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
188 * Bridge interface list entry.
190 struct bridge_iflist {
191 LIST_ENTRY(bridge_iflist) bif_next;
192 struct ifnet *bif_ifp; /* member if */
193 struct bstp_port bif_stp; /* STP state */
194 uint32_t bif_flags; /* member if flags */
195 int bif_savedcaps; /* saved capabilities */
196 uint32_t bif_addrmax; /* max # of addresses */
197 uint32_t bif_addrcnt; /* cur. # of addresses */
198 uint32_t bif_addrexceeded;/* # of address violations */
204 struct bridge_rtnode {
205 LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
206 LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
207 struct bridge_iflist *brt_dst; /* destination if */
208 unsigned long brt_expire; /* expiration time */
209 uint8_t brt_flags; /* address flags */
210 uint8_t brt_addr[ETHER_ADDR_LEN];
211 uint16_t brt_vlan; /* vlan id */
213 #define brt_ifp brt_dst->bif_ifp
216 * Software state for each bridge.
218 struct bridge_softc {
219 struct ifnet *sc_ifp; /* make this an interface */
220 LIST_ENTRY(bridge_softc) sc_list;
223 uint32_t sc_brtmax; /* max # of addresses */
224 uint32_t sc_brtcnt; /* cur. # of addresses */
225 uint32_t sc_brttimeout; /* rt timeout in seconds */
226 struct callout sc_brcallout; /* bridge callout */
227 uint32_t sc_iflist_ref; /* refcount for sc_iflist */
228 uint32_t sc_iflist_xcnt; /* refcount for sc_iflist */
229 LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
230 LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
231 LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
232 uint32_t sc_rthash_key; /* key for hash */
233 LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
234 struct bstp_state sc_stp; /* STP state */
235 uint32_t sc_brtexceeded; /* # of cache drops */
236 struct ifnet *sc_ifaddr; /* member mac copied from */
237 struct ether_addr sc_defaddr; /* Default MAC address */
240 VNET_DEFINE_STATIC(struct mtx, bridge_list_mtx);
241 #define V_bridge_list_mtx VNET(bridge_list_mtx)
242 static eventhandler_tag bridge_detach_cookie;
244 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
246 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
247 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone)
249 static int bridge_clone_create(struct if_clone *, int, caddr_t);
250 static void bridge_clone_destroy(struct ifnet *);
252 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
253 static void bridge_mutecaps(struct bridge_softc *);
254 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
256 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
257 static void bridge_init(void *);
258 static void bridge_dummynet(struct mbuf *, struct ifnet *);
259 static void bridge_stop(struct ifnet *, int);
260 static int bridge_transmit(struct ifnet *, struct mbuf *);
261 static void bridge_qflush(struct ifnet *);
262 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
263 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
265 static int bridge_enqueue(struct bridge_softc *, struct ifnet *,
267 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
269 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
272 static void bridge_timer(void *);
274 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
276 static void bridge_span(struct bridge_softc *, struct mbuf *);
278 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
279 uint16_t, struct bridge_iflist *, int, uint8_t);
280 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
282 static void bridge_rttrim(struct bridge_softc *);
283 static void bridge_rtage(struct bridge_softc *);
284 static void bridge_rtflush(struct bridge_softc *, int);
285 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
288 static void bridge_rtable_init(struct bridge_softc *);
289 static void bridge_rtable_fini(struct bridge_softc *);
291 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
292 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
293 const uint8_t *, uint16_t);
294 static int bridge_rtnode_insert(struct bridge_softc *,
295 struct bridge_rtnode *);
296 static void bridge_rtnode_destroy(struct bridge_softc *,
297 struct bridge_rtnode *);
298 static void bridge_rtable_expire(struct ifnet *, int);
299 static void bridge_state_change(struct ifnet *, int);
301 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
303 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
305 static void bridge_delete_member(struct bridge_softc *,
306 struct bridge_iflist *, int);
307 static void bridge_delete_span(struct bridge_softc *,
308 struct bridge_iflist *);
310 static int bridge_ioctl_add(struct bridge_softc *, void *);
311 static int bridge_ioctl_del(struct bridge_softc *, void *);
312 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
313 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
314 static int bridge_ioctl_scache(struct bridge_softc *, void *);
315 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
316 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
317 static int bridge_ioctl_rts(struct bridge_softc *, void *);
318 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
319 static int bridge_ioctl_sto(struct bridge_softc *, void *);
320 static int bridge_ioctl_gto(struct bridge_softc *, void *);
321 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
322 static int bridge_ioctl_flush(struct bridge_softc *, void *);
323 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
324 static int bridge_ioctl_spri(struct bridge_softc *, void *);
325 static int bridge_ioctl_ght(struct bridge_softc *, void *);
326 static int bridge_ioctl_sht(struct bridge_softc *, void *);
327 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
328 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
329 static int bridge_ioctl_gma(struct bridge_softc *, void *);
330 static int bridge_ioctl_sma(struct bridge_softc *, void *);
331 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
332 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
333 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
334 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
335 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
336 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
337 static int bridge_ioctl_grte(struct bridge_softc *, void *);
338 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
339 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
340 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
341 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
343 static int bridge_ip_checkbasic(struct mbuf **mp);
345 static int bridge_ip6_checkbasic(struct mbuf **mp);
347 static int bridge_fragment(struct ifnet *, struct mbuf **mp,
348 struct ether_header *, int, struct llc *);
349 static void bridge_linkstate(struct ifnet *ifp);
350 static void bridge_linkcheck(struct bridge_softc *sc);
353 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
354 #define VLANTAGOF(_m) \
355 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
357 static struct bstp_cb_ops bridge_ops = {
358 .bcb_state = bridge_state_change,
359 .bcb_rtage = bridge_rtable_expire
362 SYSCTL_DECL(_net_link);
363 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
365 /* only pass IP[46] packets when pfil is enabled */
366 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
367 #define V_pfil_onlyip VNET(pfil_onlyip)
368 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
369 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
370 "Only pass IP packets when pfil is enabled");
372 /* run pfil hooks on the bridge interface */
373 VNET_DEFINE_STATIC(int, pfil_bridge) = 1;
374 #define V_pfil_bridge VNET(pfil_bridge)
375 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
376 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
377 "Packet filter on the bridge interface");
379 /* layer2 filter with ipfw */
380 VNET_DEFINE_STATIC(int, pfil_ipfw);
381 #define V_pfil_ipfw VNET(pfil_ipfw)
383 /* layer2 ARP filter with ipfw */
384 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
385 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
386 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
387 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
388 "Filter ARP packets through IPFW layer2");
390 /* run pfil hooks on the member interface */
391 VNET_DEFINE_STATIC(int, pfil_member) = 1;
392 #define V_pfil_member VNET(pfil_member)
393 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
394 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
395 "Packet filter on the member interface");
397 /* run pfil hooks on the physical interface for locally destined packets */
398 VNET_DEFINE_STATIC(int, pfil_local_phys);
399 #define V_pfil_local_phys VNET(pfil_local_phys)
400 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
401 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
402 "Packet filter on the physical interface for locally destined packets");
