206 sx_init(&(_sc)->sc_sx, "if_bridge"); \
207 mtx_init(&(_sc)->sc_rt_mtx, "if_bridge rt", NULL, MTX_DEF); \
209 #define BRIDGE_LOCK_DESTROY(_sc) do { \
210 sx_destroy(&(_sc)->sc_sx); \
211 mtx_destroy(&(_sc)->sc_rt_mtx); \
213 #define BRIDGE_LOCK(_sc) sx_xlock(&(_sc)->sc_sx)
214 #define BRIDGE_UNLOCK(_sc) sx_xunlock(&(_sc)->sc_sx)
215 #define BRIDGE_LOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_XLOCKED)
216 #define BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(_sc) \
217 MPASS(in_epoch(net_epoch_preempt) || sx_xlocked(&(_sc)->sc_sx))
218 #define BRIDGE_UNLOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_UNLOCKED)
219 #define BRIDGE_RT_LOCK(_sc) mtx_lock(&(_sc)->sc_rt_mtx)
220 #define BRIDGE_RT_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_rt_mtx)
221 #define BRIDGE_RT_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_rt_mtx, MA_OWNED)
222 #define BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(_sc) \
223 MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(_sc)->sc_rt_mtx))
226 * Bridge interface list entry.
228 struct bridge_iflist {
229 CK_LIST_ENTRY(bridge_iflist) bif_next;
230 struct ifnet *bif_ifp; /* member if */
231 struct bstp_port bif_stp; /* STP state */
232 uint32_t bif_flags; /* member if flags */
233 int bif_savedcaps; /* saved capabilities */
234 uint32_t bif_addrmax; /* max # of addresses */
235 uint32_t bif_addrcnt; /* cur. # of addresses */
236 uint32_t bif_addrexceeded;/* # of address violations */
237 struct epoch_context bif_epoch_ctx;
243 struct bridge_rtnode {
244 CK_LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
245 CK_LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
246 struct bridge_iflist *brt_dst; /* destination if */
247 unsigned long brt_expire; /* expiration time */
248 uint8_t brt_flags; /* address flags */
249 uint8_t brt_addr[ETHER_ADDR_LEN];
250 uint16_t brt_vlan; /* vlan id */
251 struct vnet *brt_vnet;
252 struct epoch_context brt_epoch_ctx;
254 #define brt_ifp brt_dst->bif_ifp
257 * Software state for each bridge.
259 struct bridge_softc {
260 struct ifnet *sc_ifp; /* make this an interface */
261 LIST_ENTRY(bridge_softc) sc_list;
263 struct mtx sc_rt_mtx;
264 uint32_t sc_brtmax; /* max # of addresses */
265 uint32_t sc_brtcnt; /* cur. # of addresses */
266 uint32_t sc_brttimeout; /* rt timeout in seconds */
267 struct callout sc_brcallout; /* bridge callout */
268 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
269 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
270 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
271 uint32_t sc_rthash_key; /* key for hash */
272 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
273 struct bstp_state sc_stp; /* STP state */
274 uint32_t sc_brtexceeded; /* # of cache drops */
275 struct ifnet *sc_ifaddr; /* member mac copied from */
276 struct ether_addr sc_defaddr; /* Default MAC address */
277 struct epoch_context sc_epoch_ctx;
280 VNET_DEFINE_STATIC(struct sx, bridge_list_sx);
281 #define V_bridge_list_sx VNET(bridge_list_sx)
282 static eventhandler_tag bridge_detach_cookie;
284 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
286 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
287 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone)
289 static int bridge_clone_create(struct if_clone *, int, caddr_t);
290 static void bridge_clone_destroy(struct ifnet *);
292 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
293 static void bridge_mutecaps(struct bridge_softc *);
294 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
296 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
297 static void bridge_init(void *);
298 static void bridge_dummynet(struct mbuf *, struct ifnet *);
299 static void bridge_stop(struct ifnet *, int);
300 static int bridge_transmit(struct ifnet *, struct mbuf *);
301 static void bridge_qflush(struct ifnet *);
302 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
303 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
305 static int bridge_enqueue(struct bridge_softc *, struct ifnet *,
307 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
309 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
312 static void bridge_timer(void *);
314 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
316 static void bridge_span(struct bridge_softc *, struct mbuf *);
318 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
319 uint16_t, struct bridge_iflist *, int, uint8_t);
320 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
322 static void bridge_rttrim(struct bridge_softc *);
323 static void bridge_rtage(struct bridge_softc *);
324 static void bridge_rtflush(struct bridge_softc *, int);
325 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
328 static void bridge_rtable_init(struct bridge_softc *);
329 static void bridge_rtable_fini(struct bridge_softc *);
331 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
332 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
333 const uint8_t *, uint16_t);
334 static int bridge_rtnode_insert(struct bridge_softc *,
335 struct bridge_rtnode *);
336 static void bridge_rtnode_destroy(struct bridge_softc *,
337 struct bridge_rtnode *);
338 static void bridge_rtable_expire(struct ifnet *, int);
339 static void bridge_state_change(struct ifnet *, int);
341 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
343 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
345 static void bridge_delete_member(struct bridge_softc *,
346 struct bridge_iflist *, int);
347 static void bridge_delete_span(struct bridge_softc *,
348 struct bridge_iflist *);
350 static int bridge_ioctl_add(struct bridge_softc *, void *);
351 static int bridge_ioctl_del(struct bridge_softc *, void *);
352 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
353 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
354 static int bridge_ioctl_scache(struct bridge_softc *, void *);
355 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
356 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
357 static int bridge_ioctl_rts(struct bridge_softc *, void *);
358 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
359 static int bridge_ioctl_sto(struct bridge_softc *, void *);
360 static int bridge_ioctl_gto(struct bridge_softc *, void *);
361 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
362 static int bridge_ioctl_flush(struct bridge_softc *, void *);
363 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
364 static int bridge_ioctl_spri(struct bridge_softc *, void *);
365 static int bridge_ioctl_ght(struct bridge_softc *, void *);
366 static int bridge_ioctl_sht(struct bridge_softc *, void *);
367 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
368 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
369 static int bridge_ioctl_gma(struct bridge_softc *, void *);
370 static int bridge_ioctl_sma(struct bridge_softc *, void *);
371 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
372 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
373 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
374 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
375 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
376 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
377 static int bridge_ioctl_grte(struct bridge_softc *, void *);
378 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
379 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
380 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
381 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
383 static int bridge_ip_checkbasic(struct mbuf **mp);
385 static int bridge_ip6_checkbasic(struct mbuf **mp);
387 static int bridge_fragment(struct ifnet *, struct mbuf **mp,
388 struct ether_header *, int, struct llc *);
389 static void bridge_linkstate(struct ifnet *ifp);
390 static void bridge_linkcheck(struct bridge_softc *sc);
392 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
393 #define VLANTAGOF(_m) \
394 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
396 static struct bstp_cb_ops bridge_ops = {
397 .bcb_state = bridge_state_change,
398 .bcb_rtage = bridge_rtable_expire
401 SYSCTL_DECL(_net_link);
402 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
405 /* only pass IP[46] packets when pfil is enabled */
406 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
407 #define V_pfil_onlyip VNET(pfil_onlyip)
408 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
409 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
410 "Only pass IP packets when pfil is enabled");
412 /* run pfil hooks on the bridge interface */
413 VNET_DEFINE_STATIC(int, pfil_bridge) = 1;
414 #define V_pfil_bridge VNET(pfil_bridge)
415 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
416 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
417 "Packet filter on the bridge interface");
419 /* layer2 filter with ipfw */
420 VNET_DEFINE_STATIC(int, pfil_ipfw);
421 #define V_pfil_ipfw VNET(pfil_ipfw)
423 /* layer2 ARP filter with ipfw */
424 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
