FreeBSD/Linux Kernel Cross Reference
sys/net/if_bridge.c
1 /* $NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $ */
2
3 /*
4 * Copyright 2001 Wasabi Systems, Inc.
5 * All rights reserved.
6 *
7 * Written by Jason R. Thorpe for Wasabi Systems, Inc.
8 *
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. All advertising materials mentioning features or use of this software
18 * must display the following acknowledgement:
19 * This product includes software developed for the NetBSD Project by
20 * Wasabi Systems, Inc.
21 * 4. The name of Wasabi Systems, Inc. may not be used to endorse
22 * or promote products derived from this software without specific prior
23 * written permission.
24 *
25 * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
26 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
27 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
28 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC
29 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35 * POSSIBILITY OF SUCH DAMAGE.
36 */
37
38 /*
39 * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
40 * All rights reserved.
41 *
42 * Redistribution and use in source and binary forms, with or without
43 * modification, are permitted provided that the following conditions
44 * are met:
45 * 1. Redistributions of source code must retain the above copyright
46 * notice, this list of conditions and the following disclaimer.
47 * 2. Redistributions in binary form must reproduce the above copyright
48 * notice, this list of conditions and the following disclaimer in the
49 * documentation and/or other materials provided with the distribution.
50 *
51 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
52 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
53 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
54 * DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
55 * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
56 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
57 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
58 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
59 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
60 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
61 * POSSIBILITY OF SUCH DAMAGE.
62 *
63 * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
64 */
65
66 /*
67 * Network interface bridge support.
68 *
69 * TODO:
70 *
71 * - Currently only supports Ethernet-like interfaces (Ethernet,
72 * 802.11, VLANs on Ethernet, etc.) Figure out a nice way
73 * to bridge other types of interfaces (FDDI-FDDI, and maybe
74 * consider heterogenous bridges).
75 */
76
77 #include <sys/cdefs.h>
78 __FBSDID("$FreeBSD$");
79
80 #include "opt_inet.h"
81 #include "opt_inet6.h"
82 #include "opt_carp.h"
83
84 #include <sys/param.h>
85 #include <sys/mbuf.h>
86 #include <sys/malloc.h>
87 #include <sys/protosw.h>
88 #include <sys/systm.h>
89 #include <sys/time.h>
90 #include <sys/socket.h> /* for net/if.h */
91 #include <sys/sockio.h>
92 #include <sys/ctype.h> /* string functions */
93 #include <sys/kernel.h>
94 #include <sys/random.h>
95 #include <sys/syslog.h>
96 #include <sys/sysctl.h>
97 #include <vm/uma.h>
98 #include <sys/module.h>
99 #include <sys/proc.h>
100 #include <sys/lock.h>
101 #include <sys/mutex.h>
102
103 #include <net/bpf.h>
104 #include <net/if.h>
105 #include <net/if_clone.h>
106 #include <net/if_dl.h>
107 #include <net/if_types.h>
108 #include <net/if_var.h>
109 #include <net/pfil.h>
110
111 #include <netinet/in.h> /* for struct arpcom */
112 #include <netinet/in_systm.h>
113 #include <netinet/in_var.h>
114 #include <netinet/ip.h>
115 #include <netinet/ip_var.h>
116 #ifdef INET6
117 #include <netinet/ip6.h>
118 #include <netinet6/ip6_var.h>
119 #endif
120 #ifdef DEV_CARP
121 #include <netinet/ip_carp.h>
122 #endif
123 #include <machine/in_cksum.h>
124 #include <netinet/if_ether.h> /* for struct arpcom */
125 #include <net/bridgestp.h>
126 #include <net/if_bridgevar.h>
127 #include <net/if_llc.h>
128
129 #include <net/route.h>
130 #include <netinet/ip_fw.h>
131 #include <netinet/ip_dummynet.h>
132
133 /*
134 * Size of the route hash table. Must be a power of two.
135 */
136 #ifndef BRIDGE_RTHASH_SIZE
137 #define BRIDGE_RTHASH_SIZE 1024
138 #endif
139
140 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1)
141
142 /*
143 * Maximum number of addresses to cache.
144 */
145 #ifndef BRIDGE_RTABLE_MAX
146 #define BRIDGE_RTABLE_MAX 100
147 #endif
148
149 /*
150 * Timeout (in seconds) for entries learned dynamically.
151 */
152 #ifndef BRIDGE_RTABLE_TIMEOUT
153 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */
154 #endif
155
156 /*
157 * Number of seconds between walks of the route list.
158 */
159 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
160 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60)
161 #endif
162
163 /*
164 * List of capabilities to possibly mask on the member interface.
165 */
166 #define BRIDGE_IFCAPS_MASK IFCAP_TXCSUM
167 /*
168 * List of capabilities to disable on the member interface.
169 */
170 #define BRIDGE_IFCAPS_STRIP 0 /* IFCAP_LRO */
171
172 /*
173 * Bridge interface list entry.
174 */
175 struct bridge_iflist {
176 LIST_ENTRY(bridge_iflist) bif_next;
177 struct ifnet *bif_ifp; /* member if */
178 struct bstp_port bif_stp; /* STP state */
179 uint32_t bif_flags; /* member if flags */
180 int bif_savedcaps; /* saved capabilities */
181 };
182
183 /*
184 * Bridge route node.
185 */
186 struct bridge_rtnode {
187 LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
188 LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
189 struct ifnet *brt_ifp; /* destination if */
190 unsigned long brt_expire; /* expiration time */
191 uint8_t brt_flags; /* address flags */
192 uint8_t brt_addr[ETHER_ADDR_LEN];
193 };
194
195 /*
196 * Software state for each bridge.
197 */
198 struct bridge_softc {
199 struct ifnet *sc_ifp; /* make this an interface */
200 LIST_ENTRY(bridge_softc) sc_list;
201 struct mtx sc_mtx;
202 struct cv sc_cv;
203 uint32_t sc_brtmax; /* max # of addresses */
204 uint32_t sc_brtcnt; /* cur. # of addresses */
205 uint32_t sc_brttimeout; /* rt timeout in seconds */
206 struct callout sc_brcallout; /* bridge callout */
207 uint32_t sc_iflist_ref; /* refcount for sc_iflist */
208 uint32_t sc_iflist_xcnt; /* refcount for sc_iflist */
209 LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
210 LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
211 LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
212 uint32_t sc_rthash_key; /* key for hash */
213 LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
214 struct bstp_state sc_stp; /* STP state */
215 uint32_t sc_brtexceeded; /* # of cache drops */
216 };
217
218 static struct mtx bridge_list_mtx;
219 eventhandler_tag bridge_detach_cookie = NULL;
220
221 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
222
223 uma_zone_t bridge_rtnode_zone;
224
225 static int bridge_clone_create(struct if_clone *, int);
226 static void bridge_clone_destroy(struct ifnet *);
227
228 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
229 static void bridge_mutecaps(struct bridge_softc *);
230 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
231 int);
232 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
233 static void bridge_init(void *);
234 static void bridge_dummynet(struct mbuf *, struct ifnet *);
235 static void bridge_stop(struct ifnet *, int);
236 static void bridge_start(struct ifnet *);
237 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
238 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
239 struct rtentry *);
240 static void bridge_enqueue(struct bridge_softc *, struct ifnet *,
241 struct mbuf *);
242 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
243
244 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
245 struct mbuf *m);
246
247 static void bridge_timer(void *);
248
249 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
250 struct mbuf *, int);
251 static void bridge_span(struct bridge_softc *, struct mbuf *);
252
253 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
254 struct bridge_iflist *, int, uint8_t);
255 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *);
256 static void bridge_rttrim(struct bridge_softc *);
257 static void bridge_rtage(struct bridge_softc *);
258 static void bridge_rtflush(struct bridge_softc *, int);
259 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *);
260
261 static int bridge_rtable_init(struct bridge_softc *);
262 static void bridge_rtable_fini(struct bridge_softc *);
263
264 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
265 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
266 const uint8_t *);
267 static int bridge_rtnode_insert(struct bridge_softc *,
268 struct bridge_rtnode *);
269 static void bridge_rtnode_destroy(struct bridge_softc *,
270 struct bridge_rtnode *);
271 static void bridge_rtable_expire(struct ifnet *, int);
272 static void bridge_state_change(struct ifnet *, int);
273
274 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
275 const char *name);
276 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
277 struct ifnet *ifp);
278 static void bridge_delete_member(struct bridge_softc *,
279 struct bridge_iflist *, int);
280 static void bridge_delete_span(struct bridge_softc *,
281 struct bridge_iflist *);
282
283 static int bridge_ioctl_add(struct bridge_softc *, void *);
284 static int bridge_ioctl_del(struct bridge_softc *, void *);
285 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
286 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
287 static int bridge_ioctl_scache(struct bridge_softc *, void *);
288 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
289 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
290 static int bridge_ioctl_rts(struct bridge_softc *, void *);
291 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
292 static int bridge_ioctl_sto(struct bridge_softc *, void *);
293 static int bridge_ioctl_gto(struct bridge_softc *, void *);
294 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
295 static int bridge_ioctl_flush(struct bridge_softc *, void *);
296 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
297 static int bridge_ioctl_spri(struct bridge_softc *, void *);
298 static int bridge_ioctl_ght(struct bridge_softc *, void *);
299 static int bridge_ioctl_sht(struct bridge_softc *, void *);
300 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
301 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
302 static int bridge_ioctl_gma(struct bridge_softc *, void *);
303 static int bridge_ioctl_sma(struct bridge_softc *, void *);
304 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
305 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
306 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
307 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
308 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
309 static int bridge_ioctl_grte(struct bridge_softc *, void *);
310 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
311 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
312 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
313 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
314 int);
315 static int bridge_ip_checkbasic(struct mbuf **mp);
316 #ifdef INET6
317 static int bridge_ip6_checkbasic(struct mbuf **mp);
318 #endif /* INET6 */
319 static int bridge_fragment(struct ifnet *, struct mbuf *,
320 struct ether_header *, int, struct llc *);
321
322 static struct bstp_cb_ops bridge_ops = {
323 .bcb_state = bridge_state_change,
324 .bcb_rtage = bridge_rtable_expire
325 };
326
327 SYSCTL_DECL(_net_link);
328 SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
329
330 static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */
331 static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */
332 static int pfil_member = 1; /* run pfil hooks on the member interface */
