xref: /freebsd/sys/net/if_bridge.c (revision 6a888f62413a1a6117f5053f124c97277ed18484)
1 /*	$NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $	*/
2 
3 /*-
4  * SPDX-License-Identifier: BSD-4-Clause
5  *
6  * Copyright 2001 Wasabi Systems, Inc.
7  * All rights reserved.
8  *
9  * Written by Jason R. Thorpe for Wasabi Systems, Inc.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
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
25  *    written permission.
26  *
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.
38  */
39 
40 /*
41  * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
42  * All rights reserved.
43  *
44  * Redistribution and use in source and binary forms, with or without
45  * modification, are permitted provided that the following conditions
46  * are met:
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.
52  *
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.
64  *
65  * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
66  */
67 
68 /*
69  * Network interface bridge support.
70  *
71  * TODO:
72  *
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).
77  */
78 
79 #include "opt_inet.h"
80 #include "opt_inet6.h"
81 
82 #define	EXTERR_CATEGORY	EXTERR_CAT_BRIDGE
83 
84 #include <sys/param.h>
85 #include <sys/ctype.h>  /* string functions */
86 #include <sys/eventhandler.h>
87 #include <sys/exterrvar.h>
88 #include <sys/jail.h>
89 #include <sys/kernel.h>
90 #include <sys/lock.h>
91 #include <sys/malloc.h>
92 #include <sys/mbuf.h>
93 #include <sys/module.h>
94 #include <sys/mutex.h>
95 #include <sys/priv.h>
96 #include <sys/proc.h>
97 #include <sys/protosw.h>
98 #include <sys/random.h>
99 #include <sys/systm.h>
100 #include <sys/socket.h> /* for net/if.h */
101 #include <sys/sockio.h>
102 #include <sys/syslog.h>
103 #include <sys/sysctl.h>
104 #include <sys/time.h>
105 
106 #include <vm/uma.h>
107 
108 #include <net/bpf.h>
109 #include <net/if.h>
110 #include <net/if_clone.h>
111 #include <net/if_dl.h>
112 #include <net/if_types.h>
113 #include <net/if_var.h>
114 #include <net/if_private.h>
115 #include <net/pfil.h>
116 #include <net/vnet.h>
117 
118 #include <netinet/in.h>
119 #include <netinet/in_systm.h>
120 #include <netinet/in_var.h>
121 #include <netinet/ip.h>
122 #include <netinet/ip_var.h>
123 #ifdef INET6
124 #include <netinet/ip6.h>
125 #include <netinet6/ip6_var.h>
126 #include <netinet6/in6_ifattach.h>
127 #endif
128 #if defined(INET) || defined(INET6)
129 #include <netinet/ip_carp.h>
130 #endif
131 #include <machine/in_cksum.h>
132 #include <netinet/if_ether.h>
133 #include <net/bridgestp.h>
134 #include <net/if_bridgevar.h>
135 #include <net/if_llc.h>
136 #include <net/if_vlan_var.h>
137 
138 #include <net/route.h>
139 
140 /*
141  * At various points in the code we need to know if we're hooked into the INET
142  * and/or INET6 pfil.  Define some macros to do that based on which IP versions
143  * are enabled in the kernel.  This avoids littering the rest of the code with
144  * #ifnet INET6 to avoid referencing V_inet6_pfil_head.
145  */
146 #ifdef INET6
147 #define		PFIL_HOOKED_IN_INET6	PFIL_HOOKED_IN(V_inet6_pfil_head)
148 #define		PFIL_HOOKED_OUT_INET6	PFIL_HOOKED_OUT(V_inet6_pfil_head)
149 #else
150 #define		PFIL_HOOKED_IN_INET6	false
151 #define		PFIL_HOOKED_OUT_INET6	false
152 #endif
153 
154 #ifdef INET
155 #define		PFIL_HOOKED_IN_INET	PFIL_HOOKED_IN(V_inet_pfil_head)
156 #define		PFIL_HOOKED_OUT_INET	PFIL_HOOKED_OUT(V_inet_pfil_head)
157 #else
158 #define		PFIL_HOOKED_IN_INET	false
159 #define		PFIL_HOOKED_OUT_INET	false
160 #endif
161 
162 #define		PFIL_HOOKED_IN_46	(PFIL_HOOKED_IN_INET6 || PFIL_HOOKED_IN_INET)
163 #define		PFIL_HOOKED_OUT_46	(PFIL_HOOKED_OUT_INET6 || PFIL_HOOKED_OUT_INET)
164 
165 /*
166  * Size of the route hash table.  Must be a power of two.