425 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
426 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
427 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
428 "Filter ARP packets through IPFW layer2");
430 /* run pfil hooks on the member interface */
431 VNET_DEFINE_STATIC(int, pfil_member) = 1;
432 #define V_pfil_member VNET(pfil_member)
433 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
434 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
435 "Packet filter on the member interface");
437 /* run pfil hooks on the physical interface for locally destined packets */
438 VNET_DEFINE_STATIC(int, pfil_local_phys);
439 #define V_pfil_local_phys VNET(pfil_local_phys)
440 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
441 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
442 "Packet filter on the physical interface for locally destined packets");
444 /* log STP state changes */
445 VNET_DEFINE_STATIC(int, log_stp);
446 #define V_log_stp VNET(log_stp)
447 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
448 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
449 "Log STP state changes");
451 /* share MAC with first bridge member */
452 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
453 #define V_bridge_inherit_mac VNET(bridge_inherit_mac)
454 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
455 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
456 "Inherit MAC address from the first bridge member");
458 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
459 #define V_allow_llz_overlap VNET(allow_llz_overlap)
460 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
461 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
462 "Allow overlap of link-local scope "
463 "zones of a bridge interface and the member interfaces");
465 struct bridge_control {
466 int (*bc_func)(struct bridge_softc *, void *);
471 #define BC_F_COPYIN 0x01 /* copy arguments in */
472 #define BC_F_COPYOUT 0x02 /* copy arguments out */
473 #define BC_F_SUSER 0x04 /* do super-user check */
475 const struct bridge_control bridge_control_table[] = {
476 { bridge_ioctl_add, sizeof(struct ifbreq),
477 BC_F_COPYIN|BC_F_SUSER },
478 { bridge_ioctl_del, sizeof(struct ifbreq),
479 BC_F_COPYIN|BC_F_SUSER },
481 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
482 BC_F_COPYIN|BC_F_COPYOUT },
483 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
484 BC_F_COPYIN|BC_F_SUSER },
486 { bridge_ioctl_scache, sizeof(struct ifbrparam),
487 BC_F_COPYIN|BC_F_SUSER },
488 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
491 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
492 BC_F_COPYIN|BC_F_COPYOUT },
493 { bridge_ioctl_rts, sizeof(struct ifbaconf),
494 BC_F_COPYIN|BC_F_COPYOUT },
496 { bridge_ioctl_saddr, sizeof(struct ifbareq),
497 BC_F_COPYIN|BC_F_SUSER },
499 { bridge_ioctl_sto, sizeof(struct ifbrparam),
500 BC_F_COPYIN|BC_F_SUSER },
501 { bridge_ioctl_gto, sizeof(struct ifbrparam),
504 { bridge_ioctl_daddr, sizeof(struct ifbareq),
505 BC_F_COPYIN|BC_F_SUSER },
507 { bridge_ioctl_flush, sizeof(struct ifbreq),
508 BC_F_COPYIN|BC_F_SUSER },
510 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
512 { bridge_ioctl_spri, sizeof(struct ifbrparam),
513 BC_F_COPYIN|BC_F_SUSER },
515 { bridge_ioctl_ght, sizeof(struct ifbrparam),
517 { bridge_ioctl_sht, sizeof(struct ifbrparam),
518 BC_F_COPYIN|BC_F_SUSER },
520 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
522 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
523 BC_F_COPYIN|BC_F_SUSER },
525 { bridge_ioctl_gma, sizeof(struct ifbrparam),
527 { bridge_ioctl_sma, sizeof(struct ifbrparam),
528 BC_F_COPYIN|BC_F_SUSER },
530 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
531 BC_F_COPYIN|BC_F_SUSER },
533 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
534 BC_F_COPYIN|BC_F_SUSER },
536 { bridge_ioctl_addspan, sizeof(struct ifbreq),
537 BC_F_COPYIN|BC_F_SUSER },
538 { bridge_ioctl_delspan, sizeof(struct ifbreq),
539 BC_F_COPYIN|BC_F_SUSER },
541 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
544 { bridge_ioctl_grte, sizeof(struct ifbrparam),
547 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
548 BC_F_COPYIN|BC_F_COPYOUT },
550 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
551 BC_F_COPYIN|BC_F_SUSER },
553 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
554 BC_F_COPYIN|BC_F_SUSER },
556 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
557 BC_F_COPYIN|BC_F_SUSER },
560 const int bridge_control_table_size = nitems(bridge_control_table);
562 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
563 #define V_bridge_list VNET(bridge_list)
564 #define BRIDGE_LIST_LOCK_INIT(x) sx_init(&V_bridge_list_sx, \
566 #define BRIDGE_LIST_LOCK_DESTROY(x) sx_destroy(&V_bridge_list_sx)
567 #define BRIDGE_LIST_LOCK(x) sx_xlock(&V_bridge_list_sx)
568 #define BRIDGE_LIST_UNLOCK(x) sx_xunlock(&V_bridge_list_sx)
570 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
571 #define V_bridge_cloner VNET(bridge_cloner)
573 static const char bridge_name[] = "bridge";
576 vnet_bridge_init(const void *unused __unused)
579 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
580 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
582 BRIDGE_LIST_LOCK_INIT();
583 LIST_INIT(&V_bridge_list);
584 V_bridge_cloner = if_clone_simple(bridge_name,
585 bridge_clone_create, bridge_clone_destroy, 0);
587 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
588 vnet_bridge_init, NULL);
591 vnet_bridge_uninit(const void *unused __unused)
594 if_clone_detach(V_bridge_cloner);
595 V_bridge_cloner = NULL;
596 BRIDGE_LIST_LOCK_DESTROY();
598 /* Callbacks may use the UMA zone. */
599 epoch_drain_callbacks(net_epoch_preempt);
601 uma_zdestroy(V_bridge_rtnode_zone);
603 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
604 vnet_bridge_uninit, NULL);
607 bridge_modevent(module_t mod, int type, void *data)
612 bridge_dn_p = bridge_dummynet;
613 bridge_detach_cookie = EVENTHANDLER_REGISTER(
614 ifnet_departure_event, bridge_ifdetach, NULL,
615 EVENTHANDLER_PRI_ANY);
618 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
619 bridge_detach_cookie);
628 static moduledata_t bridge_mod = {
634 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
635 MODULE_VERSION(if_bridge, 1);
636 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
639 * handler for net.link.bridge.ipfw
642 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
644 int enable = V_pfil_ipfw;
647 error = sysctl_handle_int(oidp, &enable, 0, req);
650 if (enable != V_pfil_ipfw) {
651 V_pfil_ipfw = enable;
654 * Disable pfil so that ipfw doesnt run twice, if the user
655 * really wants both then they can re-enable pfil_bridge and/or
656 * pfil_member. Also allow non-ip packets as ipfw can filter by
668 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
669 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT,
670 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
671 "Layer2 filter with IPFW");
674 * bridge_clone_create:
676 * Create a new bridge instance.
679 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
681 struct bridge_softc *sc;
684 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
685 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
691 BRIDGE_LOCK_INIT(sc);
692 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
693 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
695 /* Initialize our routing table. */
696 bridge_rtable_init(sc);
698 callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
700 CK_LIST_INIT(&sc->sc_iflist);
701 CK_LIST_INIT(&sc->sc_spanlist);
704 if_initname(ifp, bridge_name, unit);
705 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
706 ifp->if_ioctl = bridge_ioctl;
707 ifp->if_transmit = bridge_transmit;
708 ifp->if_qflush = bridge_qflush;
709 ifp->if_init = bridge_init;
710 ifp->if_type = IFT_BRIDGE;
712 ether_gen_addr(ifp, &sc->sc_defaddr);
714 bstp_attach(&sc->sc_stp, &bridge_ops);
715 ether_ifattach(ifp, sc->sc_defaddr.octet);
716 /* Now undo some of the damage... */
717 ifp->if_baudrate = 0;
718 ifp->if_type = IFT_BRIDGE;
721 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
722 BRIDGE_LIST_UNLOCK();
728 bridge_clone_destroy_cb(struct epoch_context *ctx)
730 struct bridge_softc *sc;
732 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
734 BRIDGE_LOCK_DESTROY(sc);
739 * bridge_clone_destroy:
741 * Destroy a bridge instance.
744 bridge_clone_destroy(struct ifnet *ifp)
746 struct bridge_softc *sc = ifp->if_softc;
747 struct bridge_iflist *bif;
748 struct epoch_tracker et;
753 ifp->if_flags &= ~IFF_UP;
755 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
756 bridge_delete_member(sc, bif, 0);
758 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
759 bridge_delete_span(sc, bif);
762 /* Tear down the routing table. */
763 bridge_rtable_fini(sc);
769 callout_drain(&sc->sc_brcallout);
772 LIST_REMOVE(sc, sc_list);
773 BRIDGE_LIST_UNLOCK();
775 bstp_detach(&sc->sc_stp);
781 NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx);
787 * Handle a control request from the operator.