404 /* log STP state changes */
405 VNET_DEFINE_STATIC(int, log_stp);
406 #define V_log_stp VNET(log_stp)
407 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
408 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
409 "Log STP state changes");
411 /* share MAC with first bridge member */
412 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
413 #define V_bridge_inherit_mac VNET(bridge_inherit_mac)
414 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
415 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
416 "Inherit MAC address from the first bridge member");
418 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
419 #define V_allow_llz_overlap VNET(allow_llz_overlap)
420 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
421 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
422 "Allow overlap of link-local scope "
423 "zones of a bridge interface and the member interfaces");
425 struct bridge_control {
426 int (*bc_func)(struct bridge_softc *, void *);
431 #define BC_F_COPYIN 0x01 /* copy arguments in */
432 #define BC_F_COPYOUT 0x02 /* copy arguments out */
433 #define BC_F_SUSER 0x04 /* do super-user check */
435 const struct bridge_control bridge_control_table[] = {
436 { bridge_ioctl_add, sizeof(struct ifbreq),
437 BC_F_COPYIN|BC_F_SUSER },
438 { bridge_ioctl_del, sizeof(struct ifbreq),
439 BC_F_COPYIN|BC_F_SUSER },
441 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
442 BC_F_COPYIN|BC_F_COPYOUT },
443 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
444 BC_F_COPYIN|BC_F_SUSER },
446 { bridge_ioctl_scache, sizeof(struct ifbrparam),
447 BC_F_COPYIN|BC_F_SUSER },
448 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
451 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
452 BC_F_COPYIN|BC_F_COPYOUT },
453 { bridge_ioctl_rts, sizeof(struct ifbaconf),
454 BC_F_COPYIN|BC_F_COPYOUT },
456 { bridge_ioctl_saddr, sizeof(struct ifbareq),
457 BC_F_COPYIN|BC_F_SUSER },
459 { bridge_ioctl_sto, sizeof(struct ifbrparam),
460 BC_F_COPYIN|BC_F_SUSER },
461 { bridge_ioctl_gto, sizeof(struct ifbrparam),
464 { bridge_ioctl_daddr, sizeof(struct ifbareq),
465 BC_F_COPYIN|BC_F_SUSER },
467 { bridge_ioctl_flush, sizeof(struct ifbreq),
468 BC_F_COPYIN|BC_F_SUSER },
470 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
472 { bridge_ioctl_spri, sizeof(struct ifbrparam),
473 BC_F_COPYIN|BC_F_SUSER },
475 { bridge_ioctl_ght, sizeof(struct ifbrparam),
477 { bridge_ioctl_sht, sizeof(struct ifbrparam),
478 BC_F_COPYIN|BC_F_SUSER },
480 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
482 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
483 BC_F_COPYIN|BC_F_SUSER },
485 { bridge_ioctl_gma, sizeof(struct ifbrparam),
487 { bridge_ioctl_sma, sizeof(struct ifbrparam),
488 BC_F_COPYIN|BC_F_SUSER },
490 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
491 BC_F_COPYIN|BC_F_SUSER },
493 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
494 BC_F_COPYIN|BC_F_SUSER },
496 { bridge_ioctl_addspan, sizeof(struct ifbreq),
497 BC_F_COPYIN|BC_F_SUSER },
498 { bridge_ioctl_delspan, sizeof(struct ifbreq),
499 BC_F_COPYIN|BC_F_SUSER },
501 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
504 { bridge_ioctl_grte, sizeof(struct ifbrparam),
507 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
508 BC_F_COPYIN|BC_F_COPYOUT },
510 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
511 BC_F_COPYIN|BC_F_SUSER },
513 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
514 BC_F_COPYIN|BC_F_SUSER },
516 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
517 BC_F_COPYIN|BC_F_SUSER },
520 const int bridge_control_table_size = nitems(bridge_control_table);
522 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
523 #define V_bridge_list VNET(bridge_list)
524 #define BRIDGE_LIST_LOCK_INIT(x) mtx_init(&V_bridge_list_mtx, \
525 "if_bridge list", NULL, MTX_DEF)
526 #define BRIDGE_LIST_LOCK_DESTROY(x) mtx_destroy(&V_bridge_list_mtx)
527 #define BRIDGE_LIST_LOCK(x) mtx_lock(&V_bridge_list_mtx)
528 #define BRIDGE_LIST_UNLOCK(x) mtx_unlock(&V_bridge_list_mtx)
530 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
531 #define V_bridge_cloner VNET(bridge_cloner)
533 static const char bridge_name[] = "bridge";
536 vnet_bridge_init(const void *unused __unused)
539 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
540 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
542 BRIDGE_LIST_LOCK_INIT();
543 LIST_INIT(&V_bridge_list);
544 V_bridge_cloner = if_clone_simple(bridge_name,
545 bridge_clone_create, bridge_clone_destroy, 0);
547 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
548 vnet_bridge_init, NULL);
551 vnet_bridge_uninit(const void *unused __unused)
554 if_clone_detach(V_bridge_cloner);
555 V_bridge_cloner = NULL;
556 BRIDGE_LIST_LOCK_DESTROY();
557 uma_zdestroy(V_bridge_rtnode_zone);
559 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
560 vnet_bridge_uninit, NULL);
563 bridge_modevent(module_t mod, int type, void *data)
568 bridge_dn_p = bridge_dummynet;
569 bridge_detach_cookie = EVENTHANDLER_REGISTER(
570 ifnet_departure_event, bridge_ifdetach, NULL,
571 EVENTHANDLER_PRI_ANY);
574 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
575 bridge_detach_cookie);
584 static moduledata_t bridge_mod = {
590 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
591 MODULE_VERSION(if_bridge, 1);
592 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
595 * handler for net.link.bridge.ipfw
598 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
600 int enable = V_pfil_ipfw;
603 error = sysctl_handle_int(oidp, &enable, 0, req);
606 if (enable != V_pfil_ipfw) {
607 V_pfil_ipfw = enable;
610 * Disable pfil so that ipfw doesnt run twice, if the user
611 * really wants both then they can re-enable pfil_bridge and/or
612 * pfil_member. Also allow non-ip packets as ipfw can filter by
624 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
625 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET,
626 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
627 "Layer2 filter with IPFW");
630 * bridge_clone_create:
632 * Create a new bridge instance.
635 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
637 struct bridge_softc *sc, *sc2;
638 struct ifnet *bifp, *ifp;
640 unsigned long hostid;
642 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
643 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
649 BRIDGE_LOCK_INIT(sc);
650 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
651 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
653 /* Initialize our routing table. */
654 bridge_rtable_init(sc);
656 callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
658 LIST_INIT(&sc->sc_iflist);
659 LIST_INIT(&sc->sc_spanlist);
662 if_initname(ifp, bridge_name, unit);
663 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
664 ifp->if_ioctl = bridge_ioctl;
665 ifp->if_transmit = bridge_transmit;
666 ifp->if_qflush = bridge_qflush;
667 ifp->if_init = bridge_init;
668 ifp->if_type = IFT_BRIDGE;
671 * Generate an ethernet address with a locally administered address.
673 * Since we are using random ethernet addresses for the bridge, it is
674 * possible that we might have address collisions, so make sure that
675 * this hardware address isn't already in use on another bridge.
676 * The first try uses the hostid and falls back to arc4rand().
679 getcredhostid(curthread->td_ucred, &hostid);
681 if (fb || hostid == 0) {
682 ether_gen_addr(ifp, &sc->sc_defaddr);
684 sc->sc_defaddr.octet[0] = 0x2;
685 sc->sc_defaddr.octet[1] = (hostid >> 24) & 0xff;
686 sc->sc_defaddr.octet[2] = (hostid >> 16) & 0xff;
687 sc->sc_defaddr.octet[3] = (hostid >> 8 ) & 0xff;
688 sc->sc_defaddr.octet[4] = hostid & 0xff;
689 sc->sc_defaddr.octet[5] = ifp->if_dunit & 0xff;
695 LIST_FOREACH(sc2, &V_bridge_list, sc_list) {
697 if (memcmp(sc->sc_defaddr.octet,
698 IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
703 BRIDGE_LIST_UNLOCK();
704 } while (retry == 1);
706 bstp_attach(&sc->sc_stp, &bridge_ops);
707 ether_ifattach(ifp, sc->sc_defaddr.octet);
708 /* Now undo some of the damage... */
709 ifp->if_baudrate = 0;
710 ifp->if_type = IFT_BRIDGE;
713 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
714 BRIDGE_LIST_UNLOCK();
720 * bridge_clone_destroy:
722 * Destroy a bridge instance.
725 bridge_clone_destroy(struct ifnet *ifp)
727 struct bridge_softc *sc = ifp->if_softc;
728 struct bridge_iflist *bif;
733 ifp->if_flags &= ~IFF_UP;
735 while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
736 bridge_delete_member(sc, bif, 0);
738 while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
739 bridge_delete_span(sc, bif);
742 /* Tear down the routing table. */
743 bridge_rtable_fini(sc);
747 callout_drain(&sc->sc_brcallout);
750 LIST_REMOVE(sc, sc_list);
751 BRIDGE_LIST_UNLOCK();
753 bstp_detach(&sc->sc_stp);
757 BRIDGE_LOCK_DESTROY(sc);
764 * Handle a control request from the operator.