333 static int pfil_ipfw = 0; /* layer2 filter with ipfw */
334 static int pfil_ipfw_arp = 0; /* layer2 filter with ipfw */
335 static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for
336 locally destined packets */
337 static int log_stp = 0; /* log STP state changes */
338 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
339 &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
340 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
341 &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
342 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
343 &pfil_bridge, 0, "Packet filter on the bridge interface");
344 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
345 &pfil_member, 0, "Packet filter on the member interface");
346 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
347 &pfil_local_phys, 0,
348 "Packet filter on the physical interface for locally destined packets");
349 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
350 &log_stp, 0, "Log STP state changes");
351
352 struct bridge_control {
353 int (*bc_func)(struct bridge_softc *, void *);
354 int bc_argsize;
355 int bc_flags;
356 };
357
358 #define BC_F_COPYIN 0x01 /* copy arguments in */
359 #define BC_F_COPYOUT 0x02 /* copy arguments out */
360 #define BC_F_SUSER 0x04 /* do super-user check */
361
362 const struct bridge_control bridge_control_table[] = {
363 { bridge_ioctl_add, sizeof(struct ifbreq),
364 BC_F_COPYIN|BC_F_SUSER },
365 { bridge_ioctl_del, sizeof(struct ifbreq),
366 BC_F_COPYIN|BC_F_SUSER },
367
368 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
369 BC_F_COPYIN|BC_F_COPYOUT },
370 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
371 BC_F_COPYIN|BC_F_SUSER },
372
373 { bridge_ioctl_scache, sizeof(struct ifbrparam),
374 BC_F_COPYIN|BC_F_SUSER },
375 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
376 BC_F_COPYOUT },
377
378 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
379 BC_F_COPYIN|BC_F_COPYOUT },
380 { bridge_ioctl_rts, sizeof(struct ifbaconf),
381 BC_F_COPYIN|BC_F_COPYOUT },
382
383 { bridge_ioctl_saddr, sizeof(struct ifbareq),
384 BC_F_COPYIN|BC_F_SUSER },
385
386 { bridge_ioctl_sto, sizeof(struct ifbrparam),
387 BC_F_COPYIN|BC_F_SUSER },
388 { bridge_ioctl_gto, sizeof(struct ifbrparam),
389 BC_F_COPYOUT },
390
391 { bridge_ioctl_daddr, sizeof(struct ifbareq),
392 BC_F_COPYIN|BC_F_SUSER },
393
394 { bridge_ioctl_flush, sizeof(struct ifbreq),
395 BC_F_COPYIN|BC_F_SUSER },
396
397 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
398 BC_F_COPYOUT },
399 { bridge_ioctl_spri, sizeof(struct ifbrparam),
400 BC_F_COPYIN|BC_F_SUSER },
401
402 { bridge_ioctl_ght, sizeof(struct ifbrparam),
403 BC_F_COPYOUT },
404 { bridge_ioctl_sht, sizeof(struct ifbrparam),
405 BC_F_COPYIN|BC_F_SUSER },
406
407 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
408 BC_F_COPYOUT },
409 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
410 BC_F_COPYIN|BC_F_SUSER },
411
412 { bridge_ioctl_gma, sizeof(struct ifbrparam),
413 BC_F_COPYOUT },
414 { bridge_ioctl_sma, sizeof(struct ifbrparam),
415 BC_F_COPYIN|BC_F_SUSER },
416
417 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
418 BC_F_COPYIN|BC_F_SUSER },
419
420 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
421 BC_F_COPYIN|BC_F_SUSER },
422
423 { bridge_ioctl_addspan, sizeof(struct ifbreq),
424 BC_F_COPYIN|BC_F_SUSER },
425 { bridge_ioctl_delspan, sizeof(struct ifbreq),
426 BC_F_COPYIN|BC_F_SUSER },
427
428 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
429 BC_F_COPYOUT },
430
431 { bridge_ioctl_grte, sizeof(struct ifbrparam),
432 BC_F_COPYOUT },
433
434 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
435 BC_F_COPYIN|BC_F_COPYOUT },
436
437 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
438 BC_F_COPYIN|BC_F_SUSER },
439
440 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
441 BC_F_COPYIN|BC_F_SUSER },
442 };
443 const int bridge_control_table_size =
444 sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
445
446 static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] =
447 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
448
449 LIST_HEAD(, bridge_softc) bridge_list;
450
451 IFC_SIMPLE_DECLARE(bridge, 0);
452
453 static int
454 bridge_modevent(module_t mod, int type, void *data)
455 {
456
457 switch (type) {
458 case MOD_LOAD:
459 mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
460 if_clone_attach(&bridge_cloner);
461 bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
462 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
463 UMA_ALIGN_PTR, 0);
464 LIST_INIT(&bridge_list);
465 bridge_input_p = bridge_input;
466 bridge_output_p = bridge_output;
467 bridge_dn_p = bridge_dummynet;
468 bridge_detach_cookie = EVENTHANDLER_REGISTER(
469 ifnet_departure_event, bridge_ifdetach, NULL,
470 EVENTHANDLER_PRI_ANY);
471 break;
472 case MOD_UNLOAD:
473 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
474 bridge_detach_cookie);
475 if_clone_detach(&bridge_cloner);
476 while (!LIST_EMPTY(&bridge_list))
477 bridge_clone_destroy(LIST_FIRST(&bridge_list)->sc_ifp);
478 uma_zdestroy(bridge_rtnode_zone);
479 bridge_input_p = NULL;
480 bridge_output_p = NULL;
481 bridge_dn_p = NULL;
482 mtx_destroy(&bridge_list_mtx);
483 break;
484 default:
485 return (EOPNOTSUPP);
486 }
487 return (0);
488 }
489
490 static moduledata_t bridge_mod = {
491 "if_bridge",
492 bridge_modevent,
493 0
494 };
495
496 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
497 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
498
499 /*
500 * handler for net.link.bridge.pfil_ipfw
501 */
502 static int
503 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
504 {
505 int enable = pfil_ipfw;
506 int error;
507
508 error = sysctl_handle_int(oidp, &enable, 0, req);
509 enable = (enable) ? 1 : 0;
510
511 if (enable != pfil_ipfw) {
512 pfil_ipfw = enable;
513
514 /*
515 * Disable pfil so that ipfw doesnt run twice, if the user
516 * really wants both then they can re-enable pfil_bridge and/or
517 * pfil_member. Also allow non-ip packets as ipfw can filter by
518 * layer2 type.
519 */
520 if (pfil_ipfw) {
521 pfil_onlyip = 0;
522 pfil_bridge = 0;
523 pfil_member = 0;
524 }
525 }
526
527 return (error);
528 }
529 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
530 &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
531
532 /*
533 * bridge_clone_create:
534 *
535 * Create a new bridge instance.
536 */
537 static int
538 bridge_clone_create(struct if_clone *ifc, int unit)
539 {
540 struct bridge_softc *sc, *sc2;
541 struct ifnet *bifp, *ifp;
542 u_char eaddr[6];
543 int retry;
544
545 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
546 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
547 if (ifp == NULL) {
548 free(sc, M_DEVBUF);
549 return (ENOSPC);
550 }
551
552 BRIDGE_LOCK_INIT(sc);
553 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
554 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
555
556 /* Initialize our routing table. */
557 bridge_rtable_init(sc);
558
559 callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
560
561 LIST_INIT(&sc->sc_iflist);
562 LIST_INIT(&sc->sc_spanlist);
563
564 ifp->if_softc = sc;
565 if_initname(ifp, ifc->ifc_name, unit);
566 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
567 ifp->if_ioctl = bridge_ioctl;
568 ifp->if_start = bridge_start;
569 ifp->if_init = bridge_init;
570 ifp->if_type = IFT_BRIDGE;
571 IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
572 ifp->if_snd.ifq_drv_maxlen = ifqmaxlen;
573 IFQ_SET_READY(&ifp->if_snd);
574
575 /*
576 * Generate a random ethernet address with a locally administered
577 * address.
578 *
579 * Since we are using random ethernet addresses for the bridge, it is
580 * possible that we might have address collisions, so make sure that
581 * this hardware address isn't already in use on another bridge.
582 */
583 for (retry = 1; retry != 0;) {
584 arc4rand(eaddr, ETHER_ADDR_LEN, 1);
585 eaddr[0] &= ~1; /* clear multicast bit */
586 eaddr[0] |= 2; /* set the LAA bit */
587 retry = 0;
588 mtx_lock(&bridge_list_mtx);
589 LIST_FOREACH(sc2, &bridge_list, sc_list) {
590 bifp = sc2->sc_ifp;
591 if (memcmp(eaddr, IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0)
592 retry = 1;
593 }
594 mtx_unlock(&bridge_list_mtx);
595 }
596
597 bstp_attach(&sc->sc_stp, &bridge_ops);
598 ether_ifattach(ifp, eaddr);
599 /* Now undo some of the damage... */
600 ifp->if_baudrate = 0;
601 ifp->if_type = IFT_BRIDGE;
602
603 mtx_lock(&bridge_list_mtx);
604 LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
605 mtx_unlock(&bridge_list_mtx);
606
607 return (0);
608 }
609
610 /*
611 * bridge_clone_destroy:
612 *
613 * Destroy a bridge instance.
614 */
615 static void
616 bridge_clone_destroy(struct ifnet *ifp)
617 {
618 struct bridge_softc *sc = ifp->if_softc;
619 struct bridge_iflist *bif;
620
621 BRIDGE_LOCK(sc);
622
623 bridge_stop(ifp, 1);
624 ifp->if_flags &= ~IFF_UP;
625
626 while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
627 bridge_delete_member(sc, bif, 0);
628
629 while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
630 bridge_delete_span(sc, bif);
631 }
632
633 BRIDGE_UNLOCK(sc);
634
635 callout_drain(&sc->sc_brcallout);
636
637 mtx_lock(&bridge_list_mtx);
638 LIST_REMOVE(sc, sc_list);
639 mtx_unlock(&bridge_list_mtx);
640
641 bstp_detach(&sc->sc_stp);
642 ether_ifdetach(ifp);
643 if_free_type(ifp, IFT_ETHER);
644
645 /* Tear down the routing table. */
646 bridge_rtable_fini(sc);
647
648 BRIDGE_LOCK_DESTROY(sc);
649 free(sc, M_DEVBUF);
650 }
651
652 /*
653 * bridge_ioctl:
654 *
655 * Handle a control request from the operator.
656 */
657 static int
658 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
659 {
660 struct bridge_softc *sc = ifp->if_softc;
661 struct thread *td = curthread;
662 union {
663 struct ifbreq ifbreq;
664 struct ifbifconf ifbifconf;
665 struct ifbareq ifbareq;
666 struct ifbaconf ifbaconf;
667 struct ifbrparam ifbrparam;
668 struct ifbropreq ifbropreq;
669 } args;
670 struct ifdrv *ifd = (struct ifdrv *) data;
671 const struct bridge_control *bc;
672 int error = 0;
673
674 switch (cmd) {
675
676 case SIOCADDMULTI:
677 case SIOCDELMULTI:
678 break;
679
680 case SIOCGDRVSPEC:
681 case SIOCSDRVSPEC:
682 if (ifd->ifd_cmd >= bridge_control_table_size) {
683 error = EINVAL;
684 break;
685 }
686 bc = &bridge_control_table[ifd->ifd_cmd];
687
688 if (cmd == SIOCGDRVSPEC &&
689 (bc->bc_flags & BC_F_COPYOUT) == 0) {
690 error = EINVAL;
691 break;
692 }
693 else if (cmd == SIOCSDRVSPEC &&
694 (bc->bc_flags & BC_F_COPYOUT) != 0) {
695 error = EINVAL;
696 break;
697 }
698
699 if (bc->bc_flags & BC_F_SUSER) {
700 error = suser(td);
701 if (error)
702 break;
703 }
704
705 if (ifd->ifd_len != bc->bc_argsize ||
706 ifd->ifd_len > sizeof(args)) {
707 error = EINVAL;
708 break;
709 }
710
711 bzero(&args, sizeof(args));
712 if (bc->bc_flags & BC_F_COPYIN) {
713 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
714 if (error)
715 break;
716 }
717
718 BRIDGE_LOCK(sc);
719 error = (*bc->bc_func)(sc, &args);
720 BRIDGE_UNLOCK(sc);
721 if (error)
722 break;
723
724 if (bc->bc_flags & BC_F_COPYOUT)
725 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
726
727 break;
728
729 case SIOCSIFFLAGS:
730 if (!(ifp->if_flags & IFF_UP) &&
731 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
732 /*
733 * If interface is marked down and it is running,
734 * then stop and disable it.
735 */
736 BRIDGE_LOCK(sc);
737 bridge_stop(ifp, 1);
738 BRIDGE_UNLOCK(sc);
739 } else if ((ifp->if_flags & IFF_UP) &&
740 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
741 /*
742 * If interface is marked up and it is stopped, then
743 * start it.
744 */
745 (*ifp->if_init)(sc);
746 }
747 break;
748
749 case SIOCSIFMTU:
750 /* Do not allow the MTU to be changed on the bridge */
751 error = EINVAL;
752 break;
753
754 default:
755 /*
756 * drop the lock as ether_ioctl() will call bridge_start() and
757 * cause the lock to be recursed.
