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 */
238 struct epoch_context bif_epoch_ctx;
244 struct bridge_rtnode {
245 CK_LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
246 CK_LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
247 struct bridge_iflist *brt_dst; /* destination if */
248 unsigned long brt_expire; /* expiration time */
249 uint8_t brt_flags; /* address flags */
250 uint8_t brt_addr[ETHER_ADDR_LEN];
251 uint16_t brt_vlan; /* vlan id */
253 struct vnet *brt_vnet;
254 struct epoch_context brt_epoch_ctx;
256 #define brt_ifp brt_dst->bif_ifp
259 * Software state for each bridge.
261 struct bridge_softc {
262 struct ifnet *sc_ifp; /* make this an interface */
263 LIST_ENTRY(bridge_softc) sc_list;
265 struct mtx sc_rt_mtx;
266 uint32_t sc_brtmax; /* max # of addresses */
267 uint32_t sc_brtcnt; /* cur. # of addresses */
268 uint32_t sc_brttimeout; /* rt timeout in seconds */
269 struct callout sc_brcallout; /* bridge callout */
270 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
271 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
272 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
273 uint32_t sc_rthash_key; /* key for hash */
274 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
275 struct bstp_state sc_stp; /* STP state */
276 uint32_t sc_brtexceeded; /* # of cache drops */
277 struct ifnet *sc_ifaddr; /* member mac copied from */
278 struct ether_addr sc_defaddr; /* Default MAC address */
280 struct epoch_context sc_epoch_ctx;
283 VNET_DEFINE_STATIC(struct sx, bridge_list_sx);
284 #define V_bridge_list_sx VNET(bridge_list_sx)
285 static eventhandler_tag bridge_detach_cookie;
287 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
289 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
290 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone)
292 static int bridge_clone_create(struct if_clone *, int, caddr_t);
293 static void bridge_clone_destroy(struct ifnet *);
295 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
296 static void bridge_mutecaps(struct bridge_softc *);
297 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
299 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
300 static void bridge_init(void *);
301 static void bridge_dummynet(struct mbuf *, struct ifnet *);
302 static void bridge_stop(struct ifnet *, int);
303 static int bridge_transmit(struct ifnet *, struct mbuf *);
305 static void bridge_altq_start(if_t);
306 static int bridge_altq_transmit(if_t, struct mbuf *);
308 static void bridge_qflush(struct ifnet *);
309 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
310 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
312 static int bridge_enqueue(struct bridge_softc *, struct ifnet *,
314 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
316 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
319 static void bridge_timer(void *);
321 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
323 static void bridge_span(struct bridge_softc *, struct mbuf *);
325 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
326 uint16_t, struct bridge_iflist *, int, uint8_t);
327 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
329 static void bridge_rttrim(struct bridge_softc *);
330 static void bridge_rtage(struct bridge_softc *);
331 static void bridge_rtflush(struct bridge_softc *, int);
332 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
335 static void bridge_rtable_init(struct bridge_softc *);
336 static void bridge_rtable_fini(struct bridge_softc *);
338 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
339 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
340 const uint8_t *, uint16_t);
341 static int bridge_rtnode_insert(struct bridge_softc *,
342 struct bridge_rtnode *);
343 static void bridge_rtnode_destroy(struct bridge_softc *,
344 struct bridge_rtnode *);
345 static void bridge_rtable_expire(struct ifnet *, int);
346 static void bridge_state_change(struct ifnet *, int);
348 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
350 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
352 static void bridge_delete_member(struct bridge_softc *,
353 struct bridge_iflist *, int);
354 static void bridge_delete_span(struct bridge_softc *,
355 struct bridge_iflist *);
357 static int bridge_ioctl_add(struct bridge_softc *, void *);
358 static int bridge_ioctl_del(struct bridge_softc *, void *);
359 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
360 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
361 static int bridge_ioctl_scache(struct bridge_softc *, void *);
362 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
363 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
364 static int bridge_ioctl_rts(struct bridge_softc *, void *);
365 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
366 static int bridge_ioctl_sto(struct bridge_softc *, void *);
367 static int bridge_ioctl_gto(struct bridge_softc *, void *);
368 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
369 static int bridge_ioctl_flush(struct bridge_softc *, void *);
370 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
371 static int bridge_ioctl_spri(struct bridge_softc *, void *);
372 static int bridge_ioctl_ght(struct bridge_softc *, void *);
373 static int bridge_ioctl_sht(struct bridge_softc *, void *);
374 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
375 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
376 static int bridge_ioctl_gma(struct bridge_softc *, void *);
377 static int bridge_ioctl_sma(struct bridge_softc *, void *);
378 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
379 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
380 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
381 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
382 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
383 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
384 static int bridge_ioctl_grte(struct bridge_softc *, void *);
385 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
386 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
387 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
388 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
390 static int bridge_ip_checkbasic(struct mbuf **mp);
392 static int bridge_ip6_checkbasic(struct mbuf **mp);
394 static int bridge_fragment(struct ifnet *, struct mbuf **mp,
395 struct ether_header *, int, struct llc *);
396 static void bridge_linkstate(struct ifnet *ifp);
397 static void bridge_linkcheck(struct bridge_softc *sc);
400 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
401 #define VLANTAGOF(_m) \
402 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
404 static struct bstp_cb_ops bridge_ops = {
405 .bcb_state = bridge_state_change,
406 .bcb_rtage = bridge_rtable_expire
409 SYSCTL_DECL(_net_link);
410 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
412 /* only pass IP[46] packets when pfil is enabled */
413 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
414 #define V_pfil_onlyip VNET(pfil_onlyip)
415 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
416 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
417 "Only pass IP packets when pfil is enabled");
419 /* run pfil hooks on the bridge interface */
420 VNET_DEFINE_STATIC(int, pfil_bridge) = 1;
421 #define V_pfil_bridge VNET(pfil_bridge)
422 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
423 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
424 "Packet filter on the bridge interface");
426 /* layer2 filter with ipfw */
427 VNET_DEFINE_STATIC(int, pfil_ipfw);
428 #define V_pfil_ipfw VNET(pfil_ipfw)
430 /* layer2 ARP filter with ipfw */
431 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
432 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
433 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
434 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
435 "Filter ARP packets through IPFW layer2");
437 /* run pfil hooks on the member interface */
438 VNET_DEFINE_STATIC(int, pfil_member) = 1;
439 #define V_pfil_member VNET(pfil_member)
440 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
441 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
442 "Packet filter on the member interface");
444 /* run pfil hooks on the physical interface for locally destined packets */
445 VNET_DEFINE_STATIC(int, pfil_local_phys);
446 #define V_pfil_local_phys VNET(pfil_local_phys)
447 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
448 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
449 "Packet filter on the physical interface for locally destined packets");
451 /* log STP state changes */
452 VNET_DEFINE_STATIC(int, log_stp);
453 #define V_log_stp VNET(log_stp)
454 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
455 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
456 "Log STP state changes");
458 /* share MAC with first bridge member */
459 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
460 #define V_bridge_inherit_mac VNET(bridge_inherit_mac)
461 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
462 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
463 "Inherit MAC address from the first bridge member");
465 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