767 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
769 struct bridge_softc *sc = ifp->if_softc;
770 struct ifreq *ifr = (struct ifreq *)data;
771 struct bridge_iflist *bif;
772 struct thread *td = curthread;
774 struct ifbreq ifbreq;
775 struct ifbifconf ifbifconf;
776 struct ifbareq ifbareq;
777 struct ifbaconf ifbaconf;
778 struct ifbrparam ifbrparam;
779 struct ifbropreq ifbropreq;
781 struct ifdrv *ifd = (struct ifdrv *) data;
782 const struct bridge_control *bc;
783 int error = 0, oldmtu;
793 if (ifd->ifd_cmd >= bridge_control_table_size) {
797 bc = &bridge_control_table[ifd->ifd_cmd];
799 if (cmd == SIOCGDRVSPEC &&
800 (bc->bc_flags & BC_F_COPYOUT) == 0) {
804 else if (cmd == SIOCSDRVSPEC &&
805 (bc->bc_flags & BC_F_COPYOUT) != 0) {
810 if (bc->bc_flags & BC_F_SUSER) {
811 error = priv_check(td, PRIV_NET_BRIDGE);
816 if (ifd->ifd_len != bc->bc_argsize ||
817 ifd->ifd_len > sizeof(args)) {
822 bzero(&args, sizeof(args));
823 if (bc->bc_flags & BC_F_COPYIN) {
824 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
829 oldmtu = ifp->if_mtu;
831 error = (*bc->bc_func)(sc, &args);
837 * Bridge MTU may change during addition of the first port.
838 * If it did, do network layer specific procedure.
840 if (ifp->if_mtu != oldmtu) {
847 if (bc->bc_flags & BC_F_COPYOUT)
848 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
853 if (!(ifp->if_flags & IFF_UP) &&
854 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
856 * If interface is marked down and it is running,
857 * then stop and disable it.
862 } else if ((ifp->if_flags & IFF_UP) &&
863 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
865 * If interface is marked up and it is stopped, then
873 if (ifr->ifr_mtu < 576) {
877 if (LIST_EMPTY(&sc->sc_iflist)) {
878 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
882 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
883 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
884 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
885 " != %d\n", sc->sc_ifp->if_xname,
886 bif->bif_ifp->if_mtu,
887 bif->bif_ifp->if_xname, ifr->ifr_mtu);
893 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
898 * drop the lock as ether_ioctl() will call bridge_start() and
899 * cause the lock to be recursed.
901 error = ether_ioctl(ifp, cmd, data);
911 * Clear or restore unwanted capabilities on the member interface
914 bridge_mutecaps(struct bridge_softc *sc)
916 struct bridge_iflist *bif;
919 /* Initial bitmask of capabilities to test */
920 mask = BRIDGE_IFCAPS_MASK;
922 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
923 /* Every member must support it or its disabled */
924 mask &= bif->bif_savedcaps;
928 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
929 enabled = bif->bif_ifp->if_capenable;
930 enabled &= ~BRIDGE_IFCAPS_STRIP;
931 /* strip off mask bits and enable them again if allowed */
932 enabled &= ~BRIDGE_IFCAPS_MASK;
935 bridge_set_ifcap(sc, bif, enabled);
943 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
945 struct ifnet *ifp = bif->bif_ifp;
947 int error, mask, stuck;
949 BRIDGE_UNLOCK_ASSERT(sc);
951 bzero(&ifr, sizeof(ifr));
952 ifr.ifr_reqcap = set;
954 if (ifp->if_capenable != set) {
955 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
957 if_printf(sc->sc_ifp,
958 "error setting capabilities on %s: %d\n",
959 ifp->if_xname, error);
960 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
961 stuck = ifp->if_capenable & mask & ~set;
963 if_printf(sc->sc_ifp,
964 "can't disable some capabilities on %s: 0x%x\n",
965 ifp->if_xname, stuck);
970 * bridge_lookup_member:
972 * Lookup a bridge member interface.
974 static struct bridge_iflist *
975 bridge_lookup_member(struct bridge_softc *sc, const char *name)
977 struct bridge_iflist *bif;
980 BRIDGE_LOCK_ASSERT(sc);
982 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
984 if (strcmp(ifp->if_xname, name) == 0)
992 * bridge_lookup_member_if:
994 * Lookup a bridge member interface by ifnet*.
996 static struct bridge_iflist *
997 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
999 struct bridge_iflist *bif;
1001 BRIDGE_LOCK_ASSERT(sc);
1003 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1004 if (bif->bif_ifp == member_ifp)
1012 * bridge_delete_member:
1014 * Delete the specified member interface.
1017 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1020 struct ifnet *ifs = bif->bif_ifp;
1021 struct ifnet *fif = NULL;
1023 BRIDGE_LOCK_ASSERT(sc);
1025 if (bif->bif_flags & IFBIF_STP)
1026 bstp_disable(&bif->bif_stp);
1028 ifs->if_bridge = NULL;
1030 LIST_REMOVE(bif, bif_next);
1034 * If removing the interface that gave the bridge its mac address, set
1035 * the mac address of the bridge to the address of the next member, or
1036 * to its default address if no members are left.
1038 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1039 if (LIST_EMPTY(&sc->sc_iflist)) {
1040 bcopy(&sc->sc_defaddr,
1041 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1042 sc->sc_ifaddr = NULL;
1044 fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
1045 bcopy(IF_LLADDR(fif),
1046 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1047 sc->sc_ifaddr = fif;
1049 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1052 bridge_linkcheck(sc);
1053 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1054 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1055 KASSERT(bif->bif_addrcnt == 0,
1056 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1058 ifs->if_bridge_output = NULL;
1059 ifs->if_bridge_input = NULL;
1060 ifs->if_bridge_linkstate = NULL;
1063 switch (ifs->if_type) {
1067 * Take the interface out of promiscuous mode, but only
1068 * if it was promiscuous in the first place. It might
1069 * not be if we're in the bridge_ioctl_add() error path.
1071 if (ifs->if_flags & IFF_PROMISC)
1072 (void) ifpromisc(ifs, 0);
1080 panic("bridge_delete_member: impossible");
1084 /* reneable any interface capabilities */
1085 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1087 bstp_destroy(&bif->bif_stp); /* prepare to free */
1089 free(bif, M_DEVBUF);
1093 * bridge_delete_span:
1095 * Delete the specified span interface.
1098 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1100 BRIDGE_LOCK_ASSERT(sc);
1102 KASSERT(bif->bif_ifp->if_bridge == NULL,
1103 ("%s: not a span interface", __func__));
1105 LIST_REMOVE(bif, bif_next);
1106 free(bif, M_DEVBUF);
1110 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1112 struct ifbreq *req = arg;
1113 struct bridge_iflist *bif = NULL;
1117 ifs = ifunit(req->ifbr_ifsname);
1120 if (ifs->if_ioctl == NULL) /* must be supported */
1123 /* If it's in the span list, it can't be a member. */
1124 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1125 if (ifs == bif->bif_ifp)
1128 if (ifs->if_bridge == sc)
1131 if (ifs->if_bridge != NULL)
1134 switch (ifs->if_type) {
1138 /* permitted interface types */
1146 * Two valid inet6 addresses with link-local scope must not be
1147 * on the parent interface and the member interfaces at the
1148 * same time. This restriction is needed to prevent violation
1149 * of link-local scope zone. Attempts to add a member
1150 * interface which has inet6 addresses when the parent has
1151 * inet6 triggers removal of all inet6 addresses on the member
1155 /* Check if the parent interface has a link-local scope addr. */
1156 if (V_allow_llz_overlap == 0 &&
1157 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1159 * If any, remove all inet6 addresses from the member
1163 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1164 if (in6ifa_llaonifp(bif->bif_ifp)) {
1166 in6_ifdetach(bif->bif_ifp);
1168 if_printf(sc->sc_ifp,
1169 "IPv6 addresses on %s have been removed "
1170 "before adding it as a member to prevent "
1171 "IPv6 address scope violation.\n",
1172 bif->bif_ifp->if_xname);
1176 if (in6ifa_llaonifp(ifs)) {
1180 if_printf(sc->sc_ifp,
1181 "IPv6 addresses on %s have been removed "
1182 "before adding it as a member to prevent "
1183 "IPv6 address scope violation.\n",
1188 /* Allow the first Ethernet member to define the MTU */
1189 if (LIST_EMPTY(&sc->sc_iflist))
1190 sc->sc_ifp->if_mtu = ifs->if_mtu;
1191 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1192 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1193 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1197 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1202 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1203 bif->bif_savedcaps = ifs->if_capenable;
1206 * Assign the interface's MAC address to the bridge if it's the first
1207 * member and the MAC address of the bridge has not been changed from
1208 * the default randomly generated one.