758 */
759 error = ether_ioctl(ifp, cmd, data);
760 break;
761 }
762
763 return (error);
764 }
765
766 /*
767 * bridge_mutecaps:
768 *
769 * Clear or restore unwanted capabilities on the member interface
770 */
771 static void
772 bridge_mutecaps(struct bridge_softc *sc)
773 {
774 struct bridge_iflist *bif;
775 int enabled, mask;
776
777 /* Initial bitmask of capabilities to test */
778 mask = BRIDGE_IFCAPS_MASK;
779
780 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
781 /* Every member must support it or its disabled */
782 mask &= bif->bif_savedcaps;
783 }
784
785 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
786 enabled = bif->bif_ifp->if_capenable;
787 enabled &= ~BRIDGE_IFCAPS_STRIP;
788 /* strip off mask bits and enable them again if allowed */
789 enabled &= ~BRIDGE_IFCAPS_MASK;
790 enabled |= mask;
791
792 bridge_set_ifcap(sc, bif, enabled);
793 }
794
795 }
796
797 static void
798 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
799 {
800 struct ifnet *ifp = bif->bif_ifp;
801 struct ifreq ifr;
802 int error;
803
804 bzero(&ifr, sizeof(ifr));
805 ifr.ifr_reqcap = set;
806
807 if (ifp->if_capenable != set) {
808 IFF_LOCKGIANT(ifp);
809 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
810 IFF_UNLOCKGIANT(ifp);
811 if (error)
812 if_printf(sc->sc_ifp,
813 "error setting interface capabilities on %s\n",
814 ifp->if_xname);
815 }
816 }
817
818 /*
819 * bridge_lookup_member:
820 *
821 * Lookup a bridge member interface.
822 */
823 static struct bridge_iflist *
824 bridge_lookup_member(struct bridge_softc *sc, const char *name)
825 {
826 struct bridge_iflist *bif;
827 struct ifnet *ifp;
828
829 BRIDGE_LOCK_ASSERT(sc);
830
831 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
832 ifp = bif->bif_ifp;
833 if (strcmp(ifp->if_xname, name) == 0)
834 return (bif);
835 }
836
837 return (NULL);
838 }
839
840 /*
841 * bridge_lookup_member_if:
842 *
843 * Lookup a bridge member interface by ifnet*.
844 */
845 static struct bridge_iflist *
846 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
847 {
848 struct bridge_iflist *bif;
849
850 BRIDGE_LOCK_ASSERT(sc);
851
852 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
853 if (bif->bif_ifp == member_ifp)
854 return (bif);
855 }
856
857 return (NULL);
858 }
859
860 /*
861 * bridge_delete_member:
862 *
863 * Delete the specified member interface.
864 */
865 static void
866 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
867 int gone)
868 {
869 struct ifnet *ifs = bif->bif_ifp;
870
871 BRIDGE_LOCK_ASSERT(sc);
872
873 if (!gone) {
874 switch (ifs->if_type) {
875 case IFT_ETHER:
876 case IFT_L2VLAN:
877 /*
878 * Take the interface out of promiscuous mode.
879 */
880 (void) ifpromisc(ifs, 0);
881 break;
882
883 case IFT_GIF:
884 break;
885
886 default:
887 #ifdef DIAGNOSTIC
888 panic("bridge_delete_member: impossible");
889 #endif
890 break;
891 }
892 /* reneable any interface capabilities */
893 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
894 }
895
896 if (bif->bif_flags & IFBIF_STP)
897 bstp_disable(&bif->bif_stp);
898
899 ifs->if_bridge = NULL;
900 BRIDGE_XLOCK(sc);
901 LIST_REMOVE(bif, bif_next);
902 BRIDGE_XDROP(sc);
903
904 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
905 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
906
907 BRIDGE_UNLOCK(sc);
908 bstp_destroy(&bif->bif_stp); /* prepare to free */
909 BRIDGE_LOCK(sc);
910 free(bif, M_DEVBUF);
911 }
912
913 /*
914 * bridge_delete_span:
915 *
916 * Delete the specified span interface.
917 */
918 static void
919 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
920 {
921 BRIDGE_LOCK_ASSERT(sc);
922
923 KASSERT(bif->bif_ifp->if_bridge == NULL,
924 ("%s: not a span interface", __func__));
925
926 LIST_REMOVE(bif, bif_next);
927 free(bif, M_DEVBUF);
928 }
929
930 static int
931 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
932 {
933 struct ifbreq *req = arg;
934 struct bridge_iflist *bif = NULL;
935 struct ifnet *ifs;
936 int error = 0;
937
938 ifs = ifunit(req->ifbr_ifsname);
939 if (ifs == NULL)
940 return (ENOENT);
941 if (ifs->if_ioctl == NULL) /* must be supported */
942 return (EINVAL);
943
944 /* If it's in the span list, it can't be a member. */
945 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
946 if (ifs == bif->bif_ifp)
947 return (EBUSY);
948
949 /* Allow the first Ethernet member to define the MTU */
950 if (ifs->if_type != IFT_GIF) {
951 if (LIST_EMPTY(&sc->sc_iflist))
952 sc->sc_ifp->if_mtu = ifs->if_mtu;
953 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
954 if_printf(sc->sc_ifp, "invalid MTU for %s\n",
955 ifs->if_xname);
956 return (EINVAL);
957 }
958 }
959
960 if (ifs->if_bridge == sc)
961 return (EEXIST);
962
963 if (ifs->if_bridge != NULL)
964 return (EBUSY);
965
966 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
967 if (bif == NULL)
968 return (ENOMEM);
969
970 bif->bif_ifp = ifs;
971 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
972 bif->bif_savedcaps = ifs->if_capenable;
973
974 switch (ifs->if_type) {
975 case IFT_ETHER:
976 case IFT_L2VLAN:
977 /*
978 * Place the interface into promiscuous mode.
979 */
980 error = ifpromisc(ifs, 1);
981 if (error)
982 goto out;
983 break;
984
985 case IFT_GIF:
986 break;
987
988 default:
989 error = EINVAL;
990 goto out;
991 }
992
993 ifs->if_bridge = sc;
994 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
995 /*
996 * XXX: XLOCK HERE!?!
997 *
998 * NOTE: insert_***HEAD*** should be safe for the traversals.
999 */
1000 LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1001
1002 /* Set interface capabilities to the intersection set of all members */
1003 bridge_mutecaps(sc);
1004 out:
1005 if (error) {
1006 if (bif != NULL)
1007 free(bif, M_DEVBUF);
1008 }
1009 return (error);
1010 }
1011
1012 static int
1013 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1014 {
1015 struct ifbreq *req = arg;
1016 struct bridge_iflist *bif;
1017
1018 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1019 if (bif == NULL)
1020 return (ENOENT);
1021
1022 bridge_delete_member(sc, bif, 0);
1023
1024 return (0);
1025 }
1026
1027 static int
1028 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1029 {
1030 struct ifbreq *req = arg;
1031 struct bridge_iflist *bif;
1032 struct bstp_port *bp;
1033
1034 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1035 if (bif == NULL)
1036 return (ENOENT);
1037
1038 bp = &bif->bif_stp;
1039 req->ifbr_ifsflags = bif->bif_flags;
1040 req->ifbr_state = bp->bp_state;
1041 req->ifbr_priority = bp->bp_priority;
1042 req->ifbr_path_cost = bp->bp_path_cost;
1043 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1044 req->ifbr_proto = bp->bp_protover;
1045 req->ifbr_role = bp->bp_role;
1046 req->ifbr_stpflags = bp->bp_flags;
1047
1048 /* Copy STP state options as flags */
1049 if (bp->bp_operedge)
1050 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1051 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1052 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1053 if (bp->bp_ptp_link)
1054 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1055 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1056 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1057 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1058 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1059 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1060 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1061 return (0);
1062 }
1063
1064 static int
1065 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1066 {
1067 struct ifbreq *req = arg;
1068 struct bridge_iflist *bif;
1069 struct bstp_port *bp;
1070 int error;
1071
1072 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1073 if (bif == NULL)
1074 return (ENOENT);
1075 bp = &bif->bif_stp;
1076
1077 if (req->ifbr_ifsflags & IFBIF_SPAN)
1078 /* SPAN is readonly */
1079 return (EINVAL);
1080
1081 if (req->ifbr_ifsflags & IFBIF_STP) {
1082 if ((bif->bif_flags & IFBIF_STP) == 0) {
1083 error = bstp_enable(&bif->bif_stp);
1084 if (error)
1085 return (error);
1086 }
1087 } else {
1088 if ((bif->bif_flags & IFBIF_STP) != 0)
1089 bstp_disable(&bif->bif_stp);
1090 }
1091
1092 /* Pass on STP flags */
1093 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1094 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1095 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1096 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1097
1098 /* Save the bits relating to the bridge */
1099 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1100
1101 return (0);
1102 }
1103
1104 static int
1105 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1106 {
1107 struct ifbrparam *param = arg;
1108
1109 sc->sc_brtmax = param->ifbrp_csize;
1110 bridge_rttrim(sc);
1111
1112 return (0);
1113 }
1114
1115 static int
1116 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1117 {
1118 struct ifbrparam *param = arg;
1119
1120 param->ifbrp_csize = sc->sc_brtmax;
1121
1122 return (0);
1123 }
1124
1125 static int
1126 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1127 {
1128 struct ifbifconf *bifc = arg;
1129 struct bridge_iflist *bif;
1130 struct ifbreq breq;
1131 char *buf, *outbuf;
1132 int count, buflen, len, error = 0;
1133
1134 count = 0;
1135 LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1136 count++;
1137 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1138 count++;
1139
1140 buflen = sizeof(breq) * count;
1141 if (bifc->ifbic_len == 0) {
1142 bifc->ifbic_len = buflen;
1143 return (0);
1144 }
1145 BRIDGE_UNLOCK(sc);
1146 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1147 BRIDGE_LOCK(sc);
1148
1149 count = 0;
1150 buf = outbuf;
1151 len = min(bifc->ifbic_len, buflen);
1152 bzero(&breq, sizeof(breq));
1153 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1154 if (len < sizeof(breq))
1155 break;
1156
1157 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1158 sizeof(breq.ifbr_ifsname));
1159 /* Fill in the ifbreq structure */
1160 error = bridge_ioctl_gifflags(sc, &breq);
1161 if (error)
1162 break;
1163 memcpy(buf, &breq, sizeof(breq));
1164 count++;
1165 buf += sizeof(breq);
1166 len -= sizeof(breq);
1167 }
1168 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1169 if (len < sizeof(breq))
1170 break;
1171
1172 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1173 sizeof(breq.ifbr_ifsname));
1174 breq.ifbr_ifsflags = bif->bif_flags;
1175 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1176 memcpy(buf, &breq, sizeof(breq));
1177 count++;
1178 buf += sizeof(breq);
1179 len -= sizeof(breq);
1180 }
1181
1182 BRIDGE_UNLOCK(sc);
1183 bifc->ifbic_len = sizeof(breq) * count;
1184 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1185 BRIDGE_LOCK(sc);
1186 free(outbuf, M_TEMP);
1187 return (error);
1188 }
1189
1190 static int
1191 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1192 {
1193 struct ifbaconf *bac = arg;
1194 struct bridge_rtnode *brt;
1195 struct ifbareq bareq;
1196 char *buf, *outbuf;
1197 int count, buflen, len, error = 0;
1198
1199 if (bac->ifbac_len == 0)
1200 return (0);
1201
1202 count = 0;
1203 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1204 count++;
1205 buflen = sizeof(bareq) * count;
1206
1207 BRIDGE_UNLOCK(sc);
1208 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1209 BRIDGE_LOCK(sc);
1210
1211 count = 0;
1212 buf = outbuf;
1213 len = min(bac->ifbac_len, buflen);
1214 bzero(&bareq, sizeof(bareq));
1215 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1216 if (len < sizeof(bareq))
1217 goto out;
1218 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1219 sizeof(bareq.ifba_ifsname));
1220 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1221 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1222 time_uptime < brt->brt_expire)