790 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
792 struct bridge_softc *sc = ifp->if_softc;
793 struct ifreq *ifr = (struct ifreq *)data;
794 struct bridge_iflist *bif;
795 struct thread *td = curthread;
797 struct ifbreq ifbreq;
798 struct ifbifconf ifbifconf;
799 struct ifbareq ifbareq;
800 struct ifbaconf ifbaconf;
801 struct ifbrparam ifbrparam;
802 struct ifbropreq ifbropreq;
804 struct ifdrv *ifd = (struct ifdrv *) data;
805 const struct bridge_control *bc;
806 int error = 0, oldmtu;
817 if (ifd->ifd_cmd >= bridge_control_table_size) {
821 bc = &bridge_control_table[ifd->ifd_cmd];
823 if (cmd == SIOCGDRVSPEC &&
824 (bc->bc_flags & BC_F_COPYOUT) == 0) {
828 else if (cmd == SIOCSDRVSPEC &&
829 (bc->bc_flags & BC_F_COPYOUT) != 0) {
834 if (bc->bc_flags & BC_F_SUSER) {
835 error = priv_check(td, PRIV_NET_BRIDGE);
840 if (ifd->ifd_len != bc->bc_argsize ||
841 ifd->ifd_len > sizeof(args)) {
846 bzero(&args, sizeof(args));
847 if (bc->bc_flags & BC_F_COPYIN) {
848 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
853 oldmtu = ifp->if_mtu;
854 error = (*bc->bc_func)(sc, &args);
859 * Bridge MTU may change during addition of the first port.
860 * If it did, do network layer specific procedure.
862 if (ifp->if_mtu != oldmtu) {
869 if (bc->bc_flags & BC_F_COPYOUT)
870 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
875 if (!(ifp->if_flags & IFF_UP) &&
876 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
878 * If interface is marked down and it is running,
879 * then stop and disable it.
882 } else if ((ifp->if_flags & IFF_UP) &&
883 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
885 * If interface is marked up and it is stopped, then
895 if (ifr->ifr_mtu < 576) {
899 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
900 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
903 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
904 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
905 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
906 " != %d\n", sc->sc_ifp->if_xname,
907 bif->bif_ifp->if_mtu,
908 bif->bif_ifp->if_xname, ifr->ifr_mtu);
914 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
918 * drop the lock as ether_ioctl() will call bridge_start() and
919 * cause the lock to be recursed.
922 error = ether_ioctl(ifp, cmd, data);
935 * Clear or restore unwanted capabilities on the member interface
938 bridge_mutecaps(struct bridge_softc *sc)
940 struct bridge_iflist *bif;
943 BRIDGE_LOCK_ASSERT(sc);
945 /* Initial bitmask of capabilities to test */
946 mask = BRIDGE_IFCAPS_MASK;
948 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
949 /* Every member must support it or its disabled */
950 mask &= bif->bif_savedcaps;
953 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
954 enabled = bif->bif_ifp->if_capenable;
955 enabled &= ~BRIDGE_IFCAPS_STRIP;
956 /* strip off mask bits and enable them again if allowed */
957 enabled &= ~BRIDGE_IFCAPS_MASK;
959 bridge_set_ifcap(sc, bif, enabled);
964 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
966 struct ifnet *ifp = bif->bif_ifp;
968 int error, mask, stuck;
970 bzero(&ifr, sizeof(ifr));
971 ifr.ifr_reqcap = set;
973 if (ifp->if_capenable != set) {
974 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
976 if_printf(sc->sc_ifp,
977 "error setting capabilities on %s: %d\n",
978 ifp->if_xname, error);
979 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
980 stuck = ifp->if_capenable & mask & ~set;
982 if_printf(sc->sc_ifp,
983 "can't disable some capabilities on %s: 0x%x\n",
984 ifp->if_xname, stuck);
989 * bridge_lookup_member:
991 * Lookup a bridge member interface.
993 static struct bridge_iflist *
994 bridge_lookup_member(struct bridge_softc *sc, const char *name)
996 struct bridge_iflist *bif;
999 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1001 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1003 if (strcmp(ifp->if_xname, name) == 0)
1011 * bridge_lookup_member_if:
1013 * Lookup a bridge member interface by ifnet*.
1015 static struct bridge_iflist *
1016 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1018 struct bridge_iflist *bif;
1020 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1022 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1023 if (bif->bif_ifp == member_ifp)
1031 bridge_delete_member_cb(struct epoch_context *ctx)
1033 struct bridge_iflist *bif;
1035 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1037 free(bif, M_DEVBUF);
1041 * bridge_delete_member:
1043 * Delete the specified member interface.
1046 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1049 struct ifnet *ifs = bif->bif_ifp;
1050 struct ifnet *fif = NULL;
1051 struct bridge_iflist *bifl;
1053 BRIDGE_LOCK_ASSERT(sc);
1055 if (bif->bif_flags & IFBIF_STP)
1056 bstp_disable(&bif->bif_stp);
1058 ifs->if_bridge = NULL;
1059 CK_LIST_REMOVE(bif, bif_next);
1062 * If removing the interface that gave the bridge its mac address, set
1063 * the mac address of the bridge to the address of the next member, or
1064 * to its default address if no members are left.
1066 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1067 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1068 bcopy(&sc->sc_defaddr,
1069 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1070 sc->sc_ifaddr = NULL;
1072 bifl = CK_LIST_FIRST(&sc->sc_iflist);
1073 fif = bifl->bif_ifp;
1074 bcopy(IF_LLADDR(fif),
1075 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1076 sc->sc_ifaddr = fif;
1078 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1081 bridge_linkcheck(sc);
1082 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1084 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1085 BRIDGE_RT_UNLOCK(sc);
1086 KASSERT(bif->bif_addrcnt == 0,
1087 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1089 ifs->if_bridge_output = NULL;
1090 ifs->if_bridge_input = NULL;
1091 ifs->if_bridge_linkstate = NULL;
1093 switch (ifs->if_type) {
1097 * Take the interface out of promiscuous mode, but only
1098 * if it was promiscuous in the first place. It might
1099 * not be if we're in the bridge_ioctl_add() error path.
1101 if (ifs->if_flags & IFF_PROMISC)
1102 (void) ifpromisc(ifs, 0);
1110 panic("bridge_delete_member: impossible");
1114 /* reneable any interface capabilities */
1115 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1117 bstp_destroy(&bif->bif_stp); /* prepare to free */
1119 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1123 * bridge_delete_span:
1125 * Delete the specified span interface.
1128 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1130 BRIDGE_LOCK_ASSERT(sc);
1132 KASSERT(bif->bif_ifp->if_bridge == NULL,
1133 ("%s: not a span interface", __func__));
1135 CK_LIST_REMOVE(bif, bif_next);
1137 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1141 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1143 struct ifbreq *req = arg;
1144 struct bridge_iflist *bif = NULL;
1148 ifs = ifunit(req->ifbr_ifsname);
1151 if (ifs->if_ioctl == NULL) /* must be supported */
1154 /* If it's in the span list, it can't be a member. */
1155 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1156 if (ifs == bif->bif_ifp)
1159 if (ifs->if_bridge == sc)
1162 if (ifs->if_bridge != NULL)
1165 switch (ifs->if_type) {
1169 /* permitted interface types */
1177 * Two valid inet6 addresses with link-local scope must not be
1178 * on the parent interface and the member interfaces at the
1179 * same time. This restriction is needed to prevent violation
1180 * of link-local scope zone. Attempts to add a member
1181 * interface which has inet6 addresses when the parent has
1182 * inet6 triggers removal of all inet6 addresses on the member
1186 /* Check if the parent interface has a link-local scope addr. */
1187 if (V_allow_llz_overlap == 0 &&
1188 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1190 * If any, remove all inet6 addresses from the member
1193 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1194 if (in6ifa_llaonifp(bif->bif_ifp)) {
1195 in6_ifdetach(bif->bif_ifp);
1196 if_printf(sc->sc_ifp,
1197 "IPv6 addresses on %s have been removed "
1198 "before adding it as a member to prevent "
1199 "IPv6 address scope violation.\n",
1200 bif->bif_ifp->if_xname);
1203 if (in6ifa_llaonifp(ifs)) {
1205 if_printf(sc->sc_ifp,
1206 "IPv6 addresses on %s have been removed "
1207 "before adding it as a member to prevent "
1208 "IPv6 address scope violation.\n",
1213 /* Allow the first Ethernet member to define the MTU */
1214 if (CK_LIST_EMPTY(&sc->sc_iflist))
1215 sc->sc_ifp->if_mtu = ifs->if_mtu;
1216 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1217 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1218 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1222 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1227 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1228 bif->bif_savedcaps = ifs->if_capenable;
1231 * Assign the interface's MAC address to the bridge if it's the first
1232 * member and the MAC address of the bridge has not been changed from
1233 * the default randomly generated one.