1210 if (V_bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1211 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1212 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1213 sc->sc_ifaddr = ifs;
1214 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1217 ifs->if_bridge = sc;
1218 ifs->if_bridge_output = bridge_output;
1219 ifs->if_bridge_input = bridge_input;
1220 ifs->if_bridge_linkstate = bridge_linkstate;
1221 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1223 * XXX: XLOCK HERE!?!
1225 * NOTE: insert_***HEAD*** should be safe for the traversals.
1227 LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1229 /* Set interface capabilities to the intersection set of all members */
1230 bridge_mutecaps(sc);
1231 bridge_linkcheck(sc);
1233 /* Place the interface into promiscuous mode */
1234 switch (ifs->if_type) {
1238 error = ifpromisc(ifs, 1);
1244 bridge_delete_member(sc, bif, 0);
1249 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1251 struct ifbreq *req = arg;
1252 struct bridge_iflist *bif;
1254 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1258 bridge_delete_member(sc, bif, 0);
1264 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1266 struct ifbreq *req = arg;
1267 struct bridge_iflist *bif;
1268 struct bstp_port *bp;
1270 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1275 req->ifbr_ifsflags = bif->bif_flags;
1276 req->ifbr_state = bp->bp_state;
1277 req->ifbr_priority = bp->bp_priority;
1278 req->ifbr_path_cost = bp->bp_path_cost;
1279 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1280 req->ifbr_proto = bp->bp_protover;
1281 req->ifbr_role = bp->bp_role;
1282 req->ifbr_stpflags = bp->bp_flags;
1283 req->ifbr_addrcnt = bif->bif_addrcnt;
1284 req->ifbr_addrmax = bif->bif_addrmax;
1285 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1287 /* Copy STP state options as flags */
1288 if (bp->bp_operedge)
1289 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1290 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1291 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1292 if (bp->bp_ptp_link)
1293 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1294 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1295 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1296 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1297 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1298 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1299 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1304 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1306 struct ifbreq *req = arg;
1307 struct bridge_iflist *bif;
1308 struct bstp_port *bp;
1311 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1316 if (req->ifbr_ifsflags & IFBIF_SPAN)
1317 /* SPAN is readonly */
1320 if (req->ifbr_ifsflags & IFBIF_STP) {
1321 if ((bif->bif_flags & IFBIF_STP) == 0) {
1322 error = bstp_enable(&bif->bif_stp);
1327 if ((bif->bif_flags & IFBIF_STP) != 0)
1328 bstp_disable(&bif->bif_stp);
1331 /* Pass on STP flags */
1332 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1333 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1334 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1335 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1337 /* Save the bits relating to the bridge */
1338 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1344 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1346 struct ifbrparam *param = arg;
1348 sc->sc_brtmax = param->ifbrp_csize;
1355 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1357 struct ifbrparam *param = arg;
1359 param->ifbrp_csize = sc->sc_brtmax;
1365 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1367 struct ifbifconf *bifc = arg;
1368 struct bridge_iflist *bif;
1371 int count, buflen, len, error = 0;
1374 LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1376 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1379 buflen = sizeof(breq) * count;
1380 if (bifc->ifbic_len == 0) {
1381 bifc->ifbic_len = buflen;
1385 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1390 len = min(bifc->ifbic_len, buflen);
1391 bzero(&breq, sizeof(breq));
1392 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1393 if (len < sizeof(breq))
1396 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1397 sizeof(breq.ifbr_ifsname));
1398 /* Fill in the ifbreq structure */
1399 error = bridge_ioctl_gifflags(sc, &breq);
1402 memcpy(buf, &breq, sizeof(breq));
1404 buf += sizeof(breq);
1405 len -= sizeof(breq);
1407 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1408 if (len < sizeof(breq))
1411 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1412 sizeof(breq.ifbr_ifsname));
1413 breq.ifbr_ifsflags = bif->bif_flags;
1414 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1415 memcpy(buf, &breq, sizeof(breq));
1417 buf += sizeof(breq);
1418 len -= sizeof(breq);
1422 bifc->ifbic_len = sizeof(breq) * count;
1423 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1425 free(outbuf, M_TEMP);
1430 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1432 struct ifbaconf *bac = arg;
1433 struct bridge_rtnode *brt;
1434 struct ifbareq bareq;
1436 int count, buflen, len, error = 0;
1438 if (bac->ifbac_len == 0)
1442 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1444 buflen = sizeof(bareq) * count;
1447 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1452 len = min(bac->ifbac_len, buflen);
1453 bzero(&bareq, sizeof(bareq));
1454 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1455 if (len < sizeof(bareq))
1457 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1458 sizeof(bareq.ifba_ifsname));
1459 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1460 bareq.ifba_vlan = brt->brt_vlan;
1461 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1462 time_uptime < brt->brt_expire)
1463 bareq.ifba_expire = brt->brt_expire - time_uptime;
1465 bareq.ifba_expire = 0;
1466 bareq.ifba_flags = brt->brt_flags;
1468 memcpy(buf, &bareq, sizeof(bareq));
1470 buf += sizeof(bareq);
1471 len -= sizeof(bareq);
1475 bac->ifbac_len = sizeof(bareq) * count;
1476 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1478 free(outbuf, M_TEMP);
1483 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1485 struct ifbareq *req = arg;
1486 struct bridge_iflist *bif;
1489 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1493 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1500 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1502 struct ifbrparam *param = arg;
1504 sc->sc_brttimeout = param->ifbrp_ctime;
1509 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1511 struct ifbrparam *param = arg;
1513 param->ifbrp_ctime = sc->sc_brttimeout;
1518 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1520 struct ifbareq *req = arg;
1522 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1526 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1528 struct ifbreq *req = arg;
1530 bridge_rtflush(sc, req->ifbr_ifsflags);
1535 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1537 struct ifbrparam *param = arg;
1538 struct bstp_state *bs = &sc->sc_stp;
1540 param->ifbrp_prio = bs->bs_bridge_priority;
1545 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1547 struct ifbrparam *param = arg;
1549 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1553 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1555 struct ifbrparam *param = arg;
1556 struct bstp_state *bs = &sc->sc_stp;
1558 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1563 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1565 struct ifbrparam *param = arg;
1567 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1571 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1573 struct ifbrparam *param = arg;
1574 struct bstp_state *bs = &sc->sc_stp;
1576 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1581 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1583 struct ifbrparam *param = arg;
1585 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1589 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1591 struct ifbrparam *param = arg;
1592 struct bstp_state *bs = &sc->sc_stp;
1594 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1599 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1601 struct ifbrparam *param = arg;
1603 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1607 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1609 struct ifbreq *req = arg;
1610 struct bridge_iflist *bif;
1612 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1616 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1620 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1622 struct ifbreq *req = arg;
1623 struct bridge_iflist *bif;
1625 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1629 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1633 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1635 struct ifbreq *req = arg;
1636 struct bridge_iflist *bif;
1638 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1642 bif->bif_addrmax = req->ifbr_addrmax;
1647 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1649 struct ifbreq *req = arg;
1650 struct bridge_iflist *bif = NULL;
1653 ifs = ifunit(req->ifbr_ifsname);
1657 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1658 if (ifs == bif->bif_ifp)
1661 if (ifs->if_bridge != NULL)
1664 switch (ifs->if_type) {
1673 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1678 bif->bif_flags = IFBIF_SPAN;
1680 LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1686 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1688 struct ifbreq *req = arg;
1689 struct bridge_iflist *bif;
1692 ifs = ifunit(req->ifbr_ifsname);
1696 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1697 if (ifs == bif->bif_ifp)
1703 bridge_delete_span(sc, bif);
1709 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1711 struct ifbropreq *req = arg;
1712 struct bstp_state *bs = &sc->sc_stp;
1713 struct bstp_port *root_port;
1715 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1716 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1717 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1719 root_port = bs->bs_root_port;
1720 if (root_port == NULL)
1721 req->ifbop_root_port = 0;
1723 req->ifbop_root_port = root_port->bp_ifp->if_index;
1725 req->ifbop_holdcount = bs->bs_txholdcount;
1726 req->ifbop_priority = bs->bs_bridge_priority;
1727 req->ifbop_protocol = bs->bs_protover;
1728 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1729 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1730 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1731 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1732 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1733 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1739 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1741 struct ifbrparam *param = arg;
1743 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1748 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1750 struct ifbpstpconf *bifstp = arg;
1751 struct bridge_iflist *bif;
1752 struct bstp_port *bp;
1753 struct ifbpstpreq bpreq;
1755 int count, buflen, len, error = 0;
1758 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1759 if ((bif->bif_flags & IFBIF_STP) != 0)
1763 buflen = sizeof(bpreq) * count;
1764 if (bifstp->ifbpstp_len == 0) {
1765 bifstp->ifbpstp_len = buflen;
1770 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1775 len = min(bifstp->ifbpstp_len, buflen);
1776 bzero(&bpreq, sizeof(bpreq));
1777 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1778 if (len < sizeof(bpreq))
1781 if ((bif->bif_flags & IFBIF_STP) == 0)
1785 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1786 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1787 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1788 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1789 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1790 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1792 memcpy(buf, &bpreq, sizeof(bpreq));
1794 buf += sizeof(bpreq);
1795 len -= sizeof(bpreq);
1799 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1800 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1802 free(outbuf, M_TEMP);
1807 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1809 struct ifbrparam *param = arg;
1811 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1815 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1817 struct ifbrparam *param = arg;
1819 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1825 * Detach an interface from a bridge. Called when a member
1826 * interface is detaching.