1223 bareq.ifba_expire = brt->brt_expire - time_uptime;
1224 else
1225 bareq.ifba_expire = 0;
1226 bareq.ifba_flags = brt->brt_flags;
1227
1228 memcpy(buf, &bareq, sizeof(bareq));
1229 count++;
1230 buf += sizeof(bareq);
1231 len -= sizeof(bareq);
1232 }
1233 out:
1234 BRIDGE_UNLOCK(sc);
1235 bac->ifbac_len = sizeof(bareq) * count;
1236 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1237 BRIDGE_LOCK(sc);
1238 free(outbuf, M_TEMP);
1239 return (error);
1240 }
1241
1242 static int
1243 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1244 {
1245 struct ifbareq *req = arg;
1246 struct bridge_iflist *bif;
1247 int error;
1248
1249 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1250 if (bif == NULL)
1251 return (ENOENT);
1252
1253 error = bridge_rtupdate(sc, req->ifba_dst, bif, 1,
1254 req->ifba_flags);
1255
1256 return (error);
1257 }
1258
1259 static int
1260 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1261 {
1262 struct ifbrparam *param = arg;
1263
1264 sc->sc_brttimeout = param->ifbrp_ctime;
1265 return (0);
1266 }
1267
1268 static int
1269 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1270 {
1271 struct ifbrparam *param = arg;
1272
1273 param->ifbrp_ctime = sc->sc_brttimeout;
1274 return (0);
1275 }
1276
1277 static int
1278 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1279 {
1280 struct ifbareq *req = arg;
1281
1282 return (bridge_rtdaddr(sc, req->ifba_dst));
1283 }
1284
1285 static int
1286 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1287 {
1288 struct ifbreq *req = arg;
1289
1290 bridge_rtflush(sc, req->ifbr_ifsflags);
1291 return (0);
1292 }
1293
1294 static int
1295 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1296 {
1297 struct ifbrparam *param = arg;
1298 struct bstp_state *bs = &sc->sc_stp;
1299
1300 param->ifbrp_prio = bs->bs_bridge_priority;
1301 return (0);
1302 }
1303
1304 static int
1305 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1306 {
1307 struct ifbrparam *param = arg;
1308
1309 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1310 }
1311
1312 static int
1313 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1314 {
1315 struct ifbrparam *param = arg;
1316 struct bstp_state *bs = &sc->sc_stp;
1317
1318 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1319 return (0);
1320 }
1321
1322 static int
1323 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1324 {
1325 struct ifbrparam *param = arg;
1326
1327 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1328 }
1329
1330 static int
1331 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1332 {
1333 struct ifbrparam *param = arg;
1334 struct bstp_state *bs = &sc->sc_stp;
1335
1336 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1337 return (0);
1338 }
1339
1340 static int
1341 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1342 {
1343 struct ifbrparam *param = arg;
1344
1345 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1346 }
1347
1348 static int
1349 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1350 {
1351 struct ifbrparam *param = arg;
1352 struct bstp_state *bs = &sc->sc_stp;
1353
1354 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1355 return (0);
1356 }
1357
1358 static int
1359 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1360 {
1361 struct ifbrparam *param = arg;
1362
1363 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1364 }
1365
1366 static int
1367 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1368 {
1369 struct ifbreq *req = arg;
1370 struct bridge_iflist *bif;
1371
1372 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1373 if (bif == NULL)
1374 return (ENOENT);
1375
1376 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1377 }
1378
1379 static int
1380 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1381 {
1382 struct ifbreq *req = arg;
1383 struct bridge_iflist *bif;
1384
1385 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1386 if (bif == NULL)
1387 return (ENOENT);
1388
1389 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1390 }
1391
1392 static int
1393 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1394 {
1395 struct ifbreq *req = arg;
1396 struct bridge_iflist *bif = NULL;
1397 struct ifnet *ifs;
1398
1399 ifs = ifunit(req->ifbr_ifsname);
1400 if (ifs == NULL)
1401 return (ENOENT);
1402
1403 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1404 if (ifs == bif->bif_ifp)
1405 return (EBUSY);
1406
1407 if (ifs->if_bridge != NULL)
1408 return (EBUSY);
1409
1410 switch (ifs->if_type) {
1411 case IFT_ETHER:
1412 case IFT_GIF:
1413 case IFT_L2VLAN:
1414 break;
1415 default:
1416 return (EINVAL);
1417 }
1418
1419 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1420 if (bif == NULL)
1421 return (ENOMEM);
1422
1423 bif->bif_ifp = ifs;
1424 bif->bif_flags = IFBIF_SPAN;
1425
1426 LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1427
1428 return (0);
1429 }
1430
1431 static int
1432 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1433 {
1434 struct ifbreq *req = arg;
1435 struct bridge_iflist *bif;
1436 struct ifnet *ifs;
1437
1438 ifs = ifunit(req->ifbr_ifsname);
1439 if (ifs == NULL)
1440 return (ENOENT);
1441
1442 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1443 if (ifs == bif->bif_ifp)
1444 break;
1445
1446 if (bif == NULL)
1447 return (ENOENT);
1448
1449 bridge_delete_span(sc, bif);
1450
1451 return (0);
1452 }
1453
1454 static int
1455 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1456 {
1457 struct ifbropreq *req = arg;
1458 struct bstp_state *bs = &sc->sc_stp;
1459 struct bstp_port *root_port;
1460
1461 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1462 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1463 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1464
1465 root_port = bs->bs_root_port;
1466 if (root_port == NULL)
1467 req->ifbop_root_port = 0;
1468 else
1469 req->ifbop_root_port = root_port->bp_ifp->if_index;
1470
1471 req->ifbop_holdcount = bs->bs_txholdcount;
1472 req->ifbop_priority = bs->bs_bridge_priority;
1473 req->ifbop_protocol = bs->bs_protover;
1474 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1475 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1476 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1477 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1478 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1479 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1480
1481 return (0);
1482 }
1483
1484 static int
1485 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1486 {
1487 struct ifbrparam *param = arg;
1488
1489 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1490 return (0);
1491 }
1492
1493 static int
1494 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1495 {
1496 struct ifbpstpconf *bifstp = arg;
1497 struct bridge_iflist *bif;
1498 struct bstp_port *bp;
1499 struct ifbpstpreq bpreq;
1500 char *buf, *outbuf;
1501 int count, buflen, len, error = 0;
1502
1503 count = 0;
1504 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1505 if ((bif->bif_flags & IFBIF_STP) != 0)
1506 count++;
1507 }
1508
1509 buflen = sizeof(bpreq) * count;
1510 if (bifstp->ifbpstp_len == 0) {
1511 bifstp->ifbpstp_len = buflen;
1512 return (0);
1513 }
1514
1515 BRIDGE_UNLOCK(sc);
1516 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1517 BRIDGE_LOCK(sc);
1518
1519 count = 0;
1520 buf = outbuf;
1521 len = min(bifstp->ifbpstp_len, buflen);
1522 bzero(&bpreq, sizeof(bpreq));
1523 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1524 if (len < sizeof(bpreq))
1525 break;
1526
1527 if ((bif->bif_flags & IFBIF_STP) == 0)
1528 continue;
1529
1530 bp = &bif->bif_stp;
1531 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1532 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1533 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1534 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1535 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1536 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1537
1538 memcpy(buf, &bpreq, sizeof(bpreq));
1539 count++;
1540 buf += sizeof(bpreq);
1541 len -= sizeof(bpreq);
1542 }
1543
1544 BRIDGE_UNLOCK(sc);
1545 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1546 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1547 BRIDGE_LOCK(sc);
1548 free(outbuf, M_TEMP);
1549 return (error);
1550 }
1551
1552 static int
1553 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1554 {
1555 struct ifbrparam *param = arg;
1556
1557 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1558 }
1559
1560 static int
1561 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1562 {
1563 struct ifbrparam *param = arg;
1564
1565 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1566 }
1567
1568 /*
1569 * bridge_ifdetach:
1570 *
1571 * Detach an interface from a bridge. Called when a member
1572 * interface is detaching.
1573 */
1574 static void
1575 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1576 {
1577 struct bridge_softc *sc = ifp->if_bridge;
1578 struct bridge_iflist *bif;
1579
1580 /* Check if the interface is a bridge member */
1581 if (sc != NULL) {
1582 BRIDGE_LOCK(sc);
1583
1584 bif = bridge_lookup_member_if(sc, ifp);
1585 if (bif != NULL)
1586 bridge_delete_member(sc, bif, 1);
1587
1588 BRIDGE_UNLOCK(sc);
1589 return;
1590 }
1591
1592 /* Check if the interface is a span port */
1593 mtx_lock(&bridge_list_mtx);
1594 LIST_FOREACH(sc, &bridge_list, sc_list) {
1595 BRIDGE_LOCK(sc);
1596 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1597 if (ifp == bif->bif_ifp) {
1598 bridge_delete_span(sc, bif);
1599 break;
1600 }
1601
1602 BRIDGE_UNLOCK(sc);
1603 }
1604 mtx_unlock(&bridge_list_mtx);
1605 }
1606
1607 /*
1608 * bridge_init:
1609 *
1610 * Initialize a bridge interface.
1611 */
1612 static void
1613 bridge_init(void *xsc)
1614 {
1615 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1616 struct ifnet *ifp = sc->sc_ifp;
1617
1618 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1619 return;
1620
1621 BRIDGE_LOCK(sc);
1622 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1623 bridge_timer, sc);
1624
1625 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1626 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1627
1628 BRIDGE_UNLOCK(sc);
1629 }
1630
1631 /*
1632 * bridge_stop:
1633 *
1634 * Stop the bridge interface.
1635 */
1636 static void
1637 bridge_stop(struct ifnet *ifp, int disable)
1638 {
1639 struct bridge_softc *sc = ifp->if_softc;
1640
1641 BRIDGE_LOCK_ASSERT(sc);
1642
1643 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1644 return;
1645
1646 callout_stop(&sc->sc_brcallout);
1647 bstp_stop(&sc->sc_stp);
1648
1649 bridge_rtflush(sc, IFBF_FLUSHDYN);
1650
1651 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1652 }
1653
1654 /*
1655 * bridge_enqueue:
1656 *
1657 * Enqueue a packet on a bridge member interface.
1658 *
1659 */
1660 static void
1661 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1662 {
1663 int len, err = 0;
1664 short mflags;
1665 struct mbuf *m0;
1666
1667 len = m->m_pkthdr.len;
1668 mflags = m->m_flags;
1669
1670 /* We may be sending a fragment so traverse the mbuf */
1671 for (; m; m = m0) {
1672 m0 = m->m_nextpkt;
1673 m->m_nextpkt = NULL;
1674
1675 if (err == 0)
1676 IFQ_ENQUEUE(&dst_ifp->if_snd, m, err);
1677 }
1678
1679 if (err == 0) {
1680
1681 sc->sc_ifp->if_opackets++;
1682 sc->sc_ifp->if_obytes += len;
1683
1684 dst_ifp->if_obytes += len;
1685
1686 if (mflags & M_MCAST) {
1687 sc->sc_ifp->if_omcasts++;
1688 dst_ifp->if_omcasts++;
1689 }
1690 }
1691
1692 if ((dst_ifp->if_drv_flags & IFF_DRV_OACTIVE) == 0)
1693 (*dst_ifp->if_start)(dst_ifp);
1694 }
1695
1696 /*
1697 * bridge_dummynet:
1698 *
1699 * Receive a queued packet from dummynet and pass it on to the output
1700 * interface.