167  */
168 #ifndef BRIDGE_RTHASH_SIZE
169 #define	BRIDGE_RTHASH_SIZE		1024
170 #endif
171 
172 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
173 
174 /*
175  * Default maximum number of addresses to cache.
176  */
177 #ifndef BRIDGE_RTABLE_MAX
178 #define	BRIDGE_RTABLE_MAX		2000
179 #endif
180 
181 /*
182  * Timeout (in seconds) for entries learned dynamically.
183  */
184 #ifndef BRIDGE_RTABLE_TIMEOUT
185 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
186 #endif
187 
188 /*
189  * Number of seconds between walks of the route list.
190  */
191 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
192 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
193 #endif
194 
195 /*
196  * List of capabilities to possibly mask on the member interface.
197  */
198 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
199 					 IFCAP_TXCSUM_IPV6|IFCAP_MEXTPG)
200 
201 /*
202  * List of capabilities to strip
203  */
204 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
205 
206 /*
207  * Bridge locking
208  *
209  * The bridge relies heavily on the epoch(9) system to protect its data
210  * structures. This means we can safely use CK_LISTs while in NET_EPOCH, but we
211  * must ensure there is only one writer at a time.
212  *
213  * That is: for read accesses we only need to be in NET_EPOCH, but for write
214  * accesses we must hold:
215  *
216  *  - BRIDGE_RT_LOCK, for any change to bridge_rtnodes
217  *  - BRIDGE_LOCK, for any other change
218  *
219  * The BRIDGE_LOCK is a sleepable lock, because it is held across ioctl()
220  * calls to bridge member interfaces and these ioctl()s can sleep.
221  * The BRIDGE_RT_LOCK is a non-sleepable mutex, because it is sometimes
222  * required while we're in NET_EPOCH and then we're not allowed to sleep.
223  */
224 #define BRIDGE_LOCK_INIT(_sc)		do {			\
225 	sx_init(&(_sc)->sc_sx, "if_bridge");			\
226 	mtx_init(&(_sc)->sc_rt_mtx, "if_bridge rt", NULL, MTX_DEF);	\
227 } while (0)
228 #define BRIDGE_LOCK_DESTROY(_sc)	do {	\
229 	sx_destroy(&(_sc)->sc_sx);		\
230 	mtx_destroy(&(_sc)->sc_rt_mtx);		\
231 } while (0)
232 #define BRIDGE_LOCK(_sc)		sx_xlock(&(_sc)->sc_sx)
233 #define BRIDGE_UNLOCK(_sc)		sx_xunlock(&(_sc)->sc_sx)
234 #define BRIDGE_LOCK_ASSERT(_sc)		sx_assert(&(_sc)->sc_sx, SX_XLOCKED)
235 #define BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(_sc)	\
236 	    MPASS(in_epoch(net_epoch_preempt) || sx_xlocked(&(_sc)->sc_sx))
237 #define BRIDGE_UNLOCK_ASSERT(_sc)	sx_assert(&(_sc)->sc_sx, SX_UNLOCKED)
238 #define BRIDGE_RT_LOCK(_sc)		mtx_lock(&(_sc)->sc_rt_mtx)
239 #define BRIDGE_RT_UNLOCK(_sc)		mtx_unlock(&(_sc)->sc_rt_mtx)
240 #define BRIDGE_RT_LOCK_ASSERT(_sc)	mtx_assert(&(_sc)->sc_rt_mtx, MA_OWNED)
241 #define BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(_sc)	\
242 	    MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(_sc)->sc_rt_mtx))
243 
244 struct bridge_softc;
245 
246 /*
247  * Bridge interface list entry.
248  */
249 struct bridge_iflist {
250 	CK_LIST_ENTRY(bridge_iflist) bif_next;
251 	struct ifnet		*bif_ifp;	/* member if */
252 	struct bridge_softc	*bif_sc;	/* parent bridge */
253 	struct bstp_port	bif_stp;	/* STP state */
254 	uint32_t		bif_flags;	/* member if flags */
255 	int			bif_savedcaps;	/* saved capabilities */
256 	uint32_t		bif_addrmax;	/* max # of addresses */
257 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
258 	uint32_t		bif_addrexceeded;/* # of address violations */
259 	struct epoch_context	bif_epoch_ctx;
260 	ether_vlanid_t		bif_pvid;	/* port vlan id */
261 	ifbvlan_set_t		bif_vlan_set;	/* if allowed tagged vlans */
262 	uint16_t		bif_vlanproto;	/* vlan protocol */
263 };
264 
265 /*
266  * Bridge route node.