1829 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1831 struct bridge_softc *sc = ifp->if_bridge;
1832 struct bridge_iflist *bif;
1834 if (ifp->if_flags & IFF_RENAMING)
1836 if (V_bridge_cloner == NULL) {
1838 * This detach handler can be called after
1839 * vnet_bridge_uninit(). Just return in that case.
1843 /* Check if the interface is a bridge member */
1847 bif = bridge_lookup_member_if(sc, ifp);
1849 bridge_delete_member(sc, bif, 1);
1855 /* Check if the interface is a span port */
1857 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1859 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1860 if (ifp == bif->bif_ifp) {
1861 bridge_delete_span(sc, bif);
1867 BRIDGE_LIST_UNLOCK();
1873 * Initialize a bridge interface.
1876 bridge_init(void *xsc)
1878 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1879 struct ifnet *ifp = sc->sc_ifp;
1881 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1885 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1888 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1889 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1897 * Stop the bridge interface.
1900 bridge_stop(struct ifnet *ifp, int disable)
1902 struct bridge_softc *sc = ifp->if_softc;
1904 BRIDGE_LOCK_ASSERT(sc);
1906 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1909 callout_stop(&sc->sc_brcallout);
1910 bstp_stop(&sc->sc_stp);
1912 bridge_rtflush(sc, IFBF_FLUSHDYN);
1914 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1920 * Enqueue a packet on a bridge member interface.
1924 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1930 /* We may be sending a fragment so traverse the mbuf */
1933 m->m_nextpkt = NULL;
1934 len = m->m_pkthdr.len;
1935 mflags = m->m_flags;
1938 * If underlying interface can not do VLAN tag insertion itself
1939 * then attach a packet tag that holds it.
1941 if ((m->m_flags & M_VLANTAG) &&
1942 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1943 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1946 "unable to prepend VLAN header\n");
1947 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
1950 m->m_flags &= ~M_VLANTAG;
1953 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
1954 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1956 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
1960 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
1961 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
1962 if (mflags & M_MCAST)
1963 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
1972 * Receive a queued packet from dummynet and pass it on to the output
1975 * The mbuf has the Ethernet header already attached.
1978 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1980 struct bridge_softc *sc;
1982 sc = ifp->if_bridge;
1985 * The packet didnt originate from a member interface. This should only
1986 * ever happen if a member interface is removed while packets are
1994 if (PFIL_HOOKED(&V_inet_pfil_hook)
1996 || PFIL_HOOKED(&V_inet6_pfil_hook)
1999 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2005 bridge_enqueue(sc, ifp, m);
2011 * Send output from a bridge member interface. This
2012 * performs the bridging function for locally originated
2015 * The mbuf has the Ethernet header already attached. We must
2016 * enqueue or free the mbuf before returning.
2019 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2022 struct ether_header *eh;
2023 struct ifnet *dst_if;
2024 struct bridge_softc *sc;
2027 if (m->m_len < ETHER_HDR_LEN) {
2028 m = m_pullup(m, ETHER_HDR_LEN);
2033 eh = mtod(m, struct ether_header *);
2034 sc = ifp->if_bridge;
2035 vlan = VLANTAGOF(m);
2040 * If bridge is down, but the original output interface is up,
2041 * go ahead and send out that interface. Otherwise, the packet
2044 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2050 * If the packet is a multicast, or we don't know a better way to
2051 * get there, send to all interfaces.
2053 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2056 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2057 if (dst_if == NULL) {
2058 struct bridge_iflist *bif;
2060 int error = 0, used = 0;
2064 BRIDGE_LOCK2REF(sc, error);
2070 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2071 dst_if = bif->bif_ifp;
2073 if (dst_if->if_type == IFT_GIF)
2075 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2079 * If this is not the original output interface,
2080 * and the interface is participating in spanning
2081 * tree, make sure the port is in a state that
2082 * allows forwarding.
2084 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2085 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2088 if (LIST_NEXT(bif, bif_next) == NULL) {
2092 mc = m_copypacket(m, M_NOWAIT);
2094 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2099 bridge_enqueue(sc, dst_if, mc);
2109 * XXX Spanning tree consideration here?
2113 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2120 bridge_enqueue(sc, dst_if, m);
2127 * Do output on a bridge.
2131 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2133 struct bridge_softc *sc;
2134 struct ether_header *eh;
2135 struct ifnet *dst_if;
2140 ETHER_BPF_MTAP(ifp, m);
2142 eh = mtod(m, struct ether_header *);
2145 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2146 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2148 error = bridge_enqueue(sc, dst_if, m);
2150 bridge_broadcast(sc, ifp, m, 0);
2156 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2159 bridge_qflush(struct ifnet *ifp __unused)
2166 * The forwarding function of the bridge.
2168 * NOTE: Releases the lock on return.
2171 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2174 struct bridge_iflist *dbif;
2175 struct ifnet *src_if, *dst_if, *ifp;
2176 struct ether_header *eh;
2181 src_if = m->m_pkthdr.rcvif;
2184 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2185 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2186 vlan = VLANTAGOF(m);
2188 if ((sbif->bif_flags & IFBIF_STP) &&
2189 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2192 eh = mtod(m, struct ether_header *);
2193 dst = eh->ether_dhost;
2195 /* If the interface is learning, record the address. */
2196 if (sbif->bif_flags & IFBIF_LEARNING) {
2197 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2198 sbif, 0, IFBAF_DYNAMIC);
2200 * If the interface has addresses limits then deny any source
2201 * that is not in the cache.
2203 if (error && sbif->bif_addrmax)
2207 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2208 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2212 * At this point, the port either doesn't participate
2213 * in spanning tree or it is in the forwarding state.
2217 * If the packet is unicast, destined for someone on
2218 * "this" side of the bridge, drop it.
2220 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2221 dst_if = bridge_rtlookup(sc, dst, vlan);
2222 if (src_if == dst_if)
2226 * Check if its a reserved multicast address, any address
2227 * listed in 802.1D section 7.12.6 may not be forwarded by the
2229 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2231 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2232 dst[2] == 0xc2 && dst[3] == 0x00 &&
2233 dst[4] == 0x00 && dst[5] <= 0x0f)
2236 /* ...forward it to all interfaces. */
2237 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2242 * If we have a destination interface which is a member of our bridge,
2243 * OR this is a unicast packet, push it through the bpf(4) machinery.
2244 * For broadcast or multicast packets, don't bother because it will
2245 * be reinjected into ether_input. We do this before we pass the packets
2246 * through the pfil(9) framework, as it is possible that pfil(9) will
2247 * drop the packet, or possibly modify it, making it difficult to debug
2248 * firewall issues on the bridge.
2250 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2251 ETHER_BPF_MTAP(ifp, m);
2253 /* run the packet filter */
2254 if (PFIL_HOOKED(&V_inet_pfil_hook)
2256 || PFIL_HOOKED(&V_inet6_pfil_hook)
2260 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2267 if (dst_if == NULL) {
2268 bridge_broadcast(sc, src_if, m, 1);
2273 * At this point, we're dealing with a unicast frame
2274 * going to a different interface.