1701 *
1702 * The mbuf has the Ethernet header already attached.
1703 */
1704 static void
1705 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1706 {
1707 struct bridge_softc *sc;
1708
1709 sc = ifp->if_bridge;
1710
1711 /*
1712 * The packet didnt originate from a member interface. This should only
1713 * ever happen if a member interface is removed while packets are
1714 * queued for it.
1715 */
1716 if (sc == NULL) {
1717 m_freem(m);
1718 return;
1719 }
1720
1721 if (inet_pfil_hook.ph_busy_count >= 0
1722 #ifdef INET6
1723 || inet6_pfil_hook.ph_busy_count >= 0
1724 #endif
1725 ) {
1726 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1727 return;
1728 if (m == NULL)
1729 return;
1730 }
1731
1732 bridge_enqueue(sc, ifp, m);
1733 }
1734
1735 /*
1736 * bridge_output:
1737 *
1738 * Send output from a bridge member interface. This
1739 * performs the bridging function for locally originated
1740 * packets.
1741 *
1742 * The mbuf has the Ethernet header already attached. We must
1743 * enqueue or free the mbuf before returning.
1744 */
1745 static int
1746 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1747 struct rtentry *rt)
1748 {
1749 struct ether_header *eh;
1750 struct ifnet *dst_if;
1751 struct bridge_softc *sc;
1752
1753 if (m->m_len < ETHER_HDR_LEN) {
1754 m = m_pullup(m, ETHER_HDR_LEN);
1755 if (m == NULL)
1756 return (0);
1757 }
1758
1759 eh = mtod(m, struct ether_header *);
1760 sc = ifp->if_bridge;
1761
1762 BRIDGE_LOCK(sc);
1763
1764 /*
1765 * If bridge is down, but the original output interface is up,
1766 * go ahead and send out that interface. Otherwise, the packet
1767 * is dropped below.
1768 */
1769 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1770 dst_if = ifp;
1771 goto sendunicast;
1772 }
1773
1774 /*
1775 * If the packet is a multicast, or we don't know a better way to
1776 * get there, send to all interfaces.
1777 */
1778 if (ETHER_IS_MULTICAST(eh->ether_dhost))
1779 dst_if = NULL;
1780 else
1781 dst_if = bridge_rtlookup(sc, eh->ether_dhost);
1782 if (dst_if == NULL) {
1783 struct bridge_iflist *bif;
1784 struct mbuf *mc;
1785 int error = 0, used = 0;
1786
1787 bridge_span(sc, m);
1788
1789 BRIDGE_LOCK2REF(sc, error);
1790 if (error) {
1791 m_freem(m);
1792 return (0);
1793 }
1794
1795 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1796 dst_if = bif->bif_ifp;
1797
1798 if (dst_if->if_type == IFT_GIF)
1799 continue;
1800 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1801 continue;
1802
1803 /*
1804 * If this is not the original output interface,
1805 * and the interface is participating in spanning
1806 * tree, make sure the port is in a state that
1807 * allows forwarding.
1808 */
1809 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
1810 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1811 continue;
1812
1813 if (LIST_NEXT(bif, bif_next) == NULL) {
1814 used = 1;
1815 mc = m;
1816 } else {
1817 mc = m_copypacket(m, M_DONTWAIT);
1818 if (mc == NULL) {
1819 sc->sc_ifp->if_oerrors++;
1820 continue;
1821 }
1822 }
1823
1824 bridge_enqueue(sc, dst_if, mc);
1825 }
1826 if (used == 0)
1827 m_freem(m);
1828 BRIDGE_UNREF(sc);
1829 return (0);
1830 }
1831
1832 sendunicast:
1833 /*
1834 * XXX Spanning tree consideration here?
1835 */
1836
1837 bridge_span(sc, m);
1838 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1839 m_freem(m);
1840 BRIDGE_UNLOCK(sc);
1841 return (0);
1842 }
1843
1844 BRIDGE_UNLOCK(sc);
1845 bridge_enqueue(sc, dst_if, m);
1846 return (0);
1847 }
1848
1849 /*
1850 * bridge_start:
1851 *
1852 * Start output on a bridge.
1853 *
1854 */
1855 static void
1856 bridge_start(struct ifnet *ifp)
1857 {
1858 struct bridge_softc *sc;
1859 struct mbuf *m;
1860 struct ether_header *eh;
1861 struct ifnet *dst_if;
1862
1863 sc = ifp->if_softc;
1864
1865 ifp->if_drv_flags |= IFF_DRV_OACTIVE;
1866 for (;;) {
1867 IFQ_DEQUEUE(&ifp->if_snd, m);
1868 if (m == 0)
1869 break;
1870 ETHER_BPF_MTAP(ifp, m);
1871
1872 eh = mtod(m, struct ether_header *);
1873 dst_if = NULL;
1874
1875 BRIDGE_LOCK(sc);
1876 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
1877 dst_if = bridge_rtlookup(sc, eh->ether_dhost);
1878 }
1879
1880 if (dst_if == NULL)
1881 bridge_broadcast(sc, ifp, m, 0);
1882 else {
1883 BRIDGE_UNLOCK(sc);
1884 bridge_enqueue(sc, dst_if, m);
1885 }
1886 }
1887 ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
1888 }
1889
1890 /*
1891 * bridge_forward:
1892 *
1893 * The forwarding function of the bridge.
1894 *
1895 * NOTE: Releases the lock on return.
1896 */
1897 static void
1898 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
1899 struct mbuf *m)
1900 {
1901 struct bridge_iflist *dbif;
1902 struct ifnet *src_if, *dst_if, *ifp;
1903 struct ether_header *eh;
1904 uint8_t *dst;
1905
1906 src_if = m->m_pkthdr.rcvif;
1907 ifp = sc->sc_ifp;
1908
1909 ifp->if_ipackets++;
1910 ifp->if_ibytes += m->m_pkthdr.len;
1911
1912 if ((sbif->bif_flags & IFBIF_STP) &&
1913 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1914 goto drop;
1915
1916 eh = mtod(m, struct ether_header *);
1917 dst = eh->ether_dhost;
1918
1919 /*
1920 * If the interface is learning, and the source
1921 * address is valid and not multicast, record
1922 * the address.
1923 */
1924 if ((sbif->bif_flags & IFBIF_LEARNING) != 0 &&
1925 ETHER_IS_MULTICAST(eh->ether_shost) == 0 &&
1926 (eh->ether_shost[0] == 0 &&
1927 eh->ether_shost[1] == 0 &&
1928 eh->ether_shost[2] == 0 &&
1929 eh->ether_shost[3] == 0 &&
1930 eh->ether_shost[4] == 0 &&
1931 eh->ether_shost[5] == 0) == 0) {
1932 (void) bridge_rtupdate(sc, eh->ether_shost,
1933 sbif, 0, IFBAF_DYNAMIC);
1934 }
1935
1936 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
1937 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
1938 goto drop;
1939
1940 /*
1941 * At this point, the port either doesn't participate
1942 * in spanning tree or it is in the forwarding state.
1943 */
1944
1945 /*
1946 * If the packet is unicast, destined for someone on
1947 * "this" side of the bridge, drop it.
1948 */
1949 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
1950 dst_if = bridge_rtlookup(sc, dst);
1951 if (src_if == dst_if)
1952 goto drop;
1953 } else {
1954 /*
1955 * Check if its a reserved multicast address, any address
1956 * listed in 802.1D section 7.12.6 may not be forwarded by the
1957 * bridge.
1958 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
1959 */
1960 if (dst[0] == 0x01 && dst[1] == 0x80 &&
1961 dst[2] == 0xc2 && dst[3] == 0x00 &&
1962 dst[4] == 0x00 && dst[5] <= 0x0f)
1963 goto drop;
1964
1965 /* ...forward it to all interfaces. */
1966 ifp->if_imcasts++;
1967 dst_if = NULL;
1968 }
1969
1970 /*
1971 * If we have a destination interface which is a member of our bridge,
1972 * OR this is a unicast packet, push it through the bpf(4) machinery.
1973 * For broadcast or multicast packets, don't bother because it will
1974 * be reinjected into ether_input. We do this before we pass the packets
1975 * through the pfil(9) framework, as it is possible that pfil(9) will
1976 * drop the packet, or possibly modify it, making it difficult to debug
1977 * firewall issues on the bridge.
1978 */
1979 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
1980 ETHER_BPF_MTAP(ifp, m);
1981
1982 /* run the packet filter */
1983 if (inet_pfil_hook.ph_busy_count >= 0
1984 #ifdef INET6
1985 || inet6_pfil_hook.ph_busy_count >= 0
1986 #endif
1987 ) {
1988 BRIDGE_UNLOCK(sc);
1989 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
1990 return;
1991 if (m == NULL)
1992 return;
1993 BRIDGE_LOCK(sc);
1994 }
1995
1996 if (dst_if == NULL) {
1997 bridge_broadcast(sc, src_if, m, 1);
1998 return;
1999 }
2000
2001 /*
2002 * At this point, we're dealing with a unicast frame
2003 * going to a different interface.
2004 */
2005 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2006 goto drop;
2007
2008 dbif = bridge_lookup_member_if(sc, dst_if);
2009 if (dbif == NULL)
2010 /* Not a member of the bridge (anymore?) */
2011 goto drop;
2012
2013 /* Private segments can not talk to each other */
2014 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2015 goto drop;
2016
2017 if ((dbif->bif_flags & IFBIF_STP) &&
2018 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2019 goto drop;
2020
2021 BRIDGE_UNLOCK(sc);
2022
2023 if (inet_pfil_hook.ph_busy_count >= 0
2024 #ifdef INET6
2025 || inet6_pfil_hook.ph_busy_count >= 0
2026 #endif
2027 ) {
2028 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2029 return;
2030 if (m == NULL)
2031 return;
2032 }
2033
2034 bridge_enqueue(sc, dst_if, m);
2035 return;
2036
2037 drop:
2038 BRIDGE_UNLOCK(sc);
2039 m_freem(m);
2040 }
2041
2042 /*
2043 * bridge_input:
2044 *
2045 * Receive input from a member interface. Queue the packet for
2046 * bridging if it is not for us.
2047 */
2048 static struct mbuf *
2049 bridge_input(struct ifnet *ifp, struct mbuf *m)
2050 {
2051 struct bridge_softc *sc = ifp->if_bridge;
2052 struct bridge_iflist *bif, *bif2;
2053 struct ifnet *bifp;
2054 struct ether_header *eh;
2055 struct mbuf *mc, *mc2;
2056
2057 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2058 return (m);
2059
2060 bifp = sc->sc_ifp;
2061
2062 /*
2063 * Implement support for bridge monitoring. If this flag has been
2064 * set on this interface, discard the packet once we push it through
2065 * the bpf(4) machinery, but before we do, increment the byte and
2066 * packet counters associated with this interface.
2067 */
2068 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2069 m->m_pkthdr.rcvif = bifp;
2070 ETHER_BPF_MTAP(bifp, m);
2071 bifp->if_ipackets++;
2072 bifp->if_ibytes += m->m_pkthdr.len;
2073 m_freem(m);
2074 return (NULL);
2075 }
2076 BRIDGE_LOCK(sc);
2077 bif = bridge_lookup_member_if(sc, ifp);
2078 if (bif == NULL) {
2079 BRIDGE_UNLOCK(sc);
2080 return (m);
2081 }
2082
2083 eh = mtod(m, struct ether_header *);
2084
2085 bridge_span(sc, m);
2086
2087 if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
2088 /* Tap off 802.1D packets; they do not get forwarded. */
2089 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2090 ETHER_ADDR_LEN) == 0) {
2091 m = bstp_input(&bif->bif_stp, ifp, m);
2092 if (m == NULL) {
2093 BRIDGE_UNLOCK(sc);
2094 return (NULL);
2095 }
2096 }
2097
2098 if ((bif->bif_flags & IFBIF_STP) &&
2099 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2100 BRIDGE_UNLOCK(sc);
2101 return (m);
2102 }
2103
2104 if (bcmp(etherbroadcastaddr, eh->ether_dhost,
2105 sizeof(etherbroadcastaddr)) == 0)
2106 m->m_flags |= M_BCAST;
2107 else
2108 m->m_flags |= M_MCAST;
2109
2110 /*
2111 * Make a deep copy of the packet and enqueue the copy
2112 * for bridge processing; return the original packet for
2113 * local processing.