1235 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1236 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1237 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1238 sc->sc_ifaddr = ifs;
1239 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1242 ifs->if_bridge = sc;
1243 ifs->if_bridge_output = bridge_output;
1244 ifs->if_bridge_input = bridge_input;
1245 ifs->if_bridge_linkstate = bridge_linkstate;
1246 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1248 * XXX: XLOCK HERE!?!
1250 * NOTE: insert_***HEAD*** should be safe for the traversals.
1252 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1254 /* Set interface capabilities to the intersection set of all members */
1255 bridge_mutecaps(sc);
1256 bridge_linkcheck(sc);
1258 /* Place the interface into promiscuous mode */
1259 switch (ifs->if_type) {
1262 error = ifpromisc(ifs, 1);
1267 bridge_delete_member(sc, bif, 0);
1272 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1274 struct ifbreq *req = arg;
1275 struct bridge_iflist *bif;
1277 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1281 bridge_delete_member(sc, bif, 0);
1287 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1289 struct ifbreq *req = arg;
1290 struct bridge_iflist *bif;
1291 struct bstp_port *bp;
1293 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1298 req->ifbr_ifsflags = bif->bif_flags;
1299 req->ifbr_state = bp->bp_state;
1300 req->ifbr_priority = bp->bp_priority;
1301 req->ifbr_path_cost = bp->bp_path_cost;
1302 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1303 req->ifbr_proto = bp->bp_protover;
1304 req->ifbr_role = bp->bp_role;
1305 req->ifbr_stpflags = bp->bp_flags;
1306 req->ifbr_addrcnt = bif->bif_addrcnt;
1307 req->ifbr_addrmax = bif->bif_addrmax;
1308 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1310 /* Copy STP state options as flags */
1311 if (bp->bp_operedge)
1312 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1313 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1314 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1315 if (bp->bp_ptp_link)
1316 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1317 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1318 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1319 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1320 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1321 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1322 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1327 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1329 struct ifbreq *req = arg;
1330 struct bridge_iflist *bif;
1331 struct bstp_port *bp;
1334 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1339 if (req->ifbr_ifsflags & IFBIF_SPAN)
1340 /* SPAN is readonly */
1343 if (req->ifbr_ifsflags & IFBIF_STP) {
1344 if ((bif->bif_flags & IFBIF_STP) == 0) {
1345 error = bstp_enable(&bif->bif_stp);
1350 if ((bif->bif_flags & IFBIF_STP) != 0)
1351 bstp_disable(&bif->bif_stp);
1354 /* Pass on STP flags */
1355 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1356 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1357 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1358 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1360 /* Save the bits relating to the bridge */
1361 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1367 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1369 struct ifbrparam *param = arg;
1371 sc->sc_brtmax = param->ifbrp_csize;
1378 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1380 struct ifbrparam *param = arg;
1382 param->ifbrp_csize = sc->sc_brtmax;
1388 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1390 struct ifbifconf *bifc = arg;
1391 struct bridge_iflist *bif;
1394 int count, buflen, len, error = 0;
1397 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1399 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1402 buflen = sizeof(breq) * count;
1403 if (bifc->ifbic_len == 0) {
1404 bifc->ifbic_len = buflen;
1407 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1413 len = min(bifc->ifbic_len, buflen);
1414 bzero(&breq, sizeof(breq));
1415 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1416 if (len < sizeof(breq))
1419 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1420 sizeof(breq.ifbr_ifsname));
1421 /* Fill in the ifbreq structure */
1422 error = bridge_ioctl_gifflags(sc, &breq);
1425 memcpy(buf, &breq, sizeof(breq));
1427 buf += sizeof(breq);
1428 len -= sizeof(breq);
1430 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1431 if (len < sizeof(breq))
1434 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1435 sizeof(breq.ifbr_ifsname));
1436 breq.ifbr_ifsflags = bif->bif_flags;
1437 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1438 memcpy(buf, &breq, sizeof(breq));
1440 buf += sizeof(breq);
1441 len -= sizeof(breq);
1444 bifc->ifbic_len = sizeof(breq) * count;
1445 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1446 free(outbuf, M_TEMP);
1451 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1453 struct ifbaconf *bac = arg;
1454 struct bridge_rtnode *brt;
1455 struct ifbareq bareq;
1457 int count, buflen, len, error = 0;
1459 if (bac->ifbac_len == 0)
1463 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1465 buflen = sizeof(bareq) * count;
1467 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1473 len = min(bac->ifbac_len, buflen);
1474 bzero(&bareq, sizeof(bareq));
1475 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1476 if (len < sizeof(bareq))
1478 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1479 sizeof(bareq.ifba_ifsname));
1480 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1481 bareq.ifba_vlan = brt->brt_vlan;
1482 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1483 time_uptime < brt->brt_expire)
1484 bareq.ifba_expire = brt->brt_expire - time_uptime;
1486 bareq.ifba_expire = 0;
1487 bareq.ifba_flags = brt->brt_flags;
1489 memcpy(buf, &bareq, sizeof(bareq));
1491 buf += sizeof(bareq);
1492 len -= sizeof(bareq);
1495 bac->ifbac_len = sizeof(bareq) * count;
1496 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1497 free(outbuf, M_TEMP);
1502 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1504 struct ifbareq *req = arg;
1505 struct bridge_iflist *bif;
1506 struct epoch_tracker et;
1509 NET_EPOCH_ENTER(et);
1510 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1516 /* bridge_rtupdate() may acquire the lock. */
1517 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1525 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1527 struct ifbrparam *param = arg;
1529 sc->sc_brttimeout = param->ifbrp_ctime;
1534 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1536 struct ifbrparam *param = arg;
1538 param->ifbrp_ctime = sc->sc_brttimeout;
1543 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1545 struct ifbareq *req = arg;
1547 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1551 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1553 struct ifbreq *req = arg;
1556 bridge_rtflush(sc, req->ifbr_ifsflags);
1557 BRIDGE_RT_UNLOCK(sc);
1563 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1565 struct ifbrparam *param = arg;
1566 struct bstp_state *bs = &sc->sc_stp;
1568 param->ifbrp_prio = bs->bs_bridge_priority;
1573 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1575 struct ifbrparam *param = arg;
1577 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1581 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1583 struct ifbrparam *param = arg;
1584 struct bstp_state *bs = &sc->sc_stp;
1586 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1591 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1593 struct ifbrparam *param = arg;
1595 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1599 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1601 struct ifbrparam *param = arg;
1602 struct bstp_state *bs = &sc->sc_stp;
1604 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1609 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1611 struct ifbrparam *param = arg;
1613 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1617 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1619 struct ifbrparam *param = arg;
1620 struct bstp_state *bs = &sc->sc_stp;
1622 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1627 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1629 struct ifbrparam *param = arg;
1631 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1635 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1637 struct ifbreq *req = arg;
1638 struct bridge_iflist *bif;
1640 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1644 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1648 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1650 struct ifbreq *req = arg;
1651 struct bridge_iflist *bif;
1653 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1657 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1661 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1663 struct ifbreq *req = arg;
1664 struct bridge_iflist *bif;
1666 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1670 bif->bif_addrmax = req->ifbr_addrmax;
1675 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1677 struct ifbreq *req = arg;
1678 struct bridge_iflist *bif = NULL;
1681 ifs = ifunit(req->ifbr_ifsname);
1685 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1686 if (ifs == bif->bif_ifp)
1689 if (ifs->if_bridge != NULL)
1692 switch (ifs->if_type) {
1701 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1706 bif->bif_flags = IFBIF_SPAN;
1708 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1714 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1716 struct ifbreq *req = arg;
1717 struct bridge_iflist *bif;
1720 ifs = ifunit(req->ifbr_ifsname);
1724 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1725 if (ifs == bif->bif_ifp)
1731 bridge_delete_span(sc, bif);
1737 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1739 struct ifbropreq *req = arg;
1740 struct bstp_state *bs = &sc->sc_stp;
1741 struct bstp_port *root_port;
1743 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1744 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1745 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1747 root_port = bs->bs_root_port;
1748 if (root_port == NULL)
1749 req->ifbop_root_port = 0;
1751 req->ifbop_root_port = root_port->bp_ifp->if_index;
1753 req->ifbop_holdcount = bs->bs_txholdcount;
1754 req->ifbop_priority = bs->bs_bridge_priority;
1755 req->ifbop_protocol = bs->bs_protover;
1756 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1757 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1758 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1759 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1760 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1761 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1767 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1769 struct ifbrparam *param = arg;
1771 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1776 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1778 struct ifbpstpconf *bifstp = arg;
1779 struct bridge_iflist *bif;
1780 struct bstp_port *bp;
1781 struct ifbpstpreq bpreq;
1783 int count, buflen, len, error = 0;
1786 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1787 if ((bif->bif_flags & IFBIF_STP) != 0)
1791 buflen = sizeof(bpreq) * count;
1792 if (bifstp->ifbpstp_len == 0) {
1793 bifstp->ifbpstp_len = buflen;
1797 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1803 len = min(bifstp->ifbpstp_len, buflen);
1804 bzero(&bpreq, sizeof(bpreq));
1805 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1806 if (len < sizeof(bpreq))
1809 if ((bif->bif_flags & IFBIF_STP) == 0)
1813 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1814 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1815 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1816 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1817 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1818 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1820 memcpy(buf, &bpreq, sizeof(bpreq));
1822 buf += sizeof(bpreq);
1823 len -= sizeof(bpreq);
1826 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1827 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1828 free(outbuf, M_TEMP);
1833 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1835 struct ifbrparam *param = arg;
1837 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1841 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1843 struct ifbrparam *param = arg;
1845 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1851 * Detach an interface from a bridge. Called when a member
1852 * interface is detaching.