267  */
268 struct bridge_rtnode {
269 	CK_LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
270 	CK_LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
271 	struct bridge_iflist	*brt_dst;	/* destination if */
272 	unsigned long		brt_expire;	/* expiration time */
273 	uint8_t			brt_flags;	/* address flags */
274 	uint8_t			brt_addr[ETHER_ADDR_LEN];
275 	ether_vlanid_t		brt_vlan;	/* vlan id */
276 	struct	vnet		*brt_vnet;
277 	struct	epoch_context	brt_epoch_ctx;
278 };
279 #define	brt_ifp			brt_dst->bif_ifp
280 
281 /*
282  * Software state for each bridge.
283  */
284 struct bridge_softc {
285 	struct ifnet		*sc_ifp;	/* make this an interface */
286 	LIST_ENTRY(bridge_softc) sc_list;
287 	struct sx		sc_sx;
288 	struct mtx		sc_rt_mtx;
289 	uint32_t		sc_brtmax;	/* max # of addresses */
290 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
291 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
292 	struct callout		sc_brcallout;	/* bridge callout */
293 	CK_LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
294 	CK_LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
295 	CK_LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
296 	uint32_t		sc_rthash_key;	/* key for hash */
297 	CK_LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
298 	struct bstp_state	sc_stp;		/* STP state */
299 	uint32_t		sc_brtexceeded;	/* # of cache drops */
300 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
301 	struct ether_addr	sc_defaddr;	/* Default MAC address */
302 	if_input_fn_t		sc_if_input;	/* Saved copy of if_input */
303 	struct epoch_context	sc_epoch_ctx;
304 	ifbr_flags_t		sc_flags;	/* bridge flags */
305 	ether_vlanid_t		sc_defpvid;	/* default PVID */
306 };
307 
308 VNET_DEFINE_STATIC(struct sx, bridge_list_sx);
309 #define	V_bridge_list_sx	VNET(bridge_list_sx)
310 static eventhandler_tag bridge_detach_cookie;
311 
312 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
313 
314 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
315 #define	V_bridge_rtnode_zone	VNET(bridge_rtnode_zone)
316 
317 static int	bridge_clone_create(struct if_clone *, char *, size_t,
318 		    struct ifc_data *, struct ifnet **);
319 static int	bridge_clone_destroy(struct if_clone *, struct ifnet *, uint32_t);
320 
321 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
322 static void	bridge_mutecaps(struct bridge_softc *);
323 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
324 		    int);
325 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
326 static void	bridge_init(void *);
327 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
328 static bool	bridge_same(const void *, const void *);
329 static void	*bridge_get_softc(struct ifnet *);
330 static void	bridge_stop(struct ifnet *, int);
331 static int	bridge_transmit(struct ifnet *, struct mbuf *);
332 #ifdef ALTQ
333 static void	bridge_altq_start(if_t);
334 static int	bridge_altq_transmit(if_t, struct mbuf *);
335 #endif
336 static void	bridge_qflush(struct ifnet *);
337 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
338 static void	bridge_inject(struct ifnet *, struct mbuf *);
339 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
340 		    struct rtentry *);
341 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
342 		    struct mbuf *, struct bridge_iflist *);
343 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
344 
345 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
346 		    struct mbuf *m);
347 static bool	bridge_member_ifaddrs(void);
348 static void	bridge_timer(void *);
349 
350 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
351 		    struct mbuf *, int);
352 static void	bridge_span(struct bridge_softc *, struct mbuf *);
353 
354 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
355 		    ether_vlanid_t, struct bridge_iflist *, int, uint8_t);
356 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
357 		    ether_vlanid_t);
358 static void	bridge_rttrim(struct bridge_softc *);
359 static void	bridge_rtage(struct bridge_softc *);
360 static void	bridge_rtflush(struct bridge_softc *, int);
361 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
362 		    ether_vlanid_t);
363 static bool	bridge_vfilter_in(const struct bridge_iflist *, struct mbuf *);
364 static bool	bridge_vfilter_out(const struct bridge_iflist *,
365 		    const struct mbuf *);
366 
367 static void	bridge_rtable_init(struct bridge_softc *);
368 static void	bridge_rtable_fini(struct bridge_softc *);
369 
370 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
371 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