2114 */
2115 mc = m_dup(m, M_DONTWAIT);
2116 if (mc == NULL) {
2117 BRIDGE_UNLOCK(sc);
2118 return (m);
2119 }
2120
2121 /* Perform the bridge forwarding function with the copy. */
2122 bridge_forward(sc, bif, mc);
2123
2124 /*
2125 * Reinject the mbuf as arriving on the bridge so we have a
2126 * chance at claiming multicast packets. We can not loop back
2127 * here from ether_input as a bridge is never a member of a
2128 * bridge.
2129 */
2130 KASSERT(bifp->if_bridge == NULL,
2131 ("loop created in bridge_input"));
2132 mc2 = m_dup(m, M_DONTWAIT);
2133 if (mc2 != NULL) {
2134 /* Keep the layer3 header aligned */
2135 int i = min(mc2->m_pkthdr.len, max_protohdr);
2136 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2137 }
2138 if (mc2 != NULL) {
2139 mc2->m_pkthdr.rcvif = bifp;
2140 (*bifp->if_input)(bifp, mc2);
2141 }
2142
2143 /* Return the original packet for local processing. */
2144 return (m);
2145 }
2146
2147 if ((bif->bif_flags & IFBIF_STP) &&
2148 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2149 BRIDGE_UNLOCK(sc);
2150 return (m);
2151 }
2152
2153 #ifdef DEV_CARP
2154 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2155 || ((iface)->if_carp \
2156 && carp_forus((iface)->if_carp, eh->ether_dhost))
2157 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2158 || ((iface)->if_carp \
2159 && carp_forus((iface)->if_carp, eh->ether_shost))
2160 #else
2161 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2162 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2163 #endif
2164
2165 #ifdef INET6
2166 # define OR_PFIL_HOOKED_INET6 \
2167 || inet6_pfil_hook.ph_busy_count >= 0
2168 #else
2169 # define OR_PFIL_HOOKED_INET6
2170 #endif
2171
2172 #define GRAB_OUR_PACKETS(iface) \
2173 if ((iface)->if_type == IFT_GIF) \
2174 continue; \
2175 /* It is destined for us. */ \
2176 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2177 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2178 ) { \
2179 if ((iface)->if_type == IFT_BRIDGE) { \
2180 ETHER_BPF_MTAP(iface, m); \
2181 iface->if_ipackets++; \
2182 /* Filter on the physical interface. */ \
2183 if (pfil_local_phys && \
2184 (inet_pfil_hook.ph_busy_count >= 0 \
2185 OR_PFIL_HOOKED_INET6)) { \
2186 if (bridge_pfil(&m, NULL, ifp, \
2187 PFIL_IN) != 0 || m == NULL) { \
2188 BRIDGE_UNLOCK(sc); \
2189 return (NULL); \
2190 } \
2191 } \
2192 } \
2193 if (bif->bif_flags & IFBIF_LEARNING) \
2194 (void) bridge_rtupdate(sc, \
2195 eh->ether_shost, bif, 0, IFBAF_DYNAMIC); \
2196 m->m_pkthdr.rcvif = iface; \
2197 BRIDGE_UNLOCK(sc); \
2198 return (m); \
2199 } \
2200 \
2201 /* We just received a packet that we sent out. */ \
2202 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2203 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2204 ) { \
2205 BRIDGE_UNLOCK(sc); \
2206 m_freem(m); \
2207 return (NULL); \
2208 }
2209
2210 /*
2211 * Unicast. Make sure it's not for the bridge.
2212 */
2213 do { GRAB_OUR_PACKETS(bifp) } while (0);
2214
2215 /*
2216 * Give a chance for ifp at first priority. This will help when the
2217 * packet comes through the interface like VLAN's with the same MACs
2218 * on several interfaces from the same bridge. This also will save
2219 * some CPU cycles in case the destination interface and the input
2220 * interface (eq ifp) are the same.
2221 */
2222 do { GRAB_OUR_PACKETS(ifp) } while (0);
2223
2224 /* Now check the all bridge members. */
2225 LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2226 GRAB_OUR_PACKETS(bif2->bif_ifp)
2227 }
2228
2229 #undef OR_CARP_CHECK_WE_ARE_DST
2230 #undef OR_CARP_CHECK_WE_ARE_SRC
2231 #undef OR_PFIL_HOOKED_INET6
2232 #undef GRAB_OUR_PACKETS
2233
2234 /* Perform the bridge forwarding function. */
2235 bridge_forward(sc, bif, m);
2236
2237 return (NULL);
2238 }
2239
2240 /*
2241 * bridge_broadcast:
2242 *
2243 * Send a frame to all interfaces that are members of
2244 * the bridge, except for the one on which the packet
2245 * arrived.
2246 *
2247 * NOTE: Releases the lock on return.
2248 */
2249 static void
2250 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2251 struct mbuf *m, int runfilt)
2252 {
2253 struct bridge_iflist *dbif, *sbif;
2254 struct mbuf *mc;
2255 struct ifnet *dst_if;
2256 int error = 0, used = 0, i;
2257
2258 sbif = bridge_lookup_member_if(sc, src_if);
2259
2260 BRIDGE_LOCK2REF(sc, error);
2261 if (error) {
2262 m_freem(m);
2263 return;
2264 }
2265
2266 /* Filter on the bridge interface before broadcasting */
2267 if (runfilt && (inet_pfil_hook.ph_busy_count >= 0
2268 #ifdef INET6
2269 || inet6_pfil_hook.ph_busy_count >= 0
2270 #endif
2271 )) {
2272 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2273 goto out;
2274 if (m == NULL)
2275 goto out;
2276 }
2277
2278 LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2279 dst_if = dbif->bif_ifp;
2280 if (dst_if == src_if)
2281 continue;
2282
2283 /* Private segments can not talk to each other */
2284 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2285 continue;
2286
2287 if ((dbif->bif_flags & IFBIF_STP) &&
2288 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2289 continue;
2290
2291 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2292 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2293 continue;
2294
2295 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2296 continue;
2297
2298 if (LIST_NEXT(dbif, bif_next) == NULL) {
2299 mc = m;
2300 used = 1;
2301 } else {
2302 mc = m_dup(m, M_DONTWAIT);
2303 if (mc == NULL) {
2304 sc->sc_ifp->if_oerrors++;
2305 continue;
2306 }
2307 }
2308
2309 /*
2310 * Filter on the output interface. Pass a NULL bridge interface
2311 * pointer so we do not redundantly filter on the bridge for
2312 * each interface we broadcast on.
2313 */
2314 if (runfilt && (inet_pfil_hook.ph_busy_count >= 0
2315 #ifdef INET6
2316 || inet6_pfil_hook.ph_busy_count >= 0
2317 #endif
2318 )) {
2319 if (used == 0) {
2320 /* Keep the layer3 header aligned */
2321 i = min(mc->m_pkthdr.len, max_protohdr);
2322 mc = m_copyup(mc, i, ETHER_ALIGN);
2323 if (mc == NULL) {
2324 sc->sc_ifp->if_oerrors++;
2325 continue;
2326 }
2327 }
2328 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2329 continue;
2330 if (mc == NULL)
2331 continue;
2332 }
2333
2334 bridge_enqueue(sc, dst_if, mc);
2335 }
2336 if (used == 0)
2337 m_freem(m);
2338
2339 out:
2340 BRIDGE_UNREF(sc);
2341 }
2342
2343 /*
2344 * bridge_span:
2345 *
2346 * Duplicate a packet out one or more interfaces that are in span mode,
2347 * the original mbuf is unmodified.
2348 */
2349 static void
2350 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2351 {
2352 struct bridge_iflist *bif;
2353 struct ifnet *dst_if;
2354 struct mbuf *mc;
2355
2356 if (LIST_EMPTY(&sc->sc_spanlist))
2357 return;
2358
2359 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2360 dst_if = bif->bif_ifp;
2361
2362 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2363 continue;
2364
2365 mc = m_copypacket(m, M_DONTWAIT);
2366 if (mc == NULL) {
2367 sc->sc_ifp->if_oerrors++;
2368 continue;
2369 }
2370
2371 bridge_enqueue(sc, dst_if, mc);
2372 }
2373 }
2374
2375 /*
2376 * bridge_rtupdate:
2377 *
2378 * Add a bridge routing entry.
2379 */
2380 static int
2381 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst,
2382 struct bridge_iflist *bif, int setflags, uint8_t flags)
2383 {
2384 struct bridge_rtnode *brt;
2385 struct ifnet *dst_if = bif->bif_ifp;
2386 int error;
2387
2388 BRIDGE_LOCK_ASSERT(sc);
2389
2390 /*
2391 * A route for this destination might already exist. If so,
2392 * update it, otherwise create a new one.
2393 */
2394 if ((brt = bridge_rtnode_lookup(sc, dst)) == NULL) {
2395 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2396 sc->sc_brtexceeded++;
2397 return (ENOSPC);
2398 }
2399
2400 /*
2401 * Allocate a new bridge forwarding node, and
2402 * initialize the expiration time and Ethernet
2403 * address.
2404 */
2405 brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2406 if (brt == NULL)
2407 return (ENOMEM);
2408
2409 if (bif->bif_flags & IFBIF_STICKY)
2410 brt->brt_flags = IFBAF_STICKY;
2411 else
2412 brt->brt_flags = IFBAF_DYNAMIC;
2413
2414 brt->brt_ifp = dst_if;
2415 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2416
2417 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2418 uma_zfree(bridge_rtnode_zone, brt);
2419 return (error);
2420 }
2421 }
2422
2423 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2424 brt->brt_ifp = dst_if;
2425 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2426 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2427 if (setflags)
2428 brt->brt_flags = flags;
2429
2430 return (0);
2431 }
2432
2433 /*
2434 * bridge_rtlookup:
2435 *
2436 * Lookup the destination interface for an address.
2437 */
2438 static struct ifnet *
2439 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr)
2440 {
2441 struct bridge_rtnode *brt;
2442
2443 BRIDGE_LOCK_ASSERT(sc);
2444
2445 if ((brt = bridge_rtnode_lookup(sc, addr)) == NULL)
2446 return (NULL);
2447
2448 return (brt->brt_ifp);
2449 }
2450
2451 /*
2452 * bridge_rttrim:
2453 *
2454 * Trim the routine table so that we have a number
2455 * of routing entries less than or equal to the
2456 * maximum number.
2457 */
2458 static void
2459 bridge_rttrim(struct bridge_softc *sc)
2460 {
2461 struct bridge_rtnode *brt, *nbrt;
2462
2463 BRIDGE_LOCK_ASSERT(sc);
2464
2465 /* Make sure we actually need to do this. */
2466 if (sc->sc_brtcnt <= sc->sc_brtmax)
2467 return;
2468
2469 /* Force an aging cycle; this might trim enough addresses. */
2470 bridge_rtage(sc);
2471 if (sc->sc_brtcnt <= sc->sc_brtmax)
2472 return;
2473
2474 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2475 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2476 bridge_rtnode_destroy(sc, brt);
2477 if (sc->sc_brtcnt <= sc->sc_brtmax)
2478 return;
2479 }
2480 }
2481 }
2482
2483 /*
2484 * bridge_timer:
2485 *
2486 * Aging timer for the bridge.