1855 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1857 struct bridge_softc *sc = ifp->if_bridge;
1858 struct bridge_iflist *bif;
1860 if (ifp->if_flags & IFF_RENAMING)
1862 if (V_bridge_cloner == NULL) {
1864 * This detach handler can be called after
1865 * vnet_bridge_uninit(). Just return in that case.
1869 /* Check if the interface is a bridge member */
1873 bif = bridge_lookup_member_if(sc, ifp);
1875 bridge_delete_member(sc, bif, 1);
1881 /* Check if the interface is a span port */
1883 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1885 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1886 if (ifp == bif->bif_ifp) {
1887 bridge_delete_span(sc, bif);
1893 BRIDGE_LIST_UNLOCK();
1899 * Initialize a bridge interface.
1902 bridge_init(void *xsc)
1904 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1905 struct ifnet *ifp = sc->sc_ifp;
1907 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1911 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1914 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1915 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1923 * Stop the bridge interface.
1926 bridge_stop(struct ifnet *ifp, int disable)
1928 struct bridge_softc *sc = ifp->if_softc;
1930 BRIDGE_LOCK_ASSERT(sc);
1932 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1936 callout_stop(&sc->sc_brcallout);
1938 bstp_stop(&sc->sc_stp);
1940 bridge_rtflush(sc, IFBF_FLUSHDYN);
1941 BRIDGE_RT_UNLOCK(sc);
1943 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1949 * Enqueue a packet on a bridge member interface.
1953 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1959 /* We may be sending a fragment so traverse the mbuf */
1962 m->m_nextpkt = NULL;
1963 len = m->m_pkthdr.len;
1964 mflags = m->m_flags;
1967 * If underlying interface can not do VLAN tag insertion itself
1968 * then attach a packet tag that holds it.
1970 if ((m->m_flags & M_VLANTAG) &&
1971 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1972 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1975 "unable to prepend VLAN header\n");
1976 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
1979 m->m_flags &= ~M_VLANTAG;
1982 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
1983 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1985 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
1989 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
1990 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
1991 if (mflags & M_MCAST)
1992 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2001 * Receive a queued packet from dummynet and pass it on to the output
2004 * The mbuf has the Ethernet header already attached.
2007 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2009 struct bridge_softc *sc;
2011 sc = ifp->if_bridge;
2014 * The packet didnt originate from a member interface. This should only
2015 * ever happen if a member interface is removed while packets are
2023 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2025 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2028 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2034 bridge_enqueue(sc, ifp, m);
2040 * Send output from a bridge member interface. This
2041 * performs the bridging function for locally originated
2044 * The mbuf has the Ethernet header already attached. We must
2045 * enqueue or free the mbuf before returning.
2048 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2051 struct ether_header *eh;
2052 struct ifnet *bifp, *dst_if;
2053 struct bridge_softc *sc;
2058 if (m->m_len < ETHER_HDR_LEN) {
2059 m = m_pullup(m, ETHER_HDR_LEN);
2064 eh = mtod(m, struct ether_header *);
2065 sc = ifp->if_bridge;
2066 vlan = VLANTAGOF(m);
2071 * If bridge is down, but the original output interface is up,
2072 * go ahead and send out that interface. Otherwise, the packet
2075 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2081 * If the packet is a multicast, or we don't know a better way to
2082 * get there, send to all interfaces.
2084 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2087 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2088 /* Tap any traffic not passing back out the originating interface */
2090 ETHER_BPF_MTAP(bifp, m);
2091 if (dst_if == NULL) {
2092 struct bridge_iflist *bif;
2098 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2099 dst_if = bif->bif_ifp;
2101 if (dst_if->if_type == IFT_GIF)
2103 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2107 * If this is not the original output interface,
2108 * and the interface is participating in spanning
2109 * tree, make sure the port is in a state that
2110 * allows forwarding.
2112 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2113 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2116 if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2120 mc = m_copypacket(m, M_NOWAIT);
2122 if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2127 bridge_enqueue(sc, dst_if, mc);
2136 * XXX Spanning tree consideration here?
2140 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2145 bridge_enqueue(sc, dst_if, m);
2152 * Do output on a bridge.
2156 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2158 struct bridge_softc *sc;
2159 struct ether_header *eh;
2160 struct ifnet *dst_if;
2165 ETHER_BPF_MTAP(ifp, m);
2167 eh = mtod(m, struct ether_header *);
2169 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2170 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2171 error = bridge_enqueue(sc, dst_if, m);
2173 bridge_broadcast(sc, ifp, m, 0);
2179 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2182 bridge_qflush(struct ifnet *ifp __unused)
2189 * The forwarding function of the bridge.
2191 * NOTE: Releases the lock on return.
2194 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2197 struct bridge_iflist *dbif;
2198 struct ifnet *src_if, *dst_if, *ifp;
2199 struct ether_header *eh;
2206 src_if = m->m_pkthdr.rcvif;
2209 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2210 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2211 vlan = VLANTAGOF(m);
2213 if ((sbif->bif_flags & IFBIF_STP) &&
2214 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2217 eh = mtod(m, struct ether_header *);
2218 dst = eh->ether_dhost;
2220 /* If the interface is learning, record the address. */
2221 if (sbif->bif_flags & IFBIF_LEARNING) {
2222 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2223 sbif, 0, IFBAF_DYNAMIC);
2225 * If the interface has addresses limits then deny any source
2226 * that is not in the cache.
2228 if (error && sbif->bif_addrmax)
2232 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2233 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2237 * At this point, the port either doesn't participate
2238 * in spanning tree or it is in the forwarding state.
2242 * If the packet is unicast, destined for someone on
2243 * "this" side of the bridge, drop it.
2245 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2246 dst_if = bridge_rtlookup(sc, dst, vlan);
2247 if (src_if == dst_if)
2251 * Check if its a reserved multicast address, any address
2252 * listed in 802.1D section 7.12.6 may not be forwarded by the
2254 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2256 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2257 dst[2] == 0xc2 && dst[3] == 0x00 &&
2258 dst[4] == 0x00 && dst[5] <= 0x0f)
2261 /* ...forward it to all interfaces. */
2262 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2267 * If we have a destination interface which is a member of our bridge,
2268 * OR this is a unicast packet, push it through the bpf(4) machinery.