2276 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2279 dbif = bridge_lookup_member_if(sc, dst_if);
2281 /* Not a member of the bridge (anymore?) */
2284 /* Private segments can not talk to each other */
2285 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2288 if ((dbif->bif_flags & IFBIF_STP) &&
2289 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2294 if (PFIL_HOOKED(&V_inet_pfil_hook)
2296 || PFIL_HOOKED(&V_inet6_pfil_hook)
2299 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2305 bridge_enqueue(sc, dst_if, m);
2316 * Receive input from a member interface. Queue the packet for
2317 * bridging if it is not for us.
2319 static struct mbuf *
2320 bridge_input(struct ifnet *ifp, struct mbuf *m)
2322 struct bridge_softc *sc = ifp->if_bridge;
2323 struct bridge_iflist *bif, *bif2;
2325 struct ether_header *eh;
2326 struct mbuf *mc, *mc2;
2330 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2334 vlan = VLANTAGOF(m);
2337 * Implement support for bridge monitoring. If this flag has been
2338 * set on this interface, discard the packet once we push it through
2339 * the bpf(4) machinery, but before we do, increment the byte and
2340 * packet counters associated with this interface.
2342 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2343 m->m_pkthdr.rcvif = bifp;
2344 ETHER_BPF_MTAP(bifp, m);
2345 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2346 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2351 bif = bridge_lookup_member_if(sc, ifp);
2357 eh = mtod(m, struct ether_header *);
2361 if (m->m_flags & (M_BCAST|M_MCAST)) {
2362 /* Tap off 802.1D packets; they do not get forwarded. */
2363 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2364 ETHER_ADDR_LEN) == 0) {
2365 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2370 if ((bif->bif_flags & IFBIF_STP) &&
2371 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2377 * Make a deep copy of the packet and enqueue the copy
2378 * for bridge processing; return the original packet for
2381 mc = m_dup(m, M_NOWAIT);
2387 /* Perform the bridge forwarding function with the copy. */
2388 bridge_forward(sc, bif, mc);
2391 * Reinject the mbuf as arriving on the bridge so we have a
2392 * chance at claiming multicast packets. We can not loop back
2393 * here from ether_input as a bridge is never a member of a
2396 KASSERT(bifp->if_bridge == NULL,
2397 ("loop created in bridge_input"));
2398 mc2 = m_dup(m, M_NOWAIT);
2400 /* Keep the layer3 header aligned */
2401 int i = min(mc2->m_pkthdr.len, max_protohdr);
2402 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2405 mc2->m_pkthdr.rcvif = bifp;
2406 (*bifp->if_input)(bifp, mc2);
2409 /* Return the original packet for local processing. */
2413 if ((bif->bif_flags & IFBIF_STP) &&
2414 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2419 #if (defined(INET) || defined(INET6))
2420 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2421 || ((iface)->if_carp \
2422 && (*carp_forus_p)((iface), eh->ether_dhost))
2423 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2424 || ((iface)->if_carp \
2425 && (*carp_forus_p)((iface), eh->ether_shost))
2427 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2428 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2432 # define OR_PFIL_HOOKED_INET6 \
2433 || PFIL_HOOKED(&V_inet6_pfil_hook)
2435 # define OR_PFIL_HOOKED_INET6
2438 #define GRAB_OUR_PACKETS(iface) \
2439 if ((iface)->if_type == IFT_GIF) \
2441 /* It is destined for us. */ \
2442 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2443 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2445 if ((iface)->if_type == IFT_BRIDGE) { \
2446 ETHER_BPF_MTAP(iface, m); \
2447 if_inc_counter(iface, IFCOUNTER_IPACKETS, 1); \
2448 if_inc_counter(iface, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2449 /* Filter on the physical interface. */ \
2450 if (V_pfil_local_phys && \
2451 (PFIL_HOOKED(&V_inet_pfil_hook) \
2452 OR_PFIL_HOOKED_INET6)) { \
2453 if (bridge_pfil(&m, NULL, ifp, \
2454 PFIL_IN) != 0 || m == NULL) { \
2455 BRIDGE_UNLOCK(sc); \
2458 eh = mtod(m, struct ether_header *); \
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) { \
2465 BRIDGE_UNLOCK(sc); \
2470 m->m_pkthdr.rcvif = iface; \
2471 BRIDGE_UNLOCK(sc); \
2475 /* We just received a packet that we sent out. */ \
2476 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2477 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2479 BRIDGE_UNLOCK(sc); \
2485 * Unicast. Make sure it's not for the bridge.
2487 do { GRAB_OUR_PACKETS(bifp) } while (0);
2490 * Give a chance for ifp at first priority. This will help when the
2491 * packet comes through the interface like VLAN's with the same MACs
2492 * on several interfaces from the same bridge. This also will save
2493 * some CPU cycles in case the destination interface and the input
2494 * interface (eq ifp) are the same.
2496 do { GRAB_OUR_PACKETS(ifp) } while (0);
2498 /* Now check the all bridge members. */
2499 LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2500 GRAB_OUR_PACKETS(bif2->bif_ifp)
2503 #undef OR_CARP_CHECK_WE_ARE_DST
2504 #undef OR_CARP_CHECK_WE_ARE_SRC
2505 #undef OR_PFIL_HOOKED_INET6
2506 #undef GRAB_OUR_PACKETS
2508 /* Perform the bridge forwarding function. */
2509 bridge_forward(sc, bif, m);
2517 * Send a frame to all interfaces that are members of
2518 * the bridge, except for the one on which the packet
2521 * NOTE: Releases the lock on return.
2524 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2525 struct mbuf *m, int runfilt)
2527 struct bridge_iflist *dbif, *sbif;
2529 struct ifnet *dst_if;
2530 int error = 0, used = 0, i;
2532 sbif = bridge_lookup_member_if(sc, src_if);
2534 BRIDGE_LOCK2REF(sc, error);
2540 /* Filter on the bridge interface before broadcasting */
2541 if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2543 || PFIL_HOOKED(&V_inet6_pfil_hook)
2546 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2552 LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2553 dst_if = dbif->bif_ifp;
2554 if (dst_if == src_if)
2557 /* Private segments can not talk to each other */
2558 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2561 if ((dbif->bif_flags & IFBIF_STP) &&
2562 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2565 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2566 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2569 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2572 if (LIST_NEXT(dbif, bif_next) == NULL) {
2576 mc = m_dup(m, M_NOWAIT);
2578 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2584 * Filter on the output interface. Pass a NULL bridge interface
2585 * pointer so we do not redundantly filter on the bridge for
2586 * each interface we broadcast on.
2588 if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2590 || PFIL_HOOKED(&V_inet6_pfil_hook)
2594 /* Keep the layer3 header aligned */
2595 i = min(mc->m_pkthdr.len, max_protohdr);
2596 mc = m_copyup(mc, i, ETHER_ALIGN);
2598 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2602 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2608 bridge_enqueue(sc, dst_if, mc);
2620 * Duplicate a packet out one or more interfaces that are in span mode,
2621 * the original mbuf is unmodified.
2624 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2626 struct bridge_iflist *bif;
2627 struct ifnet *dst_if;
2630 if (LIST_EMPTY(&sc->sc_spanlist))
2633 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2634 dst_if = bif->bif_ifp;
2636 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2639 mc = m_copypacket(m, M_NOWAIT);
2641 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2645 bridge_enqueue(sc, dst_if, mc);
2652 * Add a bridge routing entry.
2655 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2656 struct bridge_iflist *bif, int setflags, uint8_t flags)
2658 struct bridge_rtnode *brt;
2661 BRIDGE_LOCK_ASSERT(sc);
2663 /* Check the source address is valid and not multicast. */
2664 if (ETHER_IS_MULTICAST(dst) ||
2665 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2666 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2669 /* 802.1p frames map to vlan 1 */
2674 * A route for this destination might already exist. If so,
2675 * update it, otherwise create a new one.
2677 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2678 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2679 sc->sc_brtexceeded++;
2682 /* Check per interface address limits (if enabled) */
2683 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2684 bif->bif_addrexceeded++;
2689 * Allocate a new bridge forwarding node, and
2690 * initialize the expiration time and Ethernet
2693 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2697 if (bif->bif_flags & IFBIF_STICKY)
2698 brt->brt_flags = IFBAF_STICKY;
2700 brt->brt_flags = IFBAF_DYNAMIC;
2702 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2703 brt->brt_vlan = vlan;
2705 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2706 uma_zfree(V_bridge_rtnode_zone, brt);
2713 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2714 brt->brt_dst != bif) {
2715 brt->brt_dst->bif_addrcnt--;
2717 brt->brt_dst->bif_addrcnt++;
2720 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2721 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2723 brt->brt_flags = flags;
2731 * Lookup the destination interface for an address.