2487 */
2488 static void
2489 bridge_timer(void *arg)
2490 {
2491 struct bridge_softc *sc = arg;
2492
2493 BRIDGE_LOCK_ASSERT(sc);
2494
2495 bridge_rtage(sc);
2496
2497 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2498 callout_reset(&sc->sc_brcallout,
2499 bridge_rtable_prune_period * hz, bridge_timer, sc);
2500 }
2501
2502 /*
2503 * bridge_rtage:
2504 *
2505 * Perform an aging cycle.
2506 */
2507 static void
2508 bridge_rtage(struct bridge_softc *sc)
2509 {
2510 struct bridge_rtnode *brt, *nbrt;
2511
2512 BRIDGE_LOCK_ASSERT(sc);
2513
2514 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2515 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2516 if (time_uptime >= brt->brt_expire)
2517 bridge_rtnode_destroy(sc, brt);
2518 }
2519 }
2520 }
2521
2522 /*
2523 * bridge_rtflush:
2524 *
2525 * Remove all dynamic addresses from the bridge.
2526 */
2527 static void
2528 bridge_rtflush(struct bridge_softc *sc, int full)
2529 {
2530 struct bridge_rtnode *brt, *nbrt;
2531
2532 BRIDGE_LOCK_ASSERT(sc);
2533
2534 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2535 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2536 bridge_rtnode_destroy(sc, brt);
2537 }
2538 }
2539
2540 /*
2541 * bridge_rtdaddr:
2542 *
2543 * Remove an address from the table.
2544 */
2545 static int
2546 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr)
2547 {
2548 struct bridge_rtnode *brt;
2549
2550 BRIDGE_LOCK_ASSERT(sc);
2551
2552 if ((brt = bridge_rtnode_lookup(sc, addr)) == NULL)
2553 return (ENOENT);
2554
2555 bridge_rtnode_destroy(sc, brt);
2556 return (0);
2557 }
2558
2559 /*
2560 * bridge_rtdelete:
2561 *
2562 * Delete routes to a speicifc member interface.
2563 */
2564 static void
2565 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2566 {
2567 struct bridge_rtnode *brt, *nbrt;
2568
2569 BRIDGE_LOCK_ASSERT(sc);
2570
2571 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2572 if (brt->brt_ifp == ifp && (full ||
2573 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2574 bridge_rtnode_destroy(sc, brt);
2575 }
2576 }
2577
2578 /*
2579 * bridge_rtable_init:
2580 *
2581 * Initialize the route table for this bridge.
2582 */
2583 static int
2584 bridge_rtable_init(struct bridge_softc *sc)
2585 {
2586 int i;
2587
2588 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2589 M_DEVBUF, M_NOWAIT);
2590 if (sc->sc_rthash == NULL)
2591 return (ENOMEM);
2592
2593 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2594 LIST_INIT(&sc->sc_rthash[i]);
2595
2596 sc->sc_rthash_key = arc4random();
2597
2598 LIST_INIT(&sc->sc_rtlist);
2599
2600 return (0);
2601 }
2602
2603 /*
2604 * bridge_rtable_fini:
2605 *
2606 * Deconstruct the route table for this bridge.
2607 */
2608 static void
2609 bridge_rtable_fini(struct bridge_softc *sc)
2610 {
2611
2612 free(sc->sc_rthash, M_DEVBUF);
2613 }
2614
2615 /*
2616 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2617 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2618 */
2619 #define mix(a, b, c) \
2620 do { \
2621 a -= b; a -= c; a ^= (c >> 13); \
2622 b -= c; b -= a; b ^= (a << 8); \
2623 c -= a; c -= b; c ^= (b >> 13); \
2624 a -= b; a -= c; a ^= (c >> 12); \
2625 b -= c; b -= a; b ^= (a << 16); \
2626 c -= a; c -= b; c ^= (b >> 5); \
2627 a -= b; a -= c; a ^= (c >> 3); \
2628 b -= c; b -= a; b ^= (a << 10); \
2629 c -= a; c -= b; c ^= (b >> 15); \
2630 } while (/*CONSTCOND*/0)
2631
2632 static __inline uint32_t
2633 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2634 {
2635 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2636
2637 b += addr[5] << 8;
2638 b += addr[4];
2639 a += addr[3] << 24;
2640 a += addr[2] << 16;
2641 a += addr[1] << 8;
2642 a += addr[0];
2643
2644 mix(a, b, c);
2645
2646 return (c & BRIDGE_RTHASH_MASK);
2647 }
2648
2649 #undef mix
2650
2651 static int
2652 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2653 {
2654 int i, d;
2655
2656 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2657 d = ((int)a[i]) - ((int)b[i]);
2658 }
2659
2660 return (d);
2661 }
2662
2663 /*
2664 * bridge_rtnode_lookup:
2665 *
2666 * Look up a bridge route node for the specified destination.
2667 */
2668 static struct bridge_rtnode *
2669 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr)
2670 {
2671 struct bridge_rtnode *brt;
2672 uint32_t hash;
2673 int dir;
2674
2675 BRIDGE_LOCK_ASSERT(sc);
2676
2677 hash = bridge_rthash(sc, addr);
2678 LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2679 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2680 if (dir == 0)
2681 return (brt);
2682 if (dir > 0)
2683 return (NULL);
2684 }
2685
2686 return (NULL);
2687 }
2688
2689 /*
2690 * bridge_rtnode_insert:
2691 *
2692 * Insert the specified bridge node into the route table. We
2693 * assume the entry is not already in the table.
2694 */
2695 static int
2696 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2697 {
2698 struct bridge_rtnode *lbrt;
2699 uint32_t hash;
2700 int dir;
2701
2702 BRIDGE_LOCK_ASSERT(sc);
2703
2704 hash = bridge_rthash(sc, brt->brt_addr);
2705
2706 lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2707 if (lbrt == NULL) {
2708 LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2709 goto out;
2710 }
2711
2712 do {
2713 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2714 if (dir == 0)
2715 return (EEXIST);
2716 if (dir > 0) {
2717 LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2718 goto out;
2719 }
2720 if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2721 LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2722 goto out;
2723 }
2724 lbrt = LIST_NEXT(lbrt, brt_hash);
2725 } while (lbrt != NULL);
2726
2727 #ifdef DIAGNOSTIC
2728 panic("bridge_rtnode_insert: impossible");
2729 #endif
2730
2731 out:
2732 LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2733 sc->sc_brtcnt++;
2734
2735 return (0);
2736 }
2737
2738 /*
2739 * bridge_rtnode_destroy:
2740 *
2741 * Destroy a bridge rtnode.
2742 */
2743 static void
2744 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2745 {
2746 BRIDGE_LOCK_ASSERT(sc);
2747
2748 LIST_REMOVE(brt, brt_hash);
2749
2750 LIST_REMOVE(brt, brt_list);
2751 sc->sc_brtcnt--;
2752 uma_zfree(bridge_rtnode_zone, brt);
2753 }
2754
2755 /*
2756 * bridge_rtable_expire:
2757 *
2758 * Set the expiry time for all routes on an interface.
2759 */
2760 static void
2761 bridge_rtable_expire(struct ifnet *ifp, int age)
2762 {
2763 struct bridge_softc *sc = ifp->if_bridge;
2764 struct bridge_rtnode *brt;
2765
2766 BRIDGE_LOCK(sc);
2767
2768 /*
2769 * If the age is zero then flush, otherwise set all the expiry times to
2770 * age for the interface
2771 */
2772 if (age == 0)
2773 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2774 else {
2775 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2776 /* Cap the expiry time to 'age' */
2777 if (brt->brt_ifp == ifp &&
2778 brt->brt_expire > time_uptime + age &&
2779 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2780 brt->brt_expire = time_uptime + age;
2781 }
2782 }
2783 BRIDGE_UNLOCK(sc);
2784 }
2785
2786 /*
2787 * bridge_state_change:
2788 *
2789 * Callback from the bridgestp code when a port changes states.
2790 */
2791 static void
2792 bridge_state_change(struct ifnet *ifp, int state)
2793 {
2794 struct bridge_softc *sc = ifp->if_bridge;
2795 static const char *stpstates[] = {
2796 "disabled",
2797 "listening",
2798 "learning",
2799 "forwarding",
2800 "blocking",
2801 "discarding"
2802 };
2803
2804 if (log_stp)
2805 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
2806 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
2807 }
2808
2809 /*
2810 * Send bridge packets through pfil if they are one of the types pfil can deal
2811 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
2812 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
2813 * that interface.
2814 */
2815 static int
2816 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
2817 {
2818 int snap, error, i, hlen;
2819 struct ether_header *eh1, eh2;
2820 struct ip_fw_args args;
2821 struct ip *ip;
2822 struct llc llc1;
2823 u_int16_t ether_type;
2824
2825 snap = 0;
2826 error = -1; /* Default error if not error == 0 */
2827
2828 #if 0
2829 /* we may return with the IP fields swapped, ensure its not shared */
2830 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
2831 #endif
2832
2833 if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
2834 return (0); /* filtering is disabled */
2835
2836 i = min((*mp)->m_pkthdr.len, max_protohdr);
2837 if ((*mp)->m_len < i) {
2838 *mp = m_pullup(*mp, i);
2839 if (*mp == NULL) {
2840 printf("%s: m_pullup failed\n", __func__);
2841 return (-1);
2842 }
2843 }
2844
2845 eh1 = mtod(*mp, struct ether_header *);
2846 ether_type = ntohs(eh1->ether_type);
2847
2848 /*
2849 * Check for SNAP/LLC.
2850 */
2851 if (ether_type < ETHERMTU) {
2852 struct llc *llc2 = (struct llc *)(eh1 + 1);
2853
2854 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
2855 llc2->llc_dsap == LLC_SNAP_LSAP &&
2856 llc2->llc_ssap == LLC_SNAP_LSAP &&
2857 llc2->llc_control == LLC_UI) {
2858 ether_type = htons(llc2->llc_un.type_snap.ether_type);
2859 snap = 1;
2860 }
2861 }
2862
2863 /*
2864 * If we're trying to filter bridge traffic, don't look at anything
2865 * other than IP and ARP traffic. If the filter doesn't understand
2866 * IPv6, don't allow IPv6 through the bridge either. This is lame
2867 * since if we really wanted, say, an AppleTalk filter, we are hosed,
2868 * but of course we don't have an AppleTalk filter to begin with.
2869 * (Note that since pfil doesn't understand ARP it will pass *ALL*
2870 * ARP traffic.)
2871 */
2872 switch (ether_type) {
2873 case ETHERTYPE_ARP:
2874 case ETHERTYPE_REVARP:
2875 if (pfil_ipfw_arp == 0)
2876 return (0); /* Automatically pass */
2877 break;
2878
2879 case ETHERTYPE_IP:
2880 #ifdef INET6
2881 case ETHERTYPE_IPV6:
2882 #endif /* INET6 */
2883 break;
2884 default:
2885 /*
2886 * Check to see if the user wants to pass non-ip
2887 * packets, these will not be checked by pfil(9) and
2888 * passed unconditionally so the default is to drop.