466 #define V_allow_llz_overlap VNET(allow_llz_overlap)
467 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
468 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
469 "Allow overlap of link-local scope "
470 "zones of a bridge interface and the member interfaces");
472 struct bridge_control {
473 int (*bc_func)(struct bridge_softc *, void *);
478 #define BC_F_COPYIN 0x01 /* copy arguments in */
479 #define BC_F_COPYOUT 0x02 /* copy arguments out */
480 #define BC_F_SUSER 0x04 /* do super-user check */
482 const struct bridge_control bridge_control_table[] = {
483 { bridge_ioctl_add, sizeof(struct ifbreq),
484 BC_F_COPYIN|BC_F_SUSER },
485 { bridge_ioctl_del, sizeof(struct ifbreq),
486 BC_F_COPYIN|BC_F_SUSER },
488 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
489 BC_F_COPYIN|BC_F_COPYOUT },
490 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
491 BC_F_COPYIN|BC_F_SUSER },
493 { bridge_ioctl_scache, sizeof(struct ifbrparam),
494 BC_F_COPYIN|BC_F_SUSER },
495 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
498 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
499 BC_F_COPYIN|BC_F_COPYOUT },
500 { bridge_ioctl_rts, sizeof(struct ifbaconf),
501 BC_F_COPYIN|BC_F_COPYOUT },
503 { bridge_ioctl_saddr, sizeof(struct ifbareq),
504 BC_F_COPYIN|BC_F_SUSER },
506 { bridge_ioctl_sto, sizeof(struct ifbrparam),
507 BC_F_COPYIN|BC_F_SUSER },
508 { bridge_ioctl_gto, sizeof(struct ifbrparam),
511 { bridge_ioctl_daddr, sizeof(struct ifbareq),
512 BC_F_COPYIN|BC_F_SUSER },
514 { bridge_ioctl_flush, sizeof(struct ifbreq),
515 BC_F_COPYIN|BC_F_SUSER },
517 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
519 { bridge_ioctl_spri, sizeof(struct ifbrparam),
520 BC_F_COPYIN|BC_F_SUSER },
522 { bridge_ioctl_ght, sizeof(struct ifbrparam),
524 { bridge_ioctl_sht, sizeof(struct ifbrparam),
525 BC_F_COPYIN|BC_F_SUSER },
527 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
529 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
530 BC_F_COPYIN|BC_F_SUSER },
532 { bridge_ioctl_gma, sizeof(struct ifbrparam),
534 { bridge_ioctl_sma, sizeof(struct ifbrparam),
535 BC_F_COPYIN|BC_F_SUSER },
537 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
538 BC_F_COPYIN|BC_F_SUSER },
540 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
541 BC_F_COPYIN|BC_F_SUSER },
543 { bridge_ioctl_addspan, sizeof(struct ifbreq),
544 BC_F_COPYIN|BC_F_SUSER },
545 { bridge_ioctl_delspan, sizeof(struct ifbreq),
546 BC_F_COPYIN|BC_F_SUSER },
548 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
551 { bridge_ioctl_grte, sizeof(struct ifbrparam),
554 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
555 BC_F_COPYIN|BC_F_COPYOUT },
557 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
558 BC_F_COPYIN|BC_F_SUSER },
560 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
561 BC_F_COPYIN|BC_F_SUSER },
563 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
564 BC_F_COPYIN|BC_F_SUSER },
567 const int bridge_control_table_size = nitems(bridge_control_table);
569 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
570 #define V_bridge_list VNET(bridge_list)
571 #define BRIDGE_LIST_LOCK_INIT(x) sx_init(&V_bridge_list_sx, \
573 #define BRIDGE_LIST_LOCK_DESTROY(x) sx_destroy(&V_bridge_list_sx)
574 #define BRIDGE_LIST_LOCK(x) sx_xlock(&V_bridge_list_sx)
575 #define BRIDGE_LIST_UNLOCK(x) sx_xunlock(&V_bridge_list_sx)
577 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
578 #define V_bridge_cloner VNET(bridge_cloner)
580 static const char bridge_name[] = "bridge";
583 vnet_bridge_init(const void *unused __unused)
586 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
587 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
589 BRIDGE_LIST_LOCK_INIT();
590 LIST_INIT(&V_bridge_list);
591 V_bridge_cloner = if_clone_simple(bridge_name,
592 bridge_clone_create, bridge_clone_destroy, 0);
594 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
595 vnet_bridge_init, NULL);
598 vnet_bridge_uninit(const void *unused __unused)
601 if_clone_detach(V_bridge_cloner);
602 V_bridge_cloner = NULL;
603 BRIDGE_LIST_LOCK_DESTROY();
605 /* Before we can destroy the uma zone, because there are callbacks that
607 epoch_drain_callbacks(net_epoch_preempt);
609 uma_zdestroy(V_bridge_rtnode_zone);
611 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
612 vnet_bridge_uninit, NULL);
615 bridge_modevent(module_t mod, int type, void *data)
620 bridge_dn_p = bridge_dummynet;
621 bridge_detach_cookie = EVENTHANDLER_REGISTER(
622 ifnet_departure_event, bridge_ifdetach, NULL,
623 EVENTHANDLER_PRI_ANY);
626 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
627 bridge_detach_cookie);
636 static moduledata_t bridge_mod = {
642 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
643 MODULE_VERSION(if_bridge, 1);
644 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
647 * handler for net.link.bridge.ipfw
650 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
652 int enable = V_pfil_ipfw;
655 error = sysctl_handle_int(oidp, &enable, 0, req);
658 if (enable != V_pfil_ipfw) {
659 V_pfil_ipfw = enable;
662 * Disable pfil so that ipfw doesnt run twice, if the user
663 * really wants both then they can re-enable pfil_bridge and/or
664 * pfil_member. Also allow non-ip packets as ipfw can filter by
676 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
677 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET,
678 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
679 "Layer2 filter with IPFW");
683 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg)
685 struct bridge_softc *sc = ifp->if_softc;
686 struct bridge_iflist *bif;
690 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
691 bridge_delete_member(sc, bif, 0);
693 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
694 bridge_delete_span(sc, bif);
699 ether_reassign(ifp, newvnet, arg);
704 * bridge_clone_create:
706 * Create a new bridge instance.
709 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
711 struct bridge_softc *sc, *sc2;
712 struct ifnet *bifp, *ifp;
714 unsigned long hostid;
716 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
717 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
723 BRIDGE_LOCK_INIT(sc);
724 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
725 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
727 /* Initialize our routing table. */
728 bridge_rtable_init(sc);
730 callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
732 CK_LIST_INIT(&sc->sc_iflist);
733 CK_LIST_INIT(&sc->sc_spanlist);
736 if_initname(ifp, bridge_name, unit);
737 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
738 ifp->if_ioctl = bridge_ioctl;
740 ifp->if_start = bridge_altq_start;
741 ifp->if_transmit = bridge_altq_transmit;
742 IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
743 ifp->if_snd.ifq_drv_maxlen = 0;
744 IFQ_SET_READY(&ifp->if_snd);
746 ifp->if_transmit = bridge_transmit;
748 ifp->if_qflush = bridge_qflush;
749 ifp->if_init = bridge_init;
750 ifp->if_type = IFT_BRIDGE;
753 * Generate an ethernet address with a locally administered address.
755 * Since we are using random ethernet addresses for the bridge, it is
756 * possible that we might have address collisions, so make sure that
757 * this hardware address isn't already in use on another bridge.
758 * The first try uses the hostid and falls back to arc4rand().
761 getcredhostid(curthread->td_ucred, &hostid);
763 if (fb || hostid == 0) {
764 ether_gen_addr(ifp, &sc->sc_defaddr);
766 sc->sc_defaddr.octet[0] = 0x2;
767 sc->sc_defaddr.octet[1] = (hostid >> 24) & 0xff;
768 sc->sc_defaddr.octet[2] = (hostid >> 16) & 0xff;
769 sc->sc_defaddr.octet[3] = (hostid >> 8 ) & 0xff;
770 sc->sc_defaddr.octet[4] = hostid & 0xff;
771 sc->sc_defaddr.octet[5] = ifp->if_dunit & 0xff;
777 LIST_FOREACH(sc2, &V_bridge_list, sc_list) {
779 if (memcmp(sc->sc_defaddr.octet,
780 IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
785 BRIDGE_LIST_UNLOCK();
786 } while (retry == 1);
788 bstp_attach(&sc->sc_stp, &bridge_ops);
789 ether_ifattach(ifp, sc->sc_defaddr.octet);
790 /* Now undo some of the damage... */
791 ifp->if_baudrate = 0;
792 ifp->if_type = IFT_BRIDGE;
794 ifp->if_reassign = bridge_reassign;
798 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
799 BRIDGE_LIST_UNLOCK();
805 bridge_clone_destroy_cb(struct epoch_context *ctx)
807 struct bridge_softc *sc;
809 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
811 BRIDGE_LOCK_DESTROY(sc);
816 * bridge_clone_destroy:
818 * Destroy a bridge instance.
821 bridge_clone_destroy(struct ifnet *ifp)
823 struct bridge_softc *sc = ifp->if_softc;
824 struct bridge_iflist *bif;
825 struct epoch_tracker et;
830 ifp->if_flags &= ~IFF_UP;
832 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
833 bridge_delete_member(sc, bif, 0);
835 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
836 bridge_delete_span(sc, bif);
839 /* Tear down the routing table. */
840 bridge_rtable_fini(sc);
844 NET_EPOCH_ENTER_ET(et);
846 callout_drain(&sc->sc_brcallout);
849 LIST_REMOVE(sc, sc_list);
850 BRIDGE_LIST_UNLOCK();
852 bstp_detach(&sc->sc_stp);
854 IFQ_PURGE(&ifp->if_snd);
856 NET_EPOCH_EXIT_ET(et);
861 epoch_call(net_epoch_preempt, &sc->sc_epoch_ctx, bridge_clone_destroy_cb);
867 * Handle a control request from the operator.