372 		    const uint8_t *, ether_vlanid_t);
373 static int	bridge_rtnode_insert(struct bridge_softc *,
374 		    struct bridge_rtnode *);
375 static void	bridge_rtnode_destroy(struct bridge_softc *,
376 		    struct bridge_rtnode *);
377 static void	bridge_rtable_expire(struct ifnet *, int);
378 static void	bridge_state_change(struct ifnet *, int);
379 
380 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
381 		    const char *name);
382 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
383 		    struct ifnet *ifp);
384 static void	bridge_delete_member(struct bridge_softc *,
385 		    struct bridge_iflist *, int);
386 static void	bridge_delete_span(struct bridge_softc *,
387 		    struct bridge_iflist *);
388 
389 static int	bridge_ioctl_add(struct bridge_softc *, void *);
390 static int	bridge_ioctl_del(struct bridge_softc *, void *);
391 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
392 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
393 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
394 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
395 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
396 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
397 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
398 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
399 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
400 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
401 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
402 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
403 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
404 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
405 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
406 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
407 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
408 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
409 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
410 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
411 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
412 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
413 static int	bridge_ioctl_sifpvid(struct bridge_softc *, void *);
414 static int	bridge_ioctl_sifvlanset(struct bridge_softc *, void *);
415 static int	bridge_ioctl_gifvlanset(struct bridge_softc *, void *);
416 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
417 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
418 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
419 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
420 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
421 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
422 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
423 static int	bridge_ioctl_gflags(struct bridge_softc *, void *);
424 static int	bridge_ioctl_sflags(struct bridge_softc *, void *);
425 static int	bridge_ioctl_gdefpvid(struct bridge_softc *, void *);
426 static int	bridge_ioctl_sdefpvid(struct bridge_softc *, void *);
427 static int	bridge_ioctl_svlanproto(struct bridge_softc *, void *);
428 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
429 		    int);
430 #ifdef INET
431 static int	bridge_ip_checkbasic(struct mbuf **mp);
432 static int	bridge_fragment(struct ifnet *, struct mbuf **mp,
433 		    struct ether_header *, int, struct llc *);
434 #endif /* INET */
435 #ifdef INET6
436 static int	bridge_ip6_checkbasic(struct mbuf **mp);
437 #endif /* INET6 */
438 static void	bridge_linkstate(struct ifnet *ifp);
439 static void	bridge_linkcheck(struct bridge_softc *sc);
440 
441 /*
442  * Use the "null" value from IEEE 802.1Q-2014 Table 9-2
443  * to indicate untagged frames.
444  */
445 #define	VLANTAGOF(_m)	\
446     ((_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : DOT1Q_VID_NULL)
447 
448 static struct bstp_cb_ops bridge_ops = {
449 	.bcb_state = bridge_state_change,
450 	.bcb_rtage = bridge_rtable_expire
451 };
452 
453 SYSCTL_DECL(_net_link);
454 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
455     "Bridge");
456 
457 /* only pass IP[46] packets when pfil is enabled */
458 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
459 #define	V_pfil_onlyip	VNET(pfil_onlyip)
460 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
461     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
462     "Only pass IP packets when pfil is enabled");
463 
464 /* run pfil hooks on the bridge interface */
465 VNET_DEFINE_STATIC(int, pfil_bridge) = 0;
466 #define	V_pfil_bridge	VNET(pfil_bridge)
467 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