2733 static struct ifnet *
2734 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2736 struct bridge_rtnode *brt;
2738 BRIDGE_LOCK_ASSERT(sc);
2740 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2743 return (brt->brt_ifp);
2749 * Trim the routine table so that we have a number
2750 * of routing entries less than or equal to the
2754 bridge_rttrim(struct bridge_softc *sc)
2756 struct bridge_rtnode *brt, *nbrt;
2758 BRIDGE_LOCK_ASSERT(sc);
2760 /* Make sure we actually need to do this. */
2761 if (sc->sc_brtcnt <= sc->sc_brtmax)
2764 /* Force an aging cycle; this might trim enough addresses. */
2766 if (sc->sc_brtcnt <= sc->sc_brtmax)
2769 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2770 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2771 bridge_rtnode_destroy(sc, brt);
2772 if (sc->sc_brtcnt <= sc->sc_brtmax)
2781 * Aging timer for the bridge.
2784 bridge_timer(void *arg)
2786 struct bridge_softc *sc = arg;
2788 BRIDGE_LOCK_ASSERT(sc);
2790 /* Destruction of rtnodes requires a proper vnet context */
2791 CURVNET_SET(sc->sc_ifp->if_vnet);
2794 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2795 callout_reset(&sc->sc_brcallout,
2796 bridge_rtable_prune_period * hz, bridge_timer, sc);
2803 * Perform an aging cycle.
2806 bridge_rtage(struct bridge_softc *sc)
2808 struct bridge_rtnode *brt, *nbrt;
2810 BRIDGE_LOCK_ASSERT(sc);
2812 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2813 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2814 if (time_uptime >= brt->brt_expire)
2815 bridge_rtnode_destroy(sc, brt);
2823 * Remove all dynamic addresses from the bridge.
2826 bridge_rtflush(struct bridge_softc *sc, int full)
2828 struct bridge_rtnode *brt, *nbrt;
2830 BRIDGE_LOCK_ASSERT(sc);
2832 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2833 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2834 bridge_rtnode_destroy(sc, brt);
2841 * Remove an address from the table.
2844 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2846 struct bridge_rtnode *brt;
2849 BRIDGE_LOCK_ASSERT(sc);
2852 * If vlan is zero then we want to delete for all vlans so the lookup
2853 * may return more than one.
2855 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2856 bridge_rtnode_destroy(sc, brt);
2860 return (found ? 0 : ENOENT);
2866 * Delete routes to a speicifc member interface.
2869 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2871 struct bridge_rtnode *brt, *nbrt;
2873 BRIDGE_LOCK_ASSERT(sc);
2875 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2876 if (brt->brt_ifp == ifp && (full ||
2877 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2878 bridge_rtnode_destroy(sc, brt);
2883 * bridge_rtable_init:
2885 * Initialize the route table for this bridge.
2888 bridge_rtable_init(struct bridge_softc *sc)
2892 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2893 M_DEVBUF, M_WAITOK);
2895 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2896 LIST_INIT(&sc->sc_rthash[i]);
2898 sc->sc_rthash_key = arc4random();
2899 LIST_INIT(&sc->sc_rtlist);
2903 * bridge_rtable_fini:
2905 * Deconstruct the route table for this bridge.
2908 bridge_rtable_fini(struct bridge_softc *sc)
2911 KASSERT(sc->sc_brtcnt == 0,
2912 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2913 free(sc->sc_rthash, M_DEVBUF);
2917 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2918 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2920 #define mix(a, b, c) \
2922 a -= b; a -= c; a ^= (c >> 13); \
2923 b -= c; b -= a; b ^= (a << 8); \
2924 c -= a; c -= b; c ^= (b >> 13); \
2925 a -= b; a -= c; a ^= (c >> 12); \
2926 b -= c; b -= a; b ^= (a << 16); \
2927 c -= a; c -= b; c ^= (b >> 5); \
2928 a -= b; a -= c; a ^= (c >> 3); \
2929 b -= c; b -= a; b ^= (a << 10); \
2930 c -= a; c -= b; c ^= (b >> 15); \
2931 } while (/*CONSTCOND*/0)
2933 static __inline uint32_t
2934 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2936 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2947 return (c & BRIDGE_RTHASH_MASK);
2953 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2957 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2958 d = ((int)a[i]) - ((int)b[i]);
2965 * bridge_rtnode_lookup:
2967 * Look up a bridge route node for the specified destination. Compare the
2968 * vlan id or if zero then just return the first match.
2970 static struct bridge_rtnode *
2971 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2973 struct bridge_rtnode *brt;
2977 BRIDGE_LOCK_ASSERT(sc);
2979 hash = bridge_rthash(sc, addr);
2980 LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2981 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2982 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2992 * bridge_rtnode_insert:
2994 * Insert the specified bridge node into the route table. We
2995 * assume the entry is not already in the table.
2998 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3000 struct bridge_rtnode *lbrt;
3004 BRIDGE_LOCK_ASSERT(sc);
3006 hash = bridge_rthash(sc, brt->brt_addr);
3008 lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
3010 LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3015 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3016 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3019 LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3022 if (LIST_NEXT(lbrt, brt_hash) == NULL) {
3023 LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3026 lbrt = LIST_NEXT(lbrt, brt_hash);
3027 } while (lbrt != NULL);
3030 panic("bridge_rtnode_insert: impossible");
3034 LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3041 * bridge_rtnode_destroy:
3043 * Destroy a bridge rtnode.
3046 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3048 BRIDGE_LOCK_ASSERT(sc);
3050 LIST_REMOVE(brt, brt_hash);
3052 LIST_REMOVE(brt, brt_list);
3054 brt->brt_dst->bif_addrcnt--;
3055 uma_zfree(V_bridge_rtnode_zone, brt);
3059 * bridge_rtable_expire:
3061 * Set the expiry time for all routes on an interface.
3064 bridge_rtable_expire(struct ifnet *ifp, int age)
3066 struct bridge_softc *sc = ifp->if_bridge;
3067 struct bridge_rtnode *brt;
3069 CURVNET_SET(ifp->if_vnet);
3073 * If the age is zero then flush, otherwise set all the expiry times to
3074 * age for the interface
3077 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3079 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3080 /* Cap the expiry time to 'age' */
3081 if (brt->brt_ifp == ifp &&
3082 brt->brt_expire > time_uptime + age &&
3083 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3084 brt->brt_expire = time_uptime + age;
3092 * bridge_state_change:
3094 * Callback from the bridgestp code when a port changes states.
3097 bridge_state_change(struct ifnet *ifp, int state)
3099 struct bridge_softc *sc = ifp->if_bridge;
3100 static const char *stpstates[] = {
3109 CURVNET_SET(ifp->if_vnet);
3111 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3112 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3117 * Send bridge packets through pfil if they are one of the types pfil can deal
3118 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3119 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3123 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3125 int snap, error, i, hlen;
3126 struct ether_header *eh1, eh2;
3129 u_int16_t ether_type;
3132 error = -1; /* Default error if not error == 0 */
3135 /* we may return with the IP fields swapped, ensure its not shared */
3136 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3139 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3140 return (0); /* filtering is disabled */
3142 i = min((*mp)->m_pkthdr.len, max_protohdr);
3143 if ((*mp)->m_len < i) {
3144 *mp = m_pullup(*mp, i);
3146 printf("%s: m_pullup failed\n", __func__);
3151 eh1 = mtod(*mp, struct ether_header *);
3152 ether_type = ntohs(eh1->ether_type);
3155 * Check for SNAP/LLC.
3157 if (ether_type < ETHERMTU) {
3158 struct llc *llc2 = (struct llc *)(eh1 + 1);
3160 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3161 llc2->llc_dsap == LLC_SNAP_LSAP &&
3162 llc2->llc_ssap == LLC_SNAP_LSAP &&
3163 llc2->llc_control == LLC_UI) {
3164 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3170 * If we're trying to filter bridge traffic, don't look at anything
3171 * other than IP and ARP traffic. If the filter doesn't understand
3172 * IPv6, don't allow IPv6 through the bridge either. This is lame
3173 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3174 * but of course we don't have an AppleTalk filter to begin with.