870 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
872 struct bridge_softc *sc = ifp->if_softc;
873 struct ifreq *ifr = (struct ifreq *)data;
874 struct bridge_iflist *bif;
875 struct thread *td = curthread;
877 struct ifbreq ifbreq;
878 struct ifbifconf ifbifconf;
879 struct ifbareq ifbareq;
880 struct ifbaconf ifbaconf;
881 struct ifbrparam ifbrparam;
882 struct ifbropreq ifbropreq;
884 struct ifdrv *ifd = (struct ifdrv *) data;
885 const struct bridge_control *bc;
886 int error = 0, oldmtu;
898 if (ifd->ifd_cmd >= bridge_control_table_size) {
902 bc = &bridge_control_table[ifd->ifd_cmd];
904 if (cmd == SIOCGDRVSPEC &&
905 (bc->bc_flags & BC_F_COPYOUT) == 0) {
909 else if (cmd == SIOCSDRVSPEC &&
910 (bc->bc_flags & BC_F_COPYOUT) != 0) {
915 if (bc->bc_flags & BC_F_SUSER) {
916 error = priv_check(td, PRIV_NET_BRIDGE);
921 if (ifd->ifd_len != bc->bc_argsize ||
922 ifd->ifd_len > sizeof(args)) {
927 bzero(&args, sizeof(args));
928 if (bc->bc_flags & BC_F_COPYIN) {
929 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
934 oldmtu = ifp->if_mtu;
935 error = (*bc->bc_func)(sc, &args);
940 * Bridge MTU may change during addition of the first port.
941 * If it did, do network layer specific procedure.
943 if (ifp->if_mtu != oldmtu) {
950 if (bc->bc_flags & BC_F_COPYOUT)
951 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
956 if (!(ifp->if_flags & IFF_UP) &&
957 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
959 * If interface is marked down and it is running,
960 * then stop and disable it.
963 } else if ((ifp->if_flags & IFF_UP) &&
964 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
966 * If interface is marked up and it is stopped, then
976 if (ifr->ifr_mtu < 576) {
980 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
981 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
984 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
985 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
986 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
987 " != %d\n", sc->sc_ifp->if_xname,
988 bif->bif_ifp->if_mtu,
989 bif->bif_ifp->if_xname, ifr->ifr_mtu);
995 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
999 * drop the lock as ether_ioctl() will call bridge_start() and
1000 * cause the lock to be recursed.
1003 error = ether_ioctl(ifp, cmd, data);
1016 * Clear or restore unwanted capabilities on the member interface
1019 bridge_mutecaps(struct bridge_softc *sc)
1021 struct bridge_iflist *bif;
1024 BRIDGE_LOCK_ASSERT(sc);
1026 /* Initial bitmask of capabilities to test */
1027 mask = BRIDGE_IFCAPS_MASK;
1029 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1030 /* Every member must support it or its disabled */
1031 mask &= bif->bif_savedcaps;
1034 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1035 enabled = bif->bif_ifp->if_capenable;
1036 enabled &= ~BRIDGE_IFCAPS_STRIP;
1037 /* strip off mask bits and enable them again if allowed */
1038 enabled &= ~BRIDGE_IFCAPS_MASK;
1040 bridge_set_ifcap(sc, bif, enabled);
1045 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1047 struct ifnet *ifp = bif->bif_ifp;
1049 int error, mask, stuck;
1051 bzero(&ifr, sizeof(ifr));
1052 ifr.ifr_reqcap = set;
1054 if (ifp->if_capenable != set) {
1055 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1057 if_printf(sc->sc_ifp,
1058 "error setting capabilities on %s: %d\n",
1059 ifp->if_xname, error);
1060 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1061 stuck = ifp->if_capenable & mask & ~set;
1063 if_printf(sc->sc_ifp,
1064 "can't disable some capabilities on %s: 0x%x\n",
1065 ifp->if_xname, stuck);
1070 * bridge_lookup_member:
1072 * Lookup a bridge member interface.
1074 static struct bridge_iflist *
1075 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1077 struct bridge_iflist *bif;
1080 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1082 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1084 if (strcmp(ifp->if_xname, name) == 0)
1092 * bridge_lookup_member_if:
1094 * Lookup a bridge member interface by ifnet*.
1096 static struct bridge_iflist *
1097 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1099 struct bridge_iflist *bif;
1101 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1103 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1104 if (bif->bif_ifp == member_ifp)
1112 bridge_delete_member_cb(struct epoch_context *ctx)
1114 struct bridge_iflist *bif;
1116 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1118 free(bif, M_DEVBUF);
1122 * bridge_delete_member:
1124 * Delete the specified member interface.
1127 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1130 struct ifnet *ifs = bif->bif_ifp;
1131 struct ifnet *fif = NULL;
1132 struct bridge_iflist *bifl;
1134 BRIDGE_LOCK_ASSERT(sc);
1136 if (bif->bif_flags & IFBIF_STP)
1137 bstp_disable(&bif->bif_stp);
1139 ifs->if_bridge = NULL;
1140 CK_LIST_REMOVE(bif, bif_next);
1143 * If removing the interface that gave the bridge its mac address, set
1144 * the mac address of the bridge to the address of the next member, or
1145 * to its default address if no members are left.
1147 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1148 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1149 bcopy(&sc->sc_defaddr,
1150 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1151 sc->sc_ifaddr = NULL;
1153 bifl = CK_LIST_FIRST(&sc->sc_iflist);
1154 fif = bifl->bif_ifp;
1155 bcopy(IF_LLADDR(fif),
1156 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1157 sc->sc_ifaddr = fif;
1159 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1162 bridge_linkcheck(sc);
1163 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1165 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1166 BRIDGE_RT_UNLOCK(sc);
1167 KASSERT(bif->bif_addrcnt == 0,
1168 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1170 ifs->if_bridge_output = NULL;
1171 ifs->if_bridge_input = NULL;
1172 ifs->if_bridge_linkstate = NULL;
1174 switch (ifs->if_type) {
1178 * Take the interface out of promiscuous mode, but only
1179 * if it was promiscuous in the first place. It might
1180 * not be if we're in the bridge_ioctl_add() error path.
1182 if (ifs->if_flags & IFF_PROMISC)
1183 (void) ifpromisc(ifs, 0);
1191 panic("bridge_delete_member: impossible");
1195 /* reneable any interface capabilities */
1196 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1198 bstp_destroy(&bif->bif_stp); /* prepare to free */
1200 epoch_call(net_epoch_preempt, &bif->bif_epoch_ctx,
1201 bridge_delete_member_cb);
1205 * bridge_delete_span:
1207 * Delete the specified span interface.
1210 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1212 BRIDGE_LOCK_ASSERT(sc);
1214 KASSERT(bif->bif_ifp->if_bridge == NULL,
1215 ("%s: not a span interface", __func__));
1217 CK_LIST_REMOVE(bif, bif_next);
1219 epoch_call(net_epoch_preempt, &bif->bif_epoch_ctx,
1220 bridge_delete_member_cb);
1224 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1226 struct ifbreq *req = arg;
1227 struct bridge_iflist *bif = NULL;
1231 ifs = ifunit(req->ifbr_ifsname);
1234 if (ifs->if_ioctl == NULL) /* must be supported */
1237 /* If it's in the span list, it can't be a member. */
1238 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1239 if (ifs == bif->bif_ifp)
1242 if (ifs->if_bridge == sc)
1245 if (ifs->if_bridge != NULL)
1248 switch (ifs->if_type) {
1252 /* permitted interface types */
1260 * Two valid inet6 addresses with link-local scope must not be
1261 * on the parent interface and the member interfaces at the
1262 * same time. This restriction is needed to prevent violation
1263 * of link-local scope zone. Attempts to add a member
1264 * interface which has inet6 addresses when the parent has
1265 * inet6 triggers removal of all inet6 addresses on the member
1269 /* Check if the parent interface has a link-local scope addr. */
1270 if (V_allow_llz_overlap == 0 &&
1271 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1273 * If any, remove all inet6 addresses from the member
1276 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1277 if (in6ifa_llaonifp(bif->bif_ifp)) {
1278 in6_ifdetach(bif->bif_ifp);
1279 if_printf(sc->sc_ifp,
1280 "IPv6 addresses on %s have been removed "
1281 "before adding it as a member to prevent "
1282 "IPv6 address scope violation.\n",
1283 bif->bif_ifp->if_xname);
1286 if (in6ifa_llaonifp(ifs)) {
1288 if_printf(sc->sc_ifp,
1289 "IPv6 addresses on %s have been removed "
1290 "before adding it as a member to prevent "
1291 "IPv6 address scope violation.\n",
1296 /* Allow the first Ethernet member to define the MTU */
1297 if (CK_LIST_EMPTY(&sc->sc_iflist))
1298 sc->sc_ifp->if_mtu = ifs->if_mtu;
1299 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1300 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1301 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1305 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1310 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1311 bif->bif_savedcaps = ifs->if_capenable;
1314 * Assign the interface's MAC address to the bridge if it's the first
1315 * member and the MAC address of the bridge has not been changed from
1316 * the default randomly generated one.
1318 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1319 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1320 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1321 sc->sc_ifaddr = ifs;
1322 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1325 ifs->if_bridge = sc;
1326 ifs->if_bridge_output = bridge_output;
1327 ifs->if_bridge_input = bridge_input;
1328 ifs->if_bridge_linkstate = bridge_linkstate;
1329 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1331 * XXX: XLOCK HERE!?!
1333 * NOTE: insert_***HEAD*** should be safe for the traversals.