468     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
469     "Packet filter on the bridge interface");
470 
471 /* layer2 filter with ipfw */
472 VNET_DEFINE_STATIC(int, pfil_ipfw);
473 #define	V_pfil_ipfw	VNET(pfil_ipfw)
474 
475 /* layer2 ARP filter with ipfw */
476 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
477 #define	V_pfil_ipfw_arp	VNET(pfil_ipfw_arp)
478 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
479     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
480     "Filter ARP packets through IPFW layer2");
481 
482 /* run pfil hooks on the member interface */
483 VNET_DEFINE_STATIC(int, pfil_member) = 0;
484 #define	V_pfil_member	VNET(pfil_member)
485 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
486     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
487     "Packet filter on the member interface");
488 
489 /* run pfil hooks on the physical interface for locally destined packets */
490 VNET_DEFINE_STATIC(int, pfil_local_phys);
491 #define	V_pfil_local_phys	VNET(pfil_local_phys)
492 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
493     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
494     "Packet filter on the physical interface for locally destined packets");
495 
496 /* log STP state changes */
497 VNET_DEFINE_STATIC(int, log_stp);
498 #define	V_log_stp	VNET(log_stp)
499 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
500     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
501     "Log STP state changes");
502 
503 /* share MAC with first bridge member */
504 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
505 #define	V_bridge_inherit_mac	VNET(bridge_inherit_mac)
506 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
507     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
508     "Inherit MAC address from the first bridge member");
509 
510 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
511 #define	V_allow_llz_overlap	VNET(allow_llz_overlap)
512 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
513     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
514     "Allow overlap of link-local scope "
515     "zones of a bridge interface and the member interfaces");
516 
517 /* log MAC address port flapping */
518 VNET_DEFINE_STATIC(bool, log_mac_flap) = true;
519 #define	V_log_mac_flap	VNET(log_mac_flap)
520 SYSCTL_BOOL(_net_link_bridge, OID_AUTO, log_mac_flap,
521     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(log_mac_flap), true,
522     "Log MAC address port flapping");
523 
524 /* allow IP addresses on bridge members */
525 VNET_DEFINE_STATIC(bool, member_ifaddrs) = true;
526 #define	V_member_ifaddrs	VNET(member_ifaddrs)
527 SYSCTL_BOOL(_net_link_bridge, OID_AUTO, member_ifaddrs,
528     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(member_ifaddrs), false,
529     "Allow layer 3 addresses on bridge members (deprecated)");
530 
531 static bool
bridge_member_ifaddrs(void)532 bridge_member_ifaddrs(void)
533 {
534 	return (V_member_ifaddrs);
535 }
536 
537 VNET_DEFINE_STATIC(int, log_interval) = 5;
538 VNET_DEFINE_STATIC(int, log_count) = 0;
539 VNET_DEFINE_STATIC(struct timeval, log_last) = { 0 };
540 
541 #define	V_log_interval	VNET(log_interval)
542 #define	V_log_count	VNET(log_count)
543 #define	V_log_last	VNET(log_last)
544 
545 struct bridge_control {
546 	int	(*bc_func)(struct bridge_softc *, void *);
547 	int	bc_argsize;
548 	int	bc_flags;
549 };
550 
551 #define	BC_F_COPYIN		0x01	/* copy arguments in */
552 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
553 #define	BC_F_SUSER		0x04	/* do super-user check */
554 
555 static const struct bridge_control bridge_control_table[] = {
556 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
557 	  BC_F_COPYIN|BC_F_SUSER },
558 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
559 	  BC_F_COPYIN|BC_F_SUSER },
560 
561 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
562 	  BC_F_COPYIN|BC_F_COPYOUT },
563 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
564 	  BC_F_COPYIN|BC_F_SUSER },
565 
566 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
567 	  BC_F_COPYIN|BC_F_SUSER },
568 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
569 	  BC_F_COPYOUT },
570 
571 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
572 	  BC_F_COPYIN|BC_F_COPYOUT },
573 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
574 	  BC_F_COPYIN|BC_F_COPYOUT },
575 
576 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
577 	  BC_F_COPYIN|BC_F_SUSER },
578 
579 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