3175 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3178 switch (ether_type) {
3180 case ETHERTYPE_REVARP:
3181 if (V_pfil_ipfw_arp == 0)
3182 return (0); /* Automatically pass */
3187 case ETHERTYPE_IPV6:
3192 * Check to see if the user wants to pass non-ip
3193 * packets, these will not be checked by pfil(9) and
3194 * passed unconditionally so the default is to drop.
3200 /* Run the packet through pfil before stripping link headers */
3201 if (PFIL_HOOKED(&V_link_pfil_hook) && V_pfil_ipfw != 0 &&
3202 dir == PFIL_OUT && ifp != NULL) {
3204 error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, 0,
3207 if (*mp == NULL || error != 0) /* packet consumed by filter */
3211 /* Strip off the Ethernet header and keep a copy. */
3212 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3213 m_adj(*mp, ETHER_HDR_LEN);
3215 /* Strip off snap header, if present */
3217 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3218 m_adj(*mp, sizeof(struct llc));
3222 * Check the IP header for alignment and errors
3224 if (dir == PFIL_IN) {
3225 switch (ether_type) {
3227 error = bridge_ip_checkbasic(mp);
3230 case ETHERTYPE_IPV6:
3231 error = bridge_ip6_checkbasic(mp);
3244 * Run the packet through pfil
3246 switch (ether_type) {
3249 * Run pfil on the member interface and the bridge, both can
3250 * be skipped by clearing pfil_member or pfil_bridge.
3253 * in_if -> bridge_if -> out_if
3255 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3256 error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3259 if (*mp == NULL || error != 0) /* filter may consume */
3262 if (V_pfil_member && ifp != NULL)
3263 error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3266 if (*mp == NULL || error != 0) /* filter may consume */
3269 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3270 error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3273 if (*mp == NULL || error != 0) /* filter may consume */
3276 /* check if we need to fragment the packet */
3277 /* bridge_fragment generates a mbuf chain of packets */
3278 /* that already include eth headers */
3279 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3280 i = (*mp)->m_pkthdr.len;
3281 if (i > ifp->if_mtu) {
3282 error = bridge_fragment(ifp, mp, &eh2, snap,
3288 /* Recalculate the ip checksum. */
3289 ip = mtod(*mp, struct ip *);
3290 hlen = ip->ip_hl << 2;
3291 if (hlen < sizeof(struct ip))
3293 if (hlen > (*mp)->m_len) {
3294 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3296 ip = mtod(*mp, struct ip *);
3301 if (hlen == sizeof(struct ip))
3302 ip->ip_sum = in_cksum_hdr(ip);
3304 ip->ip_sum = in_cksum(*mp, hlen);
3308 case ETHERTYPE_IPV6:
3309 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3310 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3313 if (*mp == NULL || error != 0) /* filter may consume */
3316 if (V_pfil_member && ifp != NULL)
3317 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3320 if (*mp == NULL || error != 0) /* filter may consume */
3323 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3324 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3341 * Finally, put everything back the way it was and return
3344 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3347 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3350 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3353 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3364 * Perform basic checks on header size since
3365 * pfil assumes ip_input has already processed
3366 * it for it. Cut-and-pasted from ip_input.c.
3367 * Given how simple the IPv6 version is,
3368 * does the IPv4 version really need to be
3371 * XXX Should we update ipstat here, or not?
3372 * XXX Right now we update ipstat but not
3376 bridge_ip_checkbasic(struct mbuf **mp)
3378 struct mbuf *m = *mp;
3386 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3387 if ((m = m_copyup(m, sizeof(struct ip),
3388 (max_linkhdr + 3) & ~3)) == NULL) {
3389 /* XXXJRT new stat, please */
3390 KMOD_IPSTAT_INC(ips_toosmall);
3393 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3394 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3395 KMOD_IPSTAT_INC(ips_toosmall);
3399 ip = mtod(m, struct ip *);
3400 if (ip == NULL) goto bad;
3402 if (ip->ip_v != IPVERSION) {
3403 KMOD_IPSTAT_INC(ips_badvers);
3406 hlen = ip->ip_hl << 2;
3407 if (hlen < sizeof(struct ip)) { /* minimum header length */
3408 KMOD_IPSTAT_INC(ips_badhlen);
3411 if (hlen > m->m_len) {
3412 if ((m = m_pullup(m, hlen)) == NULL) {
3413 KMOD_IPSTAT_INC(ips_badhlen);
3416 ip = mtod(m, struct ip *);
3417 if (ip == NULL) goto bad;
3420 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3421 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3423 if (hlen == sizeof(struct ip)) {
3424 sum = in_cksum_hdr(ip);
3426 sum = in_cksum(m, hlen);
3430 KMOD_IPSTAT_INC(ips_badsum);
3434 /* Retrieve the packet length. */
3435 len = ntohs(ip->ip_len);
3438 * Check for additional length bogosity
3441 KMOD_IPSTAT_INC(ips_badlen);
3446 * Check that the amount of data in the buffers
3447 * is as at least much as the IP header would have us expect.
3448 * Drop packet if shorter than we expect.
3450 if (m->m_pkthdr.len < len) {
3451 KMOD_IPSTAT_INC(ips_tooshort);
3455 /* Checks out, proceed */
3466 * Same as above, but for IPv6.
3467 * Cut-and-pasted from ip6_input.c.
3468 * XXX Should we update ip6stat, or not?
3471 bridge_ip6_checkbasic(struct mbuf **mp)
3473 struct mbuf *m = *mp;
3474 struct ip6_hdr *ip6;
3477 * If the IPv6 header is not aligned, slurp it up into a new
3478 * mbuf with space for link headers, in the event we forward
3479 * it. Otherwise, if it is aligned, make sure the entire base
3480 * IPv6 header is in the first mbuf of the chain.
3482 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3483 struct ifnet *inifp = m->m_pkthdr.rcvif;
3484 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3485 (max_linkhdr + 3) & ~3)) == NULL) {
3486 /* XXXJRT new stat, please */
3487 IP6STAT_INC(ip6s_toosmall);
3488 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3491 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3492 struct ifnet *inifp = m->m_pkthdr.rcvif;
3493 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3494 IP6STAT_INC(ip6s_toosmall);
3495 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3500 ip6 = mtod(m, struct ip6_hdr *);
3502 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3503 IP6STAT_INC(ip6s_badvers);
3504 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3508 /* Checks out, proceed */
3521 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3524 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3525 int snap, struct llc *llc)
3527 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3531 if (m->m_len < sizeof(struct ip) &&
3532 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3534 ip = mtod(m, struct ip *);
3536 m->m_pkthdr.csum_flags |= CSUM_IP;
3537 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3542 * Walk the chain and re-add the Ethernet header for
3545 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3546 nextpkt = mcur->m_nextpkt;
3547 mcur->m_nextpkt = NULL;
3549 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3553 mprev->m_nextpkt = nextpkt;
3556 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3559 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3563 mprev->m_nextpkt = nextpkt;
3566 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3569 * The previous two M_PREPEND could have inserted one or two
3570 * mbufs in front so we have to update the previous packet's
3573 mcur->m_nextpkt = nextpkt;
3575 mprev->m_nextpkt = mcur;
3577 /* The first mbuf in the original chain needs to be
3584 KMOD_IPSTAT_INC(ips_fragmented);
3588 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3589 m = mcur->m_nextpkt;
3596 bridge_linkstate(struct ifnet *ifp)
3598 struct bridge_softc *sc = ifp->if_bridge;
3599 struct bridge_iflist *bif;
3602 bif = bridge_lookup_member_if(sc, ifp);
3607 bridge_linkcheck(sc);
3610 bstp_linkstate(&bif->bif_stp);
3614 bridge_linkcheck(struct bridge_softc *sc)
3616 struct bridge_iflist *bif;
3617 int new_link, hasls;
3619 BRIDGE_LOCK_ASSERT(sc);
3620 new_link = LINK_STATE_DOWN;
3622 /* Our link is considered up if at least one of our ports is active */
3623 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3624 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3626 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3627 new_link = LINK_STATE_UP;
3631 if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3632 /* If no interfaces support link-state then we default to up */
3633 new_link = LINK_STATE_UP;
3635 if_link_state_change(sc->sc_ifp, new_link);