1335 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1337 /* Set interface capabilities to the intersection set of all members */
1338 bridge_mutecaps(sc);
1339 bridge_linkcheck(sc);
1341 /* Place the interface into promiscuous mode */
1342 switch (ifs->if_type) {
1345 error = ifpromisc(ifs, 1);
1350 bridge_delete_member(sc, bif, 0);
1355 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1357 struct ifbreq *req = arg;
1358 struct bridge_iflist *bif;
1360 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1364 bridge_delete_member(sc, bif, 0);
1370 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1372 struct ifbreq *req = arg;
1373 struct bridge_iflist *bif;
1374 struct bstp_port *bp;
1376 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1381 req->ifbr_ifsflags = bif->bif_flags;
1382 req->ifbr_state = bp->bp_state;
1383 req->ifbr_priority = bp->bp_priority;
1384 req->ifbr_path_cost = bp->bp_path_cost;
1385 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1386 req->ifbr_proto = bp->bp_protover;
1387 req->ifbr_role = bp->bp_role;
1388 req->ifbr_stpflags = bp->bp_flags;
1389 req->ifbr_addrcnt = bif->bif_addrcnt;
1390 req->ifbr_addrmax = bif->bif_addrmax;
1391 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1393 /* Copy STP state options as flags */
1394 if (bp->bp_operedge)
1395 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1396 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1397 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1398 if (bp->bp_ptp_link)
1399 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1400 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1401 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1402 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1403 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1404 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1405 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1410 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1412 struct ifbreq *req = arg;
1413 struct bridge_iflist *bif;
1414 struct bstp_port *bp;
1417 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1422 if (req->ifbr_ifsflags & IFBIF_SPAN)
1423 /* SPAN is readonly */
1428 if (req->ifbr_ifsflags & IFBIF_STP) {
1429 if ((bif->bif_flags & IFBIF_STP) == 0) {
1430 error = bstp_enable(&bif->bif_stp);
1437 if ((bif->bif_flags & IFBIF_STP) != 0)
1438 bstp_disable(&bif->bif_stp);
1441 /* Pass on STP flags */
1442 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1443 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1444 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1445 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1447 /* Save the bits relating to the bridge */
1448 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1456 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1458 struct ifbrparam *param = arg;
1460 sc->sc_brtmax = param->ifbrp_csize;
1467 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1469 struct ifbrparam *param = arg;
1471 param->ifbrp_csize = sc->sc_brtmax;
1477 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1479 struct ifbifconf *bifc = arg;
1480 struct bridge_iflist *bif;
1483 int count, buflen, len, error = 0;
1486 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1488 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1491 buflen = sizeof(breq) * count;
1492 if (bifc->ifbic_len == 0) {
1493 bifc->ifbic_len = buflen;
1496 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1502 len = min(bifc->ifbic_len, buflen);
1503 bzero(&breq, sizeof(breq));
1504 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1505 if (len < sizeof(breq))
1508 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1509 sizeof(breq.ifbr_ifsname));
1510 /* Fill in the ifbreq structure */
1511 error = bridge_ioctl_gifflags(sc, &breq);
1514 memcpy(buf, &breq, sizeof(breq));
1516 buf += sizeof(breq);
1517 len -= sizeof(breq);
1519 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1520 if (len < sizeof(breq))
1523 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1524 sizeof(breq.ifbr_ifsname));
1525 breq.ifbr_ifsflags = bif->bif_flags;
1526 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1527 memcpy(buf, &breq, sizeof(breq));
1529 buf += sizeof(breq);
1530 len -= sizeof(breq);
1533 bifc->ifbic_len = sizeof(breq) * count;
1534 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1535 free(outbuf, M_TEMP);
1540 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1542 struct ifbaconf *bac = arg;
1543 struct bridge_rtnode *brt;
1544 struct ifbareq bareq;
1546 int count, buflen, len, error = 0;
1548 if (bac->ifbac_len == 0)
1552 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1554 buflen = sizeof(bareq) * count;
1556 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1562 len = min(bac->ifbac_len, buflen);
1563 bzero(&bareq, sizeof(bareq));
1564 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1565 if (len < sizeof(bareq))
1567 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1568 sizeof(bareq.ifba_ifsname));
1569 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1570 bareq.ifba_vlan = brt->brt_vlan;
1571 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1572 time_uptime < brt->brt_expire)
1573 bareq.ifba_expire = brt->brt_expire - time_uptime;
1575 bareq.ifba_expire = 0;
1576 bareq.ifba_flags = brt->brt_flags;
1578 memcpy(buf, &bareq, sizeof(bareq));
1580 buf += sizeof(bareq);
1581 len -= sizeof(bareq);
1584 bac->ifbac_len = sizeof(bareq) * count;
1585 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1586 free(outbuf, M_TEMP);
1591 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1593 struct ifbareq *req = arg;
1594 struct bridge_iflist *bif;
1595 struct epoch_tracker et;
1598 NET_EPOCH_ENTER_ET(et);
1600 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1602 NET_EPOCH_EXIT_ET(et);
1606 /* bridge_rtupdate() may acquire the lock. */
1607 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1609 NET_EPOCH_EXIT_ET(et);
1615 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1617 struct ifbrparam *param = arg;
1619 sc->sc_brttimeout = param->ifbrp_ctime;
1624 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1626 struct ifbrparam *param = arg;
1628 param->ifbrp_ctime = sc->sc_brttimeout;
1633 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1635 struct ifbareq *req = arg;
1637 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1641 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1643 struct ifbreq *req = arg;
1646 bridge_rtflush(sc, req->ifbr_ifsflags);
1647 BRIDGE_RT_UNLOCK(sc);
1653 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1655 struct ifbrparam *param = arg;
1656 struct bstp_state *bs = &sc->sc_stp;
1658 param->ifbrp_prio = bs->bs_bridge_priority;
1663 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1665 struct ifbrparam *param = arg;
1667 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1671 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1673 struct ifbrparam *param = arg;
1674 struct bstp_state *bs = &sc->sc_stp;
1676 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1681 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1683 struct ifbrparam *param = arg;
1685 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1689 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1691 struct ifbrparam *param = arg;
1692 struct bstp_state *bs = &sc->sc_stp;
1694 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1699 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1701 struct ifbrparam *param = arg;
1703 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1707 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1709 struct ifbrparam *param = arg;
1710 struct bstp_state *bs = &sc->sc_stp;
1712 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1717 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1719 struct ifbrparam *param = arg;
1721 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1725 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1727 struct ifbreq *req = arg;
1728 struct bridge_iflist *bif;
1730 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1734 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1738 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1740 struct ifbreq *req = arg;
1741 struct bridge_iflist *bif;
1743 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1747 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1751 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1753 struct ifbreq *req = arg;
1754 struct bridge_iflist *bif;
1756 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1760 bif->bif_addrmax = req->ifbr_addrmax;
1765 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1767 struct ifbreq *req = arg;
1768 struct bridge_iflist *bif = NULL;
1771 ifs = ifunit(req->ifbr_ifsname);
1775 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1776 if (ifs == bif->bif_ifp)
1779 if (ifs->if_bridge != NULL)
1782 switch (ifs->if_type) {
1791 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1796 bif->bif_flags = IFBIF_SPAN;
1798 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1804 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1806 struct ifbreq *req = arg;
1807 struct bridge_iflist *bif;
1810 ifs = ifunit(req->ifbr_ifsname);
1814 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1815 if (ifs == bif->bif_ifp)
1821 bridge_delete_span(sc, bif);
1827 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1829 struct ifbropreq *req = arg;
1830 struct bstp_state *bs = &sc->sc_stp;
1831 struct bstp_port *root_port;
1833 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1834 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1835 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1837 root_port = bs->bs_root_port;
1838 if (root_port == NULL)
1839 req->ifbop_root_port = 0;
1841 req->ifbop_root_port = root_port->bp_ifp->if_index;
1843 req->ifbop_holdcount = bs->bs_txholdcount;
1844 req->ifbop_priority = bs->bs_bridge_priority;
1845 req->ifbop_protocol = bs->bs_protover;
1846 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1847 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1848 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1849 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1850 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1851 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1857 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1859 struct ifbrparam *param = arg;
1861 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1866 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1868 struct ifbpstpconf *bifstp = arg;
1869 struct bridge_iflist *bif;
1870 struct bstp_port *bp;
1871 struct ifbpstpreq bpreq;
1873 int count, buflen, len, error = 0;
1876 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1877 if ((bif->bif_flags & IFBIF_STP) != 0)
1881 buflen = sizeof(bpreq) * count;
1882 if (bifstp->ifbpstp_len == 0) {
1883 bifstp->ifbpstp_len = buflen;
1887 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1893 len = min(bifstp->ifbpstp_len, buflen);
1894 bzero(&bpreq, sizeof(bpreq));
1895 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1896 if (len < sizeof(bpreq))
1899 if ((bif->bif_flags & IFBIF_STP) == 0)
1903 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1904 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1905 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1906 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1907 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1908 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1910 memcpy(buf, &bpreq, sizeof(bpreq));
1912 buf += sizeof(bpreq);
1913 len -= sizeof(bpreq);
1916 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1917 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1918 free(outbuf, M_TEMP);
1923 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1925 struct ifbrparam *param = arg;
1927 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1931 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1933 struct ifbrparam *param = arg;
1935 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1941 * Detach an interface from a bridge. Called when a member
1942 * interface is detaching.