2889 */
2890 if (pfil_onlyip)
2891 goto bad;
2892 }
2893
2894 /* Strip off the Ethernet header and keep a copy. */
2895 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
2896 m_adj(*mp, ETHER_HDR_LEN);
2897
2898 /* Strip off snap header, if present */
2899 if (snap) {
2900 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
2901 m_adj(*mp, sizeof(struct llc));
2902 }
2903
2904 /*
2905 * Check the IP header for alignment and errors
2906 */
2907 if (dir == PFIL_IN) {
2908 switch (ether_type) {
2909 case ETHERTYPE_IP:
2910 error = bridge_ip_checkbasic(mp);
2911 break;
2912 #ifdef INET6
2913 case ETHERTYPE_IPV6:
2914 error = bridge_ip6_checkbasic(mp);
2915 break;
2916 #endif /* INET6 */
2917 default:
2918 error = 0;
2919 }
2920 if (error)
2921 goto bad;
2922 }
2923
2924 if (IPFW_LOADED && pfil_ipfw != 0 && dir == PFIL_OUT && ifp != NULL) {
2925 error = -1;
2926 args.rule = ip_dn_claim_rule(*mp);
2927 if (args.rule != NULL && fw_one_pass)
2928 goto ipfwpass; /* packet already partially processed */
2929
2930 args.m = *mp;
2931 args.oif = ifp;
2932 args.next_hop = NULL;
2933 args.eh = &eh2;
2934 args.inp = NULL; /* used by ipfw uid/gid/jail rules */
2935 i = ip_fw_chk_ptr(&args);
2936 *mp = args.m;
2937
2938 if (*mp == NULL)
2939 return (error);
2940
2941 if (DUMMYNET_LOADED && (i == IP_FW_DUMMYNET)) {
2942
2943 /* put the Ethernet header back on */
2944 M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
2945 if (*mp == NULL)
2946 return (error);
2947 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
2948
2949 /*
2950 * Pass the pkt to dummynet, which consumes it. The
2951 * packet will return to us via bridge_dummynet().
2952 */
2953 args.oif = ifp;
2954 ip_dn_io_ptr(mp, DN_TO_IFB_FWD, &args);
2955 return (error);
2956 }
2957
2958 if (i != IP_FW_PASS) /* drop */
2959 goto bad;
2960 }
2961
2962 ipfwpass:
2963 error = 0;
2964
2965 /*
2966 * Run the packet through pfil
2967 */
2968 switch (ether_type) {
2969 case ETHERTYPE_IP:
2970 /*
2971 * before calling the firewall, swap fields the same as
2972 * IP does. here we assume the header is contiguous
2973 */
2974 ip = mtod(*mp, struct ip *);
2975
2976 ip->ip_len = ntohs(ip->ip_len);
2977 ip->ip_off = ntohs(ip->ip_off);
2978
2979 /*
2980 * Run pfil on the member interface and the bridge, both can
2981 * be skipped by clearing pfil_member or pfil_bridge.
2982 *
2983 * Keep the order:
2984 * in_if -> bridge_if -> out_if
2985 */
2986 if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
2987 error = pfil_run_hooks(&inet_pfil_hook, mp, bifp,
2988 dir, NULL);
2989
2990 if (*mp == NULL || error != 0) /* filter may consume */
2991 break;
2992
2993 if (pfil_member && ifp != NULL)
2994 error = pfil_run_hooks(&inet_pfil_hook, mp, ifp,
2995 dir, NULL);
2996
2997 if (*mp == NULL || error != 0) /* filter may consume */
2998 break;
2999
3000 if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3001 error = pfil_run_hooks(&inet_pfil_hook, mp, bifp,
3002 dir, NULL);
3003
3004 if (*mp == NULL || error != 0) /* filter may consume */
3005 break;
3006
3007 /* check if we need to fragment the packet */
3008 if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3009 i = (*mp)->m_pkthdr.len;
3010 if (i > ifp->if_mtu) {
3011 error = bridge_fragment(ifp, *mp, &eh2, snap,
3012 &llc1);
3013 return (error);
3014 }
3015 }
3016
3017 /* Recalculate the ip checksum and restore byte ordering */
3018 ip = mtod(*mp, struct ip *);
3019 hlen = ip->ip_hl << 2;
3020 if (hlen < sizeof(struct ip))
3021 goto bad;
3022 if (hlen > (*mp)->m_len) {
3023 if ((*mp = m_pullup(*mp, hlen)) == 0)
3024 goto bad;
3025 ip = mtod(*mp, struct ip *);
3026 if (ip == NULL)
3027 goto bad;
3028 }
3029 ip->ip_len = htons(ip->ip_len);
3030 ip->ip_off = htons(ip->ip_off);
3031 ip->ip_sum = 0;
3032 if (hlen == sizeof(struct ip))
3033 ip->ip_sum = in_cksum_hdr(ip);
3034 else
3035 ip->ip_sum = in_cksum(*mp, hlen);
3036
3037 break;
3038 #ifdef INET6
3039 case ETHERTYPE_IPV6:
3040 if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3041 error = pfil_run_hooks(&inet6_pfil_hook, mp, bifp,
3042 dir, NULL);
3043
3044 if (*mp == NULL || error != 0) /* filter may consume */
3045 break;
3046
3047 if (pfil_member && ifp != NULL)
3048 error = pfil_run_hooks(&inet6_pfil_hook, mp, ifp,
3049 dir, NULL);
3050
3051 if (*mp == NULL || error != 0) /* filter may consume */
3052 break;
3053
3054 if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3055 error = pfil_run_hooks(&inet6_pfil_hook, mp, bifp,
3056 dir, NULL);
3057 break;
3058 #endif
3059 default:
3060 error = 0;
3061 break;
3062 }
3063
3064 if (*mp == NULL)
3065 return (error);
3066 if (error != 0)
3067 goto bad;
3068
3069 error = -1;
3070
3071 /*
3072 * Finally, put everything back the way it was and return
3073 */
3074 if (snap) {
3075 M_PREPEND(*mp, sizeof(struct llc), M_DONTWAIT);
3076 if (*mp == NULL)
3077 return (error);
3078 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3079 }
3080
3081 M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3082 if (*mp == NULL)
3083 return (error);
3084 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3085
3086 return (0);
3087
3088 bad:
3089 m_freem(*mp);
3090 *mp = NULL;
3091 return (error);
3092 }
3093
3094 /*
3095 * Perform basic checks on header size since
3096 * pfil assumes ip_input has already processed
3097 * it for it. Cut-and-pasted from ip_input.c.
3098 * Given how simple the IPv6 version is,
3099 * does the IPv4 version really need to be
3100 * this complicated?
3101 *
3102 * XXX Should we update ipstat here, or not?
3103 * XXX Right now we update ipstat but not
3104 * XXX csum_counter.
3105 */
3106 static int
3107 bridge_ip_checkbasic(struct mbuf **mp)
3108 {
3109 struct mbuf *m = *mp;
3110 struct ip *ip;
3111 int len, hlen;
3112 u_short sum;
3113
3114 if (*mp == NULL)
3115 return (-1);
3116
3117 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3118 if ((m = m_copyup(m, sizeof(struct ip),
3119 (max_linkhdr + 3) & ~3)) == NULL) {
3120 /* XXXJRT new stat, please */
3121 ipstat.ips_toosmall++;
3122 goto bad;
3123 }
3124 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3125 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3126 ipstat.ips_toosmall++;
3127 goto bad;
3128 }
3129 }
3130 ip = mtod(m, struct ip *);
3131 if (ip == NULL) goto bad;
3132
3133 if (ip->ip_v != IPVERSION) {
3134 ipstat.ips_badvers++;
3135 goto bad;
3136 }
3137 hlen = ip->ip_hl << 2;
3138 if (hlen < sizeof(struct ip)) { /* minimum header length */
3139 ipstat.ips_badhlen++;
3140 goto bad;
3141 }
3142 if (hlen > m->m_len) {
3143 if ((m = m_pullup(m, hlen)) == 0) {
3144 ipstat.ips_badhlen++;
3145 goto bad;
3146 }
3147 ip = mtod(m, struct ip *);
3148 if (ip == NULL) goto bad;
3149 }
3150
3151 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3152 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3153 } else {
3154 if (hlen == sizeof(struct ip)) {
3155 sum = in_cksum_hdr(ip);
3156 } else {
3157 sum = in_cksum(m, hlen);
3158 }
3159 }
3160 if (sum) {
3161 ipstat.ips_badsum++;
3162 goto bad;
3163 }
3164
3165 /* Retrieve the packet length. */
3166 len = ntohs(ip->ip_len);
3167
3168 /*
3169 * Check for additional length bogosity
3170 */
3171 if (len < hlen) {
3172 ipstat.ips_badlen++;
3173 goto bad;
3174 }
3175
3176 /*
3177 * Check that the amount of data in the buffers
3178 * is as at least much as the IP header would have us expect.
3179 * Drop packet if shorter than we expect.
3180 */
3181 if (m->m_pkthdr.len < len) {
3182 ipstat.ips_tooshort++;
3183 goto bad;
3184 }
3185
3186 /* Checks out, proceed */
3187 *mp = m;
3188 return (0);
3189
3190 bad:
3191 *mp = m;
3192 return (-1);
3193 }
3194
3195 #ifdef INET6
3196 /*
3197 * Same as above, but for IPv6.
3198 * Cut-and-pasted from ip6_input.c.
3199 * XXX Should we update ip6stat, or not?
3200 */
3201 static int
3202 bridge_ip6_checkbasic(struct mbuf **mp)
3203 {
3204 struct mbuf *m = *mp;
3205 struct ip6_hdr *ip6;
3206
3207 /*
3208 * If the IPv6 header is not aligned, slurp it up into a new
3209 * mbuf with space for link headers, in the event we forward
3210 * it. Otherwise, if it is aligned, make sure the entire base
3211 * IPv6 header is in the first mbuf of the chain.
3212 */
3213 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3214 struct ifnet *inifp = m->m_pkthdr.rcvif;
3215 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3216 (max_linkhdr + 3) & ~3)) == NULL) {
3217 /* XXXJRT new stat, please */
3218 ip6stat.ip6s_toosmall++;
3219 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3220 goto bad;
3221 }
3222 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3223 struct ifnet *inifp = m->m_pkthdr.rcvif;
3224 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3225 ip6stat.ip6s_toosmall++;
3226 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3227 goto bad;
3228 }
3229 }
3230
3231 ip6 = mtod(m, struct ip6_hdr *);
3232
3233 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3234 ip6stat.ip6s_badvers++;
3235 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3236 goto bad;
3237 }
3238
3239 /* Checks out, proceed */
3240 *mp = m;
3241 return (0);
3242
3243 bad:
3244 *mp = m;
3245 return (-1);
3246 }
3247 #endif /* INET6 */
3248
3249 /*
3250 * bridge_fragment:
3251 *
3252 * Return a fragmented mbuf chain.
3253 */
3254 static int
3255 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3256 int snap, struct llc *llc)
3257 {
3258 struct mbuf *m0;
3259 struct ip *ip;
3260 int error = -1;
3261
3262 if (m->m_len < sizeof(struct ip) &&
3263 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3264 goto out;
3265 ip = mtod(m, struct ip *);
3266
3267 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist,
3268 CSUM_DELAY_IP);
3269 if (error)
3270 goto out;
3271
3272 /* walk the chain and re-add the Ethernet header */
3273 for (m0 = m; m0; m0 = m0->m_nextpkt) {
3274 if (error == 0) {
3275 if (snap) {
3276 M_PREPEND(m0, sizeof(struct llc), M_DONTWAIT);
3277 if (m0 == NULL) {
3278 error = ENOBUFS;
3279 continue;
3280 }
3281 bcopy(llc, mtod(m0, caddr_t),
3282 sizeof(struct llc));
3283 }
3284 M_PREPEND(m0, ETHER_HDR_LEN, M_DONTWAIT);
3285 if (m0 == NULL) {
3286 error = ENOBUFS;
3287 continue;
3288 }
3289 bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3290 } else
3291 m_freem(m);
3292 }
3293
3294 if (error == 0)
3295 ipstat.ips_fragmented++;
3296
3297 return (error);
3298
3299 out:
3300 if (m != NULL)
3301 m_freem(m);
3302 return (error);
3303 }
Cache object: 9c53cad19a9bbc5880e1e3b33d6e6760
|