2269 * For broadcast or multicast packets, don't bother because it will
2270 * be reinjected into ether_input. We do this before we pass the packets
2271 * through the pfil(9) framework, as it is possible that pfil(9) will
2272 * drop the packet, or possibly modify it, making it difficult to debug
2273 * firewall issues on the bridge.
2275 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2276 ETHER_BPF_MTAP(ifp, m);
2278 /* run the packet filter */
2279 if (PFIL_HOOKED_IN(V_inet_pfil_head)
2281 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2284 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2290 if (dst_if == NULL) {
2291 bridge_broadcast(sc, src_if, m, 1);
2296 * At this point, we're dealing with a unicast frame
2297 * going to a different interface.
2299 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2302 dbif = bridge_lookup_member_if(sc, dst_if);
2304 /* Not a member of the bridge (anymore?) */
2307 /* Private segments can not talk to each other */
2308 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2311 if ((dbif->bif_flags & IFBIF_STP) &&
2312 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2315 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2317 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2320 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2326 bridge_enqueue(sc, dst_if, m);
2336 * Receive input from a member interface. Queue the packet for
2337 * bridging if it is not for us.
2339 static struct mbuf *
2340 bridge_input(struct ifnet *ifp, struct mbuf *m)
2342 struct bridge_softc *sc = ifp->if_bridge;
2343 struct bridge_iflist *bif, *bif2;
2345 struct ether_header *eh;
2346 struct mbuf *mc, *mc2;
2352 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2356 vlan = VLANTAGOF(m);
2359 * Implement support for bridge monitoring. If this flag has been
2360 * set on this interface, discard the packet once we push it through
2361 * the bpf(4) machinery, but before we do, increment the byte and
2362 * packet counters associated with this interface.
2364 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2365 m->m_pkthdr.rcvif = bifp;
2366 ETHER_BPF_MTAP(bifp, m);
2367 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2368 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2372 bif = bridge_lookup_member_if(sc, ifp);
2377 eh = mtod(m, struct ether_header *);
2381 if (m->m_flags & (M_BCAST|M_MCAST)) {
2382 /* Tap off 802.1D packets; they do not get forwarded. */
2383 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2384 ETHER_ADDR_LEN) == 0) {
2385 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2389 if ((bif->bif_flags & IFBIF_STP) &&
2390 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2395 * Make a deep copy of the packet and enqueue the copy
2396 * for bridge processing; return the original packet for
2399 mc = m_dup(m, M_NOWAIT);
2404 /* Perform the bridge forwarding function with the copy. */
2405 bridge_forward(sc, bif, mc);
2408 * Reinject the mbuf as arriving on the bridge so we have a
2409 * chance at claiming multicast packets. We can not loop back
2410 * here from ether_input as a bridge is never a member of a
2413 KASSERT(bifp->if_bridge == NULL,
2414 ("loop created in bridge_input"));
2415 mc2 = m_dup(m, M_NOWAIT);
2417 /* Keep the layer3 header aligned */
2418 int i = min(mc2->m_pkthdr.len, max_protohdr);
2419 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2422 mc2->m_pkthdr.rcvif = bifp;
2423 (*bifp->if_input)(bifp, mc2);
2426 /* Return the original packet for local processing. */
2430 if ((bif->bif_flags & IFBIF_STP) &&
2431 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2435 #if (defined(INET) || defined(INET6))
2436 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2437 || ((iface)->if_carp \
2438 && (*carp_forus_p)((iface), eh->ether_dhost))
2439 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2440 || ((iface)->if_carp \
2441 && (*carp_forus_p)((iface), eh->ether_shost))
2443 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2444 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2448 # define OR_PFIL_HOOKED_INET6 \
2449 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2451 # define OR_PFIL_HOOKED_INET6
2454 #define GRAB_OUR_PACKETS(iface) \
2455 if ((iface)->if_type == IFT_GIF) \
2457 /* It is destined for us. */ \
2458 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2459 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2461 if (bif->bif_flags & IFBIF_LEARNING) { \
2462 error = bridge_rtupdate(sc, eh->ether_shost, \
2463 vlan, bif, 0, IFBAF_DYNAMIC); \
2464 if (error && bif->bif_addrmax) { \
2469 m->m_pkthdr.rcvif = iface; \
2470 if ((iface) == ifp) { \
2471 /* Skip bridge processing... src == dest */ \
2474 /* It's passing over or to the bridge, locally. */ \
2475 ETHER_BPF_MTAP(bifp, m); \
2476 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \
2477 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2478 /* Filter on the physical interface. */ \
2479 if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \
2480 OR_PFIL_HOOKED_INET6)) { \
2481 if (bridge_pfil(&m, NULL, ifp, \
2482 PFIL_IN) != 0 || m == NULL) { \
2486 if ((iface) != bifp) \
2487 ETHER_BPF_MTAP(iface, m); \
2491 /* We just received a packet that we sent out. */ \
2492 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2493 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2500 * Unicast. Make sure it's not for the bridge.
2502 do { GRAB_OUR_PACKETS(bifp) } while (0);
2505 * Give a chance for ifp at first priority. This will help when the
2506 * packet comes through the interface like VLAN's with the same MACs
2507 * on several interfaces from the same bridge. This also will save
2508 * some CPU cycles in case the destination interface and the input
2509 * interface (eq ifp) are the same.
2511 do { GRAB_OUR_PACKETS(ifp) } while (0);
2513 /* Now check the all bridge members. */
2514 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2515 GRAB_OUR_PACKETS(bif2->bif_ifp)
2518 #undef OR_CARP_CHECK_WE_ARE_DST
2519 #undef OR_CARP_CHECK_WE_ARE_SRC
2520 #undef OR_PFIL_HOOKED_INET6
2521 #undef GRAB_OUR_PACKETS
2523 /* Perform the bridge forwarding function. */
2524 bridge_forward(sc, bif, m);
2532 * Send a frame to all interfaces that are members of
2533 * the bridge, except for the one on which the packet
2536 * NOTE: Releases the lock on return.
2539 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2540 struct mbuf *m, int runfilt)
2542 struct bridge_iflist *dbif, *sbif;
2544 struct ifnet *dst_if;
2549 sbif = bridge_lookup_member_if(sc, src_if);
2551 /* Filter on the bridge interface before broadcasting */
2552 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2554 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2557 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2563 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2564 dst_if = dbif->bif_ifp;
2565 if (dst_if == src_if)
2568 /* Private segments can not talk to each other */
2569 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2572 if ((dbif->bif_flags & IFBIF_STP) &&
2573 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2576 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2577 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2580 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2583 if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2587 mc = m_dup(m, M_NOWAIT);
2589 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2595 * Filter on the output interface. Pass a NULL bridge interface
2596 * pointer so we do not redundantly filter on the bridge for
2597 * each interface we broadcast on.
2599 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2601 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2605 /* Keep the layer3 header aligned */
2606 i = min(mc->m_pkthdr.len, max_protohdr);
2607 mc = m_copyup(mc, i, ETHER_ALIGN);
2609 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2613 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2619 bridge_enqueue(sc, dst_if, mc);
2628 * Duplicate a packet out one or more interfaces that are in span mode,
2629 * the original mbuf is unmodified.
2632 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2634 struct bridge_iflist *bif;
2635 struct ifnet *dst_if;
2640 if (CK_LIST_EMPTY(&sc->sc_spanlist))
2643 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2644 dst_if = bif->bif_ifp;
2646 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2649 mc = m_copypacket(m, M_NOWAIT);
2651 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2655 bridge_enqueue(sc, dst_if, mc);
2662 * Add a bridge routing entry.
2665 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2666 struct bridge_iflist *bif, int setflags, uint8_t flags)
2668 struct bridge_rtnode *brt;
2671 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
2673 /* Check the source address is valid and not multicast. */
2674 if (ETHER_IS_MULTICAST(dst) ||
2675 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2676 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2679 /* 802.1p frames map to vlan 1 */
2684 * A route for this destination might already exist. If so,
2685 * update it, otherwise create a new one.