1945 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1947 struct bridge_softc *sc = ifp->if_bridge;
1948 struct bridge_iflist *bif;
1950 if (ifp->if_flags & IFF_RENAMING)
1952 if (V_bridge_cloner == NULL) {
1954 * This detach handler can be called after
1955 * vnet_bridge_uninit(). Just return in that case.
1959 /* Check if the interface is a bridge member */
1963 bif = bridge_lookup_member_if(sc, ifp);
1965 bridge_delete_member(sc, bif, 1);
1971 /* Check if the interface is a span port */
1973 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1975 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1976 if (ifp == bif->bif_ifp) {
1977 bridge_delete_span(sc, bif);
1983 BRIDGE_LIST_UNLOCK();
1989 * Initialize a bridge interface.
1992 bridge_init(void *xsc)
1994 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1995 struct ifnet *ifp = sc->sc_ifp;
1997 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
2001 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
2004 ifp->if_drv_flags |= IFF_DRV_RUNNING;
2005 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
2013 * Stop the bridge interface.
2016 bridge_stop(struct ifnet *ifp, int disable)
2018 struct bridge_softc *sc = ifp->if_softc;
2020 BRIDGE_LOCK_ASSERT(sc);
2022 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2026 callout_stop(&sc->sc_brcallout);
2028 bstp_stop(&sc->sc_stp);
2030 bridge_rtflush(sc, IFBF_FLUSHDYN);
2031 BRIDGE_RT_UNLOCK(sc);
2033 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
2039 * Enqueue a packet on a bridge member interface.
2043 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
2049 /* We may be sending a fragment so traverse the mbuf */
2052 m->m_nextpkt = NULL;
2053 len = m->m_pkthdr.len;
2054 mflags = m->m_flags;
2057 * If underlying interface can not do VLAN tag insertion itself
2058 * then attach a packet tag that holds it.
2060 if ((m->m_flags & M_VLANTAG) &&
2061 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
2062 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
2065 "unable to prepend VLAN header\n");
2066 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2069 m->m_flags &= ~M_VLANTAG;
2072 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2073 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2075 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2079 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2080 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2081 if (mflags & M_MCAST)
2082 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2091 * Receive a queued packet from dummynet and pass it on to the output
2094 * The mbuf has the Ethernet header already attached.
2097 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2099 struct bridge_softc *sc;
2101 sc = ifp->if_bridge;
2104 * The packet didnt originate from a member interface. This should only
2105 * ever happen if a member interface is removed while packets are
2113 if (PFIL_HOOKED(&V_inet_pfil_hook)
2115 || PFIL_HOOKED(&V_inet6_pfil_hook)
2118 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2124 bridge_enqueue(sc, ifp, m);
2130 * Send output from a bridge member interface. This
2131 * performs the bridging function for locally originated
2134 * The mbuf has the Ethernet header already attached. We must
2135 * enqueue or free the mbuf before returning.
2138 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2141 struct epoch_tracker et;
2142 struct ether_header *eh;
2143 struct ifnet *dst_if;
2144 struct bridge_softc *sc;
2147 NET_EPOCH_ENTER_ET(et);
2149 if (m->m_len < ETHER_HDR_LEN) {
2150 m = m_pullup(m, ETHER_HDR_LEN);
2152 NET_EPOCH_EXIT_ET(et);
2157 eh = mtod(m, struct ether_header *);
2158 sc = ifp->if_bridge;
2159 vlan = VLANTAGOF(m);
2162 * If bridge is down, but the original output interface is up,
2163 * go ahead and send out that interface. Otherwise, the packet
2166 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2172 * If the packet is a multicast, or we don't know a better way to
2173 * get there, send to all interfaces.
2175 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2178 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2179 if (dst_if == NULL) {
2180 struct bridge_iflist *bif;
2186 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2187 dst_if = bif->bif_ifp;
2189 if (dst_if->if_type == IFT_GIF)
2191 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2195 * If this is not the original output interface,
2196 * and the interface is participating in spanning
2197 * tree, make sure the port is in a state that
2198 * allows forwarding.
2200 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2201 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2204 if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2208 mc = m_copypacket(m, M_NOWAIT);
2210 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2215 bridge_enqueue(sc, dst_if, mc);
2219 NET_EPOCH_EXIT_ET(et);
2225 * XXX Spanning tree consideration here?
2229 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2231 NET_EPOCH_EXIT_ET(et);
2235 bridge_enqueue(sc, dst_if, m);
2236 NET_EPOCH_EXIT_ET(et);
2243 * Do output on a bridge.
2247 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2249 struct epoch_tracker et;
2250 struct bridge_softc *sc;
2251 struct ether_header *eh;
2252 struct ifnet *dst_if;
2255 NET_EPOCH_ENTER_ET(et);
2259 ETHER_BPF_MTAP(ifp, m);
2261 eh = mtod(m, struct ether_header *);
2263 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2264 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2265 error = bridge_enqueue(sc, dst_if, m);
2267 bridge_broadcast(sc, ifp, m, 0);
2269 NET_EPOCH_EXIT_ET(et);
2276 bridge_altq_start(if_t ifp)
2278 struct ifaltq *ifq = &ifp->if_snd;
2282 IFQ_DEQUEUE_NOLOCK(ifq, m);
2284 bridge_transmit(ifp, m);
2285 IFQ_DEQUEUE_NOLOCK(ifq, m);
2291 bridge_altq_transmit(if_t ifp, struct mbuf *m)
2295 if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
2296 IFQ_ENQUEUE(&ifp->if_snd, m, err);
2298 bridge_altq_start(ifp);
2300 err = bridge_transmit(ifp, m);
2307 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2310 bridge_qflush(struct ifnet *ifp __unused)
2317 * The forwarding function of the bridge.
2319 * NOTE: Releases the lock on return.
2322 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2325 struct bridge_iflist *dbif;
2326 struct ifnet *src_if, *dst_if, *ifp;
2327 struct ether_header *eh;
2332 MPASS(in_epoch(net_epoch_preempt));
2334 src_if = m->m_pkthdr.rcvif;
2337 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2338 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2339 vlan = VLANTAGOF(m);
2341 if ((sbif->bif_flags & IFBIF_STP) &&
2342 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2345 eh = mtod(m, struct ether_header *);
2346 dst = eh->ether_dhost;
2348 /* If the interface is learning, record the address. */
2349 if (sbif->bif_flags & IFBIF_LEARNING) {
2350 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2351 sbif, 0, IFBAF_DYNAMIC);
2353 * If the interface has addresses limits then deny any source
2354 * that is not in the cache.
2356 if (error && sbif->bif_addrmax)
2360 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2361 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2365 * At this point, the port either doesn't participate
2366 * in spanning tree or it is in the forwarding state.
2370 * If the packet is unicast, destined for someone on
2371 * "this" side of the bridge, drop it.
2373 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2374 dst_if = bridge_rtlookup(sc, dst, vlan);
2375 if (src_if == dst_if)
2379 * Check if its a reserved multicast address, any address
2380 * listed in 802.1D section 7.12.6 may not be forwarded by the
2382 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2384 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2385 dst[2] == 0xc2 && dst[3] == 0x00 &&
2386 dst[4] == 0x00 && dst[5] <= 0x0f)
2389 /* ...forward it to all interfaces. */
2390 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2395 * If we have a destination interface which is a member of our bridge,
2396 * OR this is a unicast packet, push it through the bpf(4) machinery.
2397 * For broadcast or multicast packets, don't bother because it will
2398 * be reinjected into ether_input. We do this before we pass the packets
2399 * through the pfil(9) framework, as it is possible that pfil(9) will
2400 * drop the packet, or possibly modify it, making it difficult to debug
2401 * firewall issues on the bridge.
2403 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2404 ETHER_BPF_MTAP(ifp, m);
2406 /* run the packet filter */
2407 if (PFIL_HOOKED(&V_inet_pfil_hook)
2409 || PFIL_HOOKED(&V_inet6_pfil_hook)
2412 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2418 if (dst_if == NULL) {
2419 bridge_broadcast(sc, src_if, m, 1);
2424 * At this point, we're dealing with a unicast frame
2425 * going to a different interface.
2427 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2430 dbif = bridge_lookup_member_if(sc, dst_if);
2432 /* Not a member of the bridge (anymore?) */
2435 /* Private segments can not talk to each other */
2436 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2439 if ((dbif->bif_flags & IFBIF_STP) &&
2440 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2443 if (PFIL_HOOKED(&V_inet_pfil_hook)
2445 || PFIL_HOOKED(&V_inet6_pfil_hook)
2448 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2454 bridge_enqueue(sc, dst_if, m);
2464 * Receive input from a member interface. Queue the packet for
2465 * bridging if it is not for us.
2467 static struct mbuf *
2468 bridge_input(struct ifnet *ifp, struct mbuf *m)
2470 struct epoch_tracker et;
2471 struct bridge_softc *sc = ifp->if_bridge;
2472 struct bridge_iflist *bif, *bif2;
2474 struct ether_header *eh;
2475 struct mbuf *mc, *mc2;
2479 NET_EPOCH_ENTER_ET(et);
2481 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2482 NET_EPOCH_EXIT_ET(et);
2487 vlan = VLANTAGOF(m);
2490 * Implement support for bridge monitoring. If this flag has been
2491 * set on this interface, discard the packet once we push it through
2492 * the bpf(4) machinery, but before we do, increment the byte and
2493 * packet counters associated with this interface.
2495 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2496 m->m_pkthdr.rcvif = bifp;
2497 ETHER_BPF_MTAP(bifp, m);
2498 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2499 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2501 NET_EPOCH_EXIT_ET(et);
2504 bif = bridge_lookup_member_if(sc, ifp);
2506 NET_EPOCH_EXIT_ET(et);
2510 eh = mtod(m, struct ether_header *);
2514 if (m->m_flags & (M_BCAST|M_MCAST)) {
2515 /* Tap off 802.1D packets; they do not get forwarded. */
2516 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2517 ETHER_ADDR_LEN) == 0) {
2518 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2519 NET_EPOCH_EXIT_ET(et);
2523 if ((bif->bif_flags & IFBIF_STP) &&
2524 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2525 NET_EPOCH_EXIT_ET(et);
2530 * Make a deep copy of the packet and enqueue the copy
2531 * for bridge processing; return the original packet for
2534 mc = m_dup(m, M_NOWAIT);
2536 NET_EPOCH_EXIT_ET(et);
2540 /* Perform the bridge forwarding function with the copy. */
2541 bridge_forward(sc, bif, mc);
2544 * Reinject the mbuf as arriving on the bridge so we have a
2545 * chance at claiming multicast packets. We can not loop back
2546 * here from ether_input as a bridge is never a member of a
2549 KASSERT(bifp->if_bridge == NULL,
2550 ("loop created in bridge_input"));
2551 mc2 = m_dup(m, M_NOWAIT);
2553 /* Keep the layer3 header aligned */
2554 int i = min(mc2->m_pkthdr.len, max_protohdr);
2555 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2558 mc2->m_pkthdr.rcvif = bifp;
2559 (*bifp->if_input)(bifp, mc2);
2562 /* Return the original packet for local processing. */
2563 NET_EPOCH_EXIT_ET(et);
2567 if ((bif->bif_flags & IFBIF_STP) &&
2568 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2569 NET_EPOCH_EXIT_ET(et);
2573 #if (defined(INET) || defined(INET6))
2574 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2575 || ((iface)->if_carp \
2576 && (*carp_forus_p)((iface), eh->ether_dhost))
2577 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2578 || ((iface)->if_carp \
2579 && (*carp_forus_p)((iface), eh->ether_shost))
2581 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2582 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2586 # define OR_PFIL_HOOKED_INET6 \
2587 || PFIL_HOOKED(&V_inet6_pfil_hook)
2589 # define OR_PFIL_HOOKED_INET6
2592 #define GRAB_OUR_PACKETS(iface) \
2593 if ((iface)->if_type == IFT_GIF) \
2595 /* It is destined for us. */ \
2596 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2597 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2599 if ((iface)->if_type == IFT_BRIDGE) { \
2600 ETHER_BPF_MTAP(iface, m); \
2601 if_inc_counter(iface, IFCOUNTER_IPACKETS, 1); \
2602 if_inc_counter(iface, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2603 /* Filter on the physical interface. */ \
2604 if (V_pfil_local_phys && \
2605 (PFIL_HOOKED(&V_inet_pfil_hook) \
2606 OR_PFIL_HOOKED_INET6)) { \
2607 if (bridge_pfil(&m, NULL, ifp, \
2608 PFIL_IN) != 0 || m == NULL) { \
2609 NET_EPOCH_EXIT_ET(et); \
2612 eh = mtod(m, struct ether_header *); \
2615 if (bif->bif_flags & IFBIF_LEARNING) { \
2616 error = bridge_rtupdate(sc, eh->ether_shost, \
2617 vlan, bif, 0, IFBAF_DYNAMIC); \
2618 if (error && bif->bif_addrmax) { \
2620 NET_EPOCH_EXIT_ET(et); \
2624 m->m_pkthdr.rcvif = iface; \
2625 NET_EPOCH_EXIT_ET(et); \
2629 /* We just received a packet that we sent out. */ \
2630 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2631 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2634 NET_EPOCH_EXIT_ET(et); \
2639 * Unicast. Make sure it's not for the bridge.
2641 do { GRAB_OUR_PACKETS(bifp) } while (0);
2644 * Give a chance for ifp at first priority. This will help when the
2645 * packet comes through the interface like VLAN's with the same MACs
2646 * on several interfaces from the same bridge. This also will save
2647 * some CPU cycles in case the destination interface and the input
2648 * interface (eq ifp) are the same.
2650 do { GRAB_OUR_PACKETS(ifp) } while (0);
2652 /* Now check the all bridge members. */
2653 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2654 GRAB_OUR_PACKETS(bif2->bif_ifp)
2657 #undef OR_CARP_CHECK_WE_ARE_DST
2658 #undef OR_CARP_CHECK_WE_ARE_SRC
2659 #undef OR_PFIL_HOOKED_INET6
2660 #undef GRAB_OUR_PACKETS
2662 /* Perform the bridge forwarding function. */
2663 bridge_forward(sc, bif, m);
2665 NET_EPOCH_EXIT_ET(et);
2672 * Send a frame to all interfaces that are members of
2673 * the bridge, except for the one on which the packet
2676 * NOTE: Releases the lock on return.
2679 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2680 struct mbuf *m, int runfilt)
2682 struct bridge_iflist *dbif, *sbif;
2684 struct ifnet *dst_if;
2687 MPASS(in_epoch(net_epoch_preempt));
2689 sbif = bridge_lookup_member_if(sc, src_if);
2691 /* Filter on the bridge interface before broadcasting */
2692 if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2694 || PFIL_HOOKED(&V_inet6_pfil_hook)
2697 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2703 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2704 dst_if = dbif->bif_ifp;
2705 if (dst_if == src_if)
2708 /* Private segments can not talk to each other */
2709 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2712 if ((dbif->bif_flags & IFBIF_STP) &&
2713 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2716 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2717 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2720 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2723 if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2727 mc = m_dup(m, M_NOWAIT);
2729 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2735 * Filter on the output interface. Pass a NULL bridge interface
2736 * pointer so we do not redundantly filter on the bridge for
2737 * each interface we broadcast on.
2739 if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2741 || PFIL_HOOKED(&V_inet6_pfil_hook)
2745 /* Keep the layer3 header aligned */
2746 i = min(mc->m_pkthdr.len, max_protohdr);
2747 mc = m_copyup(mc, i, ETHER_ALIGN);
2749 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2753 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2759 bridge_enqueue(sc, dst_if, mc);
2768 * Duplicate a packet out one or more interfaces that are in span mode,
2769 * the original mbuf is unmodified.
2772 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2774 struct bridge_iflist *bif;
2775 struct ifnet *dst_if;
2778 MPASS(in_epoch(net_epoch_preempt));
2780 if (CK_LIST_EMPTY(&sc->sc_spanlist))
2783 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2784 dst_if = bif->bif_ifp;
2786 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2789 mc = m_copypacket(m, M_NOWAIT);
2791 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2795 bridge_enqueue(sc, dst_if, mc);
2802 * Add a bridge routing entry.
2805 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2806 struct bridge_iflist *bif, int setflags, uint8_t flags)
2808 struct bridge_rtnode *brt;
2811 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
2813 /* Check the source address is valid and not multicast. */
2814 if (ETHER_IS_MULTICAST(dst) ||
2815 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2816 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2819 /* 802.1p frames map to vlan 1 */
2824 * A route for this destination might already exist. If so,
2825 * update it, otherwise create a new one.