2687 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2690 /* Check again, now that we have the lock. There could have
2691 * been a race and we only want to insert this once. */
2692 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) {
2693 BRIDGE_RT_UNLOCK(sc);
2697 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2698 sc->sc_brtexceeded++;
2699 BRIDGE_RT_UNLOCK(sc);
2702 /* Check per interface address limits (if enabled) */
2703 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2704 bif->bif_addrexceeded++;
2705 BRIDGE_RT_UNLOCK(sc);
2710 * Allocate a new bridge forwarding node, and
2711 * initialize the expiration time and Ethernet
2714 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2716 BRIDGE_RT_UNLOCK(sc);
2719 brt->brt_vnet = curvnet;
2721 if (bif->bif_flags & IFBIF_STICKY)
2722 brt->brt_flags = IFBAF_STICKY;
2724 brt->brt_flags = IFBAF_DYNAMIC;
2726 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2727 brt->brt_vlan = vlan;
2729 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2730 uma_zfree(V_bridge_rtnode_zone, brt);
2731 BRIDGE_RT_UNLOCK(sc);
2737 BRIDGE_RT_UNLOCK(sc);
2740 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2741 brt->brt_dst != bif) {
2743 brt->brt_dst->bif_addrcnt--;
2745 brt->brt_dst->bif_addrcnt++;
2746 BRIDGE_RT_UNLOCK(sc);
2749 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2750 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2752 brt->brt_flags = flags;
2760 * Lookup the destination interface for an address.
2762 static struct ifnet *
2763 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2765 struct bridge_rtnode *brt;
2769 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2772 return (brt->brt_ifp);
2778 * Trim the routine table so that we have a number
2779 * of routing entries less than or equal to the
2783 bridge_rttrim(struct bridge_softc *sc)
2785 struct bridge_rtnode *brt, *nbrt;
2788 BRIDGE_RT_LOCK_ASSERT(sc);
2790 /* Make sure we actually need to do this. */
2791 if (sc->sc_brtcnt <= sc->sc_brtmax)
2794 /* Force an aging cycle; this might trim enough addresses. */
2796 if (sc->sc_brtcnt <= sc->sc_brtmax)
2799 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2800 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2801 bridge_rtnode_destroy(sc, brt);
2802 if (sc->sc_brtcnt <= sc->sc_brtmax)
2811 * Aging timer for the bridge.
2814 bridge_timer(void *arg)
2816 struct bridge_softc *sc = arg;
2818 BRIDGE_RT_LOCK_ASSERT(sc);
2820 /* Destruction of rtnodes requires a proper vnet context */
2821 CURVNET_SET(sc->sc_ifp->if_vnet);
2824 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2825 callout_reset(&sc->sc_brcallout,
2826 bridge_rtable_prune_period * hz, bridge_timer, sc);
2833 * Perform an aging cycle.
2836 bridge_rtage(struct bridge_softc *sc)
2838 struct bridge_rtnode *brt, *nbrt;
2840 BRIDGE_RT_LOCK_ASSERT(sc);
2842 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2843 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2844 if (time_uptime >= brt->brt_expire)
2845 bridge_rtnode_destroy(sc, brt);
2853 * Remove all dynamic addresses from the bridge.
2856 bridge_rtflush(struct bridge_softc *sc, int full)
2858 struct bridge_rtnode *brt, *nbrt;
2860 BRIDGE_RT_LOCK_ASSERT(sc);
2862 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2863 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2864 bridge_rtnode_destroy(sc, brt);
2871 * Remove an address from the table.
2874 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2876 struct bridge_rtnode *brt;
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 BRIDGE_RT_UNLOCK(sc);
2892 return (found ? 0 : ENOENT);
2898 * Delete routes to a speicifc member interface.
2901 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2903 struct bridge_rtnode *brt, *nbrt;
2905 BRIDGE_RT_LOCK_ASSERT(sc);
2907 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2908 if (brt->brt_ifp == ifp && (full ||
2909 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2910 bridge_rtnode_destroy(sc, brt);
2915 * bridge_rtable_init:
2917 * Initialize the route table for this bridge.
2920 bridge_rtable_init(struct bridge_softc *sc)
2924 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2925 M_DEVBUF, M_WAITOK);
2927 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2928 CK_LIST_INIT(&sc->sc_rthash[i]);
2930 sc->sc_rthash_key = arc4random();
2931 CK_LIST_INIT(&sc->sc_rtlist);
2935 * bridge_rtable_fini:
2937 * Deconstruct the route table for this bridge.
2940 bridge_rtable_fini(struct bridge_softc *sc)
2943 KASSERT(sc->sc_brtcnt == 0,
2944 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2945 free(sc->sc_rthash, M_DEVBUF);
2949 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2950 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2952 #define mix(a, b, c) \
2954 a -= b; a -= c; a ^= (c >> 13); \
2955 b -= c; b -= a; b ^= (a << 8); \
2956 c -= a; c -= b; c ^= (b >> 13); \
2957 a -= b; a -= c; a ^= (c >> 12); \
2958 b -= c; b -= a; b ^= (a << 16); \
2959 c -= a; c -= b; c ^= (b >> 5); \
2960 a -= b; a -= c; a ^= (c >> 3); \
2961 b -= c; b -= a; b ^= (a << 10); \
2962 c -= a; c -= b; c ^= (b >> 15); \
2963 } while (/*CONSTCOND*/0)
2965 static __inline uint32_t
2966 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2968 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2979 return (c & BRIDGE_RTHASH_MASK);
2985 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2989 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2990 d = ((int)a[i]) - ((int)b[i]);
2997 * bridge_rtnode_lookup:
2999 * Look up a bridge route node for the specified destination. Compare the
3000 * vlan id or if zero then just return the first match.
3002 static struct bridge_rtnode *
3003 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3005 struct bridge_rtnode *brt;
3009 BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3011 hash = bridge_rthash(sc, addr);
3012 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3013 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3014 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3024 * bridge_rtnode_insert:
3026 * Insert the specified bridge node into the route table. We
3027 * assume the entry is not already in the table.
3030 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3032 struct bridge_rtnode *lbrt;
3036 BRIDGE_RT_LOCK_ASSERT(sc);
3038 hash = bridge_rthash(sc, brt->brt_addr);
3040 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3042 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3047 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3048 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3051 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3054 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3055 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3058 lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3059 } while (lbrt != NULL);
3062 panic("bridge_rtnode_insert: impossible");
3066 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3073 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3075 struct bridge_rtnode *brt;
3077 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3079 CURVNET_SET(brt->brt_vnet);
3080 uma_zfree(V_bridge_rtnode_zone, brt);
3085 * bridge_rtnode_destroy:
3087 * Destroy a bridge rtnode.
3090 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3092 BRIDGE_RT_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;
3114 CURVNET_SET(ifp->if_vnet);
3118 * If the age is zero then flush, otherwise set all the expiry times to
3119 * age for the interface
3122 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3124 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3125 /* Cap the expiry time to 'age' */
3126 if (brt->brt_ifp == ifp &&
3127 brt->brt_expire > time_uptime + age &&
3128 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3129 brt->brt_expire = time_uptime + age;
3132 BRIDGE_RT_UNLOCK(sc);
3137 * bridge_state_change:
3139 * Callback from the bridgestp code when a port changes states.
3142 bridge_state_change(struct ifnet *ifp, int state)
3144 struct bridge_softc *sc = ifp->if_bridge;
3145 static const char *stpstates[] = {
3154 CURVNET_SET(ifp->if_vnet);
3156 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3157 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3162 * Send bridge packets through pfil if they are one of the types pfil can deal
3163 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3164 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3168 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3170 int snap, error, i, hlen;
3171 struct ether_header *eh1, eh2;
3174 u_int16_t ether_type;
3178 error = -1; /* Default error if not error == 0 */
3181 /* we may return with the IP fields swapped, ensure its not shared */
3182 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3185 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3186 return (0); /* filtering is disabled */
3188 i = min((*mp)->m_pkthdr.len, max_protohdr);
3189 if ((*mp)->m_len < i) {
3190 *mp = m_pullup(*mp, i);
3192 printf("%s: m_pullup failed\n", __func__);
3197 eh1 = mtod(*mp, struct ether_header *);
3198 ether_type = ntohs(eh1->ether_type);
3201 * Check for SNAP/LLC.