2827 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2830 /* Check again, now that we have the lock. There could have
2831 * been a race and we only want to insert this once. */
2832 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) {
2833 BRIDGE_RT_UNLOCK(sc);
2837 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2838 sc->sc_brtexceeded++;
2839 BRIDGE_RT_UNLOCK(sc);
2842 /* Check per interface address limits (if enabled) */
2843 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2844 bif->bif_addrexceeded++;
2845 BRIDGE_RT_UNLOCK(sc);
2850 * Allocate a new bridge forwarding node, and
2851 * initialize the expiration time and Ethernet
2854 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2856 BRIDGE_RT_UNLOCK(sc);
2859 brt->brt_vnet = curvnet;
2861 if (bif->bif_flags & IFBIF_STICKY)
2862 brt->brt_flags = IFBAF_STICKY;
2864 brt->brt_flags = IFBAF_DYNAMIC;
2866 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2867 brt->brt_vlan = vlan;
2869 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2870 uma_zfree(V_bridge_rtnode_zone, brt);
2871 BRIDGE_RT_UNLOCK(sc);
2877 BRIDGE_RT_UNLOCK(sc);
2880 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2881 brt->brt_dst != bif) {
2883 brt->brt_dst->bif_addrcnt--;
2885 brt->brt_dst->bif_addrcnt++;
2886 BRIDGE_RT_UNLOCK(sc);
2889 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2890 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2892 brt->brt_flags = flags;
2900 * Lookup the destination interface for an address.
2902 static struct ifnet *
2903 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2905 struct bridge_rtnode *brt;
2907 MPASS(in_epoch(net_epoch_preempt));
2909 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2912 return (brt->brt_ifp);
2918 * Trim the routine table so that we have a number
2919 * of routing entries less than or equal to the
2923 bridge_rttrim(struct bridge_softc *sc)
2925 struct bridge_rtnode *brt, *nbrt;
2927 MPASS(in_epoch(net_epoch_preempt));
2928 BRIDGE_RT_LOCK_ASSERT(sc);
2930 /* Make sure we actually need to do this. */
2931 if (sc->sc_brtcnt <= sc->sc_brtmax)
2934 /* Force an aging cycle; this might trim enough addresses. */
2936 if (sc->sc_brtcnt <= sc->sc_brtmax)
2939 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2940 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2941 bridge_rtnode_destroy(sc, brt);
2942 if (sc->sc_brtcnt <= sc->sc_brtmax)
2951 * Aging timer for the bridge.
2954 bridge_timer(void *arg)
2956 struct bridge_softc *sc = arg;
2958 BRIDGE_RT_LOCK_ASSERT(sc);
2960 /* Destruction of rtnodes requires a proper vnet context */
2961 CURVNET_SET(sc->sc_ifp->if_vnet);
2964 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2965 callout_reset(&sc->sc_brcallout,
2966 bridge_rtable_prune_period * hz, bridge_timer, sc);
2973 * Perform an aging cycle.
2976 bridge_rtage(struct bridge_softc *sc)
2978 struct bridge_rtnode *brt, *nbrt;
2980 BRIDGE_RT_LOCK_ASSERT(sc);
2982 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2983 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2984 if (time_uptime >= brt->brt_expire)
2985 bridge_rtnode_destroy(sc, brt);
2993 * Remove all dynamic addresses from the bridge.
2996 bridge_rtflush(struct bridge_softc *sc, int full)
2998 struct bridge_rtnode *brt, *nbrt;
3000 BRIDGE_RT_LOCK_ASSERT(sc);
3002 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3003 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3004 bridge_rtnode_destroy(sc, brt);
3011 * Remove an address from the table.
3014 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3016 struct bridge_rtnode *brt;
3022 * If vlan is zero then we want to delete for all vlans so the lookup
3023 * may return more than one.
3025 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
3026 bridge_rtnode_destroy(sc, brt);
3030 BRIDGE_RT_UNLOCK(sc);
3032 return (found ? 0 : ENOENT);
3038 * Delete routes to a speicifc member interface.
3041 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
3043 struct bridge_rtnode *brt, *nbrt;
3045 BRIDGE_RT_LOCK_ASSERT(sc);
3047 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3048 if (brt->brt_ifp == ifp && (full ||
3049 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
3050 bridge_rtnode_destroy(sc, brt);
3055 * bridge_rtable_init:
3057 * Initialize the route table for this bridge.
3060 bridge_rtable_init(struct bridge_softc *sc)
3064 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
3065 M_DEVBUF, M_WAITOK);
3067 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
3068 CK_LIST_INIT(&sc->sc_rthash[i]);
3070 sc->sc_rthash_key = arc4random();
3071 CK_LIST_INIT(&sc->sc_rtlist);
3075 * bridge_rtable_fini:
3077 * Deconstruct the route table for this bridge.
3080 bridge_rtable_fini(struct bridge_softc *sc)
3083 KASSERT(sc->sc_brtcnt == 0,
3084 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3085 free(sc->sc_rthash, M_DEVBUF);
3089 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3090 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3092 #define mix(a, b, c) \
3094 a -= b; a -= c; a ^= (c >> 13); \
3095 b -= c; b -= a; b ^= (a << 8); \
3096 c -= a; c -= b; c ^= (b >> 13); \
3097 a -= b; a -= c; a ^= (c >> 12); \
3098 b -= c; b -= a; b ^= (a << 16); \
3099 c -= a; c -= b; c ^= (b >> 5); \
3100 a -= b; a -= c; a ^= (c >> 3); \
3101 b -= c; b -= a; b ^= (a << 10); \
3102 c -= a; c -= b; c ^= (b >> 15); \
3103 } while (/*CONSTCOND*/0)
3105 static __inline uint32_t
3106 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3108 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3119 return (c & BRIDGE_RTHASH_MASK);
3125 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3129 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3130 d = ((int)a[i]) - ((int)b[i]);
3137 * bridge_rtnode_lookup:
3139 * Look up a bridge route node for the specified destination. Compare the
3140 * vlan id or if zero then just return the first match.
3142 static struct bridge_rtnode *
3143 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3145 struct bridge_rtnode *brt;
3149 BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3151 hash = bridge_rthash(sc, addr);
3152 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3153 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3154 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3164 * bridge_rtnode_insert:
3166 * Insert the specified bridge node into the route table. We
3167 * assume the entry is not already in the table.
3170 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3172 struct bridge_rtnode *lbrt;
3176 BRIDGE_RT_LOCK_ASSERT(sc);
3178 hash = bridge_rthash(sc, brt->brt_addr);
3180 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3182 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3187 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3188 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3191 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3194 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3195 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3198 lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3199 } while (lbrt != NULL);
3202 panic("bridge_rtnode_insert: impossible");
3206 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3213 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3215 struct bridge_rtnode *brt;
3217 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3219 CURVNET_SET(brt->brt_vnet);
3220 uma_zfree(V_bridge_rtnode_zone, brt);
3225 * bridge_rtnode_destroy:
3227 * Destroy a bridge rtnode.
3230 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3233 BRIDGE_RT_LOCK_ASSERT(sc);
3235 CK_LIST_REMOVE(brt, brt_hash);
3237 CK_LIST_REMOVE(brt, brt_list);
3239 brt->brt_dst->bif_addrcnt--;
3241 epoch_call(net_epoch_preempt, &brt->brt_epoch_ctx,
3242 bridge_rtnode_destroy_cb);
3246 * bridge_rtable_expire:
3248 * Set the expiry time for all routes on an interface.
3251 bridge_rtable_expire(struct ifnet *ifp, int age)
3253 struct bridge_softc *sc = ifp->if_bridge;
3254 struct bridge_rtnode *brt;
3256 CURVNET_SET(ifp->if_vnet);
3260 * If the age is zero then flush, otherwise set all the expiry times to
3261 * age for the interface
3264 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3266 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3267 /* Cap the expiry time to 'age' */
3268 if (brt->brt_ifp == ifp &&
3269 brt->brt_expire > time_uptime + age &&
3270 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3271 brt->brt_expire = time_uptime + age;
3274 BRIDGE_RT_UNLOCK(sc);
3279 * bridge_state_change:
3281 * Callback from the bridgestp code when a port changes states.
3284 bridge_state_change(struct ifnet *ifp, int state)
3286 struct bridge_softc *sc = ifp->if_bridge;
3287 static const char *stpstates[] = {
3296 CURVNET_SET(ifp->if_vnet);
3298 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3299 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3304 * Send bridge packets through pfil if they are one of the types pfil can deal
3305 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3306 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3310 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3312 int snap, error, i, hlen;
3313 struct ether_header *eh1, eh2;
3316 u_int16_t ether_type;
3319 error = -1; /* Default error if not error == 0 */
3322 /* we may return with the IP fields swapped, ensure its not shared */
3323 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3326 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3327 return (0); /* filtering is disabled */
3329 i = min((*mp)->m_pkthdr.len, max_protohdr);
3330 if ((*mp)->m_len < i) {
3331 *mp = m_pullup(*mp, i);
3333 printf("%s: m_pullup failed\n", __func__);
3338 eh1 = mtod(*mp, struct ether_header *);
3339 ether_type = ntohs(eh1->ether_type);
3342 * Check for SNAP/LLC.
3344 if (ether_type < ETHERMTU) {
3345 struct llc *llc2 = (struct llc *)(eh1 + 1);
3347 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3348 llc2->llc_dsap == LLC_SNAP_LSAP &&
3349 llc2->llc_ssap == LLC_SNAP_LSAP &&
3350 llc2->llc_control == LLC_UI) {
3351 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3357 * If we're trying to filter bridge traffic, don't look at anything
3358 * other than IP and ARP traffic. If the filter doesn't understand
3359 * IPv6, don't allow IPv6 through the bridge either. This is lame
3360 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3361 * but of course we don't have an AppleTalk filter to begin with.
3362 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3365 switch (ether_type) {
3367 case ETHERTYPE_REVARP:
3368 if (V_pfil_ipfw_arp == 0)
3369 return (0); /* Automatically pass */
3374 case ETHERTYPE_IPV6:
3379 * Check to see if the user wants to pass non-ip
3380 * packets, these will not be checked by pfil(9) and
3381 * passed unconditionally so the default is to drop.
3387 /* Run the packet through pfil before stripping link headers */
3388 if (PFIL_HOOKED(&V_link_pfil_hook) && V_pfil_ipfw != 0 &&
3389 dir == PFIL_OUT && ifp != NULL) {
3391 error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, 0,
3394 if (*mp == NULL || error != 0) /* packet consumed by filter */
3398 /* Strip off the Ethernet header and keep a copy. */
3399 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3400 m_adj(*mp, ETHER_HDR_LEN);
3402 /* Strip off snap header, if present */
3404 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3405 m_adj(*mp, sizeof(struct llc));
3409 * Check the IP header for alignment and errors
3411 if (dir == PFIL_IN) {
3412 switch (ether_type) {
3414 error = bridge_ip_checkbasic(mp);
3417 case ETHERTYPE_IPV6:
3418 error = bridge_ip6_checkbasic(mp);
3431 * Run the packet through pfil
3433 switch (ether_type) {
3436 * Run pfil on the member interface and the bridge, both can
3437 * be skipped by clearing pfil_member or pfil_bridge.
3440 * in_if -> bridge_if -> out_if
3442 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3443 error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3446 if (*mp == NULL || error != 0) /* filter may consume */
3449 if (V_pfil_member && ifp != NULL)
3450 error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3453 if (*mp == NULL || error != 0) /* filter may consume */
3456 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3457 error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3460 if (*mp == NULL || error != 0) /* filter may consume */
3463 /* check if we need to fragment the packet */
3464 /* bridge_fragment generates a mbuf chain of packets */
3465 /* that already include eth headers */
3466 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3467 i = (*mp)->m_pkthdr.len;
3468 if (i > ifp->if_mtu) {
3469 error = bridge_fragment(ifp, mp, &eh2, snap,
3475 /* Recalculate the ip checksum. */
3476 ip = mtod(*mp, struct ip *);
3477 hlen = ip->ip_hl << 2;
3478 if (hlen < sizeof(struct ip))
3480 if (hlen > (*mp)->m_len) {
3481 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3483 ip = mtod(*mp, struct ip *);
3488 if (hlen == sizeof(struct ip))
3489 ip->ip_sum = in_cksum_hdr(ip);
3491 ip->ip_sum = in_cksum(*mp, hlen);
3495 case ETHERTYPE_IPV6:
3496 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3497 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3500 if (*mp == NULL || error != 0) /* filter may consume */
3503 if (V_pfil_member && ifp != NULL)
3504 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3507 if (*mp == NULL || error != 0) /* filter may consume */
3510 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3511 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3528 * Finally, put everything back the way it was and return
3531 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3534 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3537 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3540 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3551 * Perform basic checks on header size since
3552 * pfil assumes ip_input has already processed
3553 * it for it. Cut-and-pasted from ip_input.c.
3554 * Given how simple the IPv6 version is,
3555 * does the IPv4 version really need to be
3558 * XXX Should we update ipstat here, or not?
3559 * XXX Right now we update ipstat but not
3563 bridge_ip_checkbasic(struct mbuf **mp)
3565 struct mbuf *m = *mp;
3573 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3574 if ((m = m_copyup(m, sizeof(struct ip),
3575 (max_linkhdr + 3) & ~3)) == NULL) {
3576 /* XXXJRT new stat, please */
3577 KMOD_IPSTAT_INC(ips_toosmall);
3580 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3581 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3582 KMOD_IPSTAT_INC(ips_toosmall);
3586 ip = mtod(m, struct ip *);
3587 if (ip == NULL) goto bad;
3589 if (ip->ip_v != IPVERSION) {
3590 KMOD_IPSTAT_INC(ips_badvers);
3593 hlen = ip->ip_hl << 2;
3594 if (hlen < sizeof(struct ip)) { /* minimum header length */
3595 KMOD_IPSTAT_INC(ips_badhlen);
3598 if (hlen > m->m_len) {
3599 if ((m = m_pullup(m, hlen)) == NULL) {
3600 KMOD_IPSTAT_INC(ips_badhlen);
3603 ip = mtod(m, struct ip *);
3604 if (ip == NULL) goto bad;
3607 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3608 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3610 if (hlen == sizeof(struct ip)) {
3611 sum = in_cksum_hdr(ip);
3613 sum = in_cksum(m, hlen);
3617 KMOD_IPSTAT_INC(ips_badsum);
3621 /* Retrieve the packet length. */
3622 len = ntohs(ip->ip_len);
3625 * Check for additional length bogosity
3628 KMOD_IPSTAT_INC(ips_badlen);
3633 * Check that the amount of data in the buffers
3634 * is as at least much as the IP header would have us expect.
3635 * Drop packet if shorter than we expect.
3637 if (m->m_pkthdr.len < len) {
3638 KMOD_IPSTAT_INC(ips_tooshort);
3642 /* Checks out, proceed */
3653 * Same as above, but for IPv6.
3654 * Cut-and-pasted from ip6_input.c.
3655 * XXX Should we update ip6stat, or not?
3658 bridge_ip6_checkbasic(struct mbuf **mp)
3660 struct mbuf *m = *mp;
3661 struct ip6_hdr *ip6;
3664 * If the IPv6 header is not aligned, slurp it up into a new
3665 * mbuf with space for link headers, in the event we forward
3666 * it. Otherwise, if it is aligned, make sure the entire base
3667 * IPv6 header is in the first mbuf of the chain.
3669 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3670 struct ifnet *inifp = m->m_pkthdr.rcvif;
3671 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3672 (max_linkhdr + 3) & ~3)) == NULL) {
3673 /* XXXJRT new stat, please */
3674 IP6STAT_INC(ip6s_toosmall);
3675 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3678 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3679 struct ifnet *inifp = m->m_pkthdr.rcvif;
3680 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3681 IP6STAT_INC(ip6s_toosmall);
3682 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3687 ip6 = mtod(m, struct ip6_hdr *);
3689 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3690 IP6STAT_INC(ip6s_badvers);
3691 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3695 /* Checks out, proceed */
3708 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3711 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3712 int snap, struct llc *llc)
3714 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3718 if (m->m_len < sizeof(struct ip) &&
3719 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3721 ip = mtod(m, struct ip *);
3723 m->m_pkthdr.csum_flags |= CSUM_IP;
3724 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3729 * Walk the chain and re-add the Ethernet header for
3732 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3733 nextpkt = mcur->m_nextpkt;
3734 mcur->m_nextpkt = NULL;
3736 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3740 mprev->m_nextpkt = nextpkt;
3743 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3746 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3750 mprev->m_nextpkt = nextpkt;
3753 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3756 * The previous two M_PREPEND could have inserted one or two
3757 * mbufs in front so we have to update the previous packet's
3760 mcur->m_nextpkt = nextpkt;
3762 mprev->m_nextpkt = mcur;
3764 /* The first mbuf in the original chain needs to be
3771 KMOD_IPSTAT_INC(ips_fragmented);
3775 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3776 m = mcur->m_nextpkt;
3783 bridge_linkstate(struct ifnet *ifp)
3785 struct bridge_softc *sc = ifp->if_bridge;
3786 struct bridge_iflist *bif;
3787 struct epoch_tracker et;
3789 NET_EPOCH_ENTER_ET(et);
3791 bif = bridge_lookup_member_if(sc, ifp);
3793 NET_EPOCH_EXIT_ET(et);
3796 bridge_linkcheck(sc);
3798 bstp_linkstate(&bif->bif_stp);
3800 NET_EPOCH_EXIT_ET(et);
3804 bridge_linkcheck(struct bridge_softc *sc)
3806 struct bridge_iflist *bif;
3807 int new_link, hasls;
3809 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3811 new_link = LINK_STATE_DOWN;
3813 /* Our link is considered up if at least one of our ports is active */
3814 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3815 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3817 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3818 new_link = LINK_STATE_UP;
3822 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3823 /* If no interfaces support link-state then we default to up */
3824 new_link = LINK_STATE_UP;
3826 if_link_state_change(sc->sc_ifp, new_link);