3203 if (ether_type < ETHERMTU) {
3204 struct llc *llc2 = (struct llc *)(eh1 + 1);
3206 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3207 llc2->llc_dsap == LLC_SNAP_LSAP &&
3208 llc2->llc_ssap == LLC_SNAP_LSAP &&
3209 llc2->llc_control == LLC_UI) {
3210 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3216 * If we're trying to filter bridge traffic, don't look at anything
3217 * other than IP and ARP traffic. If the filter doesn't understand
3218 * IPv6, don't allow IPv6 through the bridge either. This is lame
3219 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3220 * but of course we don't have an AppleTalk filter to begin with.
3221 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3224 switch (ether_type) {
3226 case ETHERTYPE_REVARP:
3227 if (V_pfil_ipfw_arp == 0)
3228 return (0); /* Automatically pass */
3233 case ETHERTYPE_IPV6:
3238 * Check to see if the user wants to pass non-ip
3239 * packets, these will not be checked by pfil(9) and
3240 * passed unconditionally so the default is to drop.
3246 /* Run the packet through pfil before stripping link headers */
3247 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3248 dir == PFIL_OUT && ifp != NULL) {
3249 switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) {
3257 /* Strip off the Ethernet header and keep a copy. */
3258 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3259 m_adj(*mp, ETHER_HDR_LEN);
3261 /* Strip off snap header, if present */
3263 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3264 m_adj(*mp, sizeof(struct llc));
3268 * Check the IP header for alignment and errors
3270 if (dir == PFIL_IN) {
3271 switch (ether_type) {
3273 error = bridge_ip_checkbasic(mp);
3276 case ETHERTYPE_IPV6:
3277 error = bridge_ip6_checkbasic(mp);
3290 * Run the packet through pfil
3293 switch (ether_type) {
3296 * Run pfil on the member interface and the bridge, both can
3297 * be skipped by clearing pfil_member or pfil_bridge.
3300 * in_if -> bridge_if -> out_if
3302 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3303 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3307 if (V_pfil_member && ifp != NULL && (rv =
3308 pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) !=
3312 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3313 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3317 /* check if we need to fragment the packet */
3318 /* bridge_fragment generates a mbuf chain of packets */
3319 /* that already include eth headers */
3320 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3321 i = (*mp)->m_pkthdr.len;
3322 if (i > ifp->if_mtu) {
3323 error = bridge_fragment(ifp, mp, &eh2, snap,
3329 /* Recalculate the ip checksum. */
3330 ip = mtod(*mp, struct ip *);
3331 hlen = ip->ip_hl << 2;
3332 if (hlen < sizeof(struct ip))
3334 if (hlen > (*mp)->m_len) {
3335 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3337 ip = mtod(*mp, struct ip *);
3342 if (hlen == sizeof(struct ip))
3343 ip->ip_sum = in_cksum_hdr(ip);
3345 ip->ip_sum = in_cksum(*mp, hlen);
3349 case ETHERTYPE_IPV6:
3350 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3351 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3355 if (V_pfil_member && ifp != NULL && (rv =
3356 pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) !=
3360 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3361 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3380 * Finally, put everything back the way it was and return
3383 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3386 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3389 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3392 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3403 * Perform basic checks on header size since
3404 * pfil assumes ip_input has already processed
3405 * it for it. Cut-and-pasted from ip_input.c.
3406 * Given how simple the IPv6 version is,
3407 * does the IPv4 version really need to be
3410 * XXX Should we update ipstat here, or not?
3411 * XXX Right now we update ipstat but not
3415 bridge_ip_checkbasic(struct mbuf **mp)
3417 struct mbuf *m = *mp;
3425 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3426 if ((m = m_copyup(m, sizeof(struct ip),
3427 (max_linkhdr + 3) & ~3)) == NULL) {
3428 /* XXXJRT new stat, please */
3429 KMOD_IPSTAT_INC(ips_toosmall);
3432 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3433 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3434 KMOD_IPSTAT_INC(ips_toosmall);
3438 ip = mtod(m, struct ip *);
3439 if (ip == NULL) goto bad;
3441 if (ip->ip_v != IPVERSION) {
3442 KMOD_IPSTAT_INC(ips_badvers);
3445 hlen = ip->ip_hl << 2;
3446 if (hlen < sizeof(struct ip)) { /* minimum header length */
3447 KMOD_IPSTAT_INC(ips_badhlen);
3450 if (hlen > m->m_len) {
3451 if ((m = m_pullup(m, hlen)) == NULL) {
3452 KMOD_IPSTAT_INC(ips_badhlen);
3455 ip = mtod(m, struct ip *);
3456 if (ip == NULL) goto bad;
3459 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3460 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3462 if (hlen == sizeof(struct ip)) {
3463 sum = in_cksum_hdr(ip);
3465 sum = in_cksum(m, hlen);
3469 KMOD_IPSTAT_INC(ips_badsum);
3473 /* Retrieve the packet length. */
3474 len = ntohs(ip->ip_len);
3477 * Check for additional length bogosity
3480 KMOD_IPSTAT_INC(ips_badlen);
3485 * Check that the amount of data in the buffers
3486 * is as at least much as the IP header would have us expect.
3487 * Drop packet if shorter than we expect.
3489 if (m->m_pkthdr.len < len) {
3490 KMOD_IPSTAT_INC(ips_tooshort);
3494 /* Checks out, proceed */
3505 * Same as above, but for IPv6.
3506 * Cut-and-pasted from ip6_input.c.
3507 * XXX Should we update ip6stat, or not?
3510 bridge_ip6_checkbasic(struct mbuf **mp)
3512 struct mbuf *m = *mp;
3513 struct ip6_hdr *ip6;
3516 * If the IPv6 header is not aligned, slurp it up into a new
3517 * mbuf with space for link headers, in the event we forward
3518 * it. Otherwise, if it is aligned, make sure the entire base
3519 * IPv6 header is in the first mbuf of the chain.
3521 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3522 struct ifnet *inifp = m->m_pkthdr.rcvif;
3523 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3524 (max_linkhdr + 3) & ~3)) == NULL) {
3525 /* XXXJRT new stat, please */
3526 IP6STAT_INC(ip6s_toosmall);
3527 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3530 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3531 struct ifnet *inifp = m->m_pkthdr.rcvif;
3532 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3533 IP6STAT_INC(ip6s_toosmall);
3534 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3539 ip6 = mtod(m, struct ip6_hdr *);
3541 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3542 IP6STAT_INC(ip6s_badvers);
3543 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3547 /* Checks out, proceed */
3560 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3563 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3564 int snap, struct llc *llc)
3566 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3570 if (m->m_len < sizeof(struct ip) &&
3571 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3573 ip = mtod(m, struct ip *);
3575 m->m_pkthdr.csum_flags |= CSUM_IP;
3576 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3581 * Walk the chain and re-add the Ethernet header for
3584 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3585 nextpkt = mcur->m_nextpkt;
3586 mcur->m_nextpkt = NULL;
3588 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3592 mprev->m_nextpkt = nextpkt;
3595 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3598 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3602 mprev->m_nextpkt = nextpkt;
3605 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3608 * The previous two M_PREPEND could have inserted one or two
3609 * mbufs in front so we have to update the previous packet's
3612 mcur->m_nextpkt = nextpkt;
3614 mprev->m_nextpkt = mcur;
3616 /* The first mbuf in the original chain needs to be
3623 KMOD_IPSTAT_INC(ips_fragmented);
3627 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3628 m = mcur->m_nextpkt;
3635 bridge_linkstate(struct ifnet *ifp)
3637 struct bridge_softc *sc = ifp->if_bridge;
3638 struct bridge_iflist *bif;
3639 struct epoch_tracker et;
3641 NET_EPOCH_ENTER(et);
3643 bif = bridge_lookup_member_if(sc, ifp);
3648 bridge_linkcheck(sc);
3650 bstp_linkstate(&bif->bif_stp);
3656 bridge_linkcheck(struct bridge_softc *sc)
3658 struct bridge_iflist *bif;
3659 int new_link, hasls;
3661 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3663 new_link = LINK_STATE_DOWN;
3665 /* Our link is considered up if at least one of our ports is active */
3666 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3667 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3669 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3670 new_link = LINK_STATE_UP;
3674 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3675 /* If no interfaces support link-state then we default to up */
3676 new_link = LINK_STATE_UP;
3678 if_link_state_change(sc->sc_ifp, new_link);