580 	  BC_F_COPYIN|BC_F_SUSER },
581 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
582 	  BC_F_COPYOUT },
583 
584 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
585 	  BC_F_COPYIN|BC_F_SUSER },
586 
587 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
588 	  BC_F_COPYIN|BC_F_SUSER },
589 
590 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
591 	  BC_F_COPYOUT },
592 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
593 	  BC_F_COPYIN|BC_F_SUSER },
594 
595 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
596 	  BC_F_COPYOUT },
597 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
598 	  BC_F_COPYIN|BC_F_SUSER },
599 
600 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
601 	  BC_F_COPYOUT },
602 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
603 	  BC_F_COPYIN|BC_F_SUSER },
604 
605 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
606 	  BC_F_COPYOUT },
607 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
608 	  BC_F_COPYIN|BC_F_SUSER },
609 
610 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
611 	  BC_F_COPYIN|BC_F_SUSER },
612 
613 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
614 	  BC_F_COPYIN|BC_F_SUSER },
615 
616 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
617 	  BC_F_COPYIN|BC_F_SUSER },
618 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
619 	  BC_F_COPYIN|BC_F_SUSER },
620 
621 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
622 	  BC_F_COPYOUT },
623 
624 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
625 	  BC_F_COPYOUT },
626 
627 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
628 	  BC_F_COPYIN|BC_F_COPYOUT },
629 
630 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
631 	  BC_F_COPYIN|BC_F_SUSER },
632 
633 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
634 	  BC_F_COPYIN|BC_F_SUSER },
635 
636 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
637 	  BC_F_COPYIN|BC_F_SUSER },
638 
639 	{ bridge_ioctl_sifpvid,		sizeof(struct ifbreq),
640 	  BC_F_COPYIN|BC_F_SUSER },
641 
642 	{ bridge_ioctl_sifvlanset,	sizeof(struct ifbif_vlan_req),
643 	  BC_F_COPYIN|BC_F_SUSER },
644 
645 	{ bridge_ioctl_gifvlanset,	sizeof(struct ifbif_vlan_req),
646 	  BC_F_COPYIN|BC_F_COPYOUT },
647 
648 	{ bridge_ioctl_gflags,		sizeof(struct ifbrparam),
649 	  BC_F_COPYOUT },
650 
651 	{ bridge_ioctl_sflags,		sizeof(struct ifbrparam),
652 	  BC_F_COPYIN|BC_F_SUSER },
653 
654 	{ bridge_ioctl_gdefpvid,	sizeof(struct ifbrparam),
655 	  BC_F_COPYOUT },
656 
657 	{ bridge_ioctl_sdefpvid,	sizeof(struct ifbrparam),
658 	  BC_F_COPYIN|BC_F_SUSER },
659 
660 	{ bridge_ioctl_svlanproto,	sizeof(struct ifbreq),
661 	  BC_F_COPYIN|BC_F_SUSER },
662 };
663 static const int bridge_control_table_size = nitems(bridge_control_table);
664 
665 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list) =
666     LIST_HEAD_INITIALIZER();
667 #define	V_bridge_list	VNET(bridge_list)
668 #define	BRIDGE_LIST_LOCK_INIT(x)	sx_init(&V_bridge_list_sx,	\
669 					    "if_bridge list")
670 #define	BRIDGE_LIST_LOCK_DESTROY(x)	sx_destroy(&V_bridge_list_sx)
671 #define	BRIDGE_LIST_LOCK(x)		sx_xlock(&V_bridge_list_sx)
672 #define	BRIDGE_LIST_UNLOCK(x)		sx_xunlock(&V_bridge_list_sx)
673 
674 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
675 #define	V_bridge_cloner	VNET(bridge_cloner)
676 
677 static const char bridge_name[] = "bridge";
678 
679 static void
vnet_bridge_init(const void * unused __unused)680 vnet_bridge_init(const void *unused __unused)
681 {
682 
683 	V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
684 	    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
685 	    UMA_ALIGN_PTR, 0);
686 	BRIDGE_LIST_LOCK_INIT();
687 
688 	struct if_clone_addreq req = {
689 		.create_f = bridge_clone_create,
690 		.destroy_f = bridge_clone_destroy,
691 		.flags = IFC_F_AUTOUNIT,
692 	};
693 	V_bridge_cloner = ifc_attach_cloner(bridge_name, &req);
694 }
695 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
696     vnet_bridge_init, NULL);
697 
698 static void
vnet_bridge_uninit(const void * unused __unused)699 vnet_bridge_uninit(const void *unused __unused)
700 {
701 
702 	ifc_detach_cloner(V_bridge_cloner);
703 	V_bridge_cloner = NULL;
704 	BRIDGE_LIST_LOCK_DESTROY();
705 
706 	/* Callbacks may use the UMA zone. */
707 	NET_EPOCH_DRAIN_CALLBACKS();
708 
709 	uma_zdestroy(V_bridge_rtnode_zone);
710 }
711 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
712     vnet_bridge_uninit, NULL);
713 
714 static int
bridge_modevent(module_t mod,int type,void * data)715 bridge_modevent(module_t mod, int type, void *data)
716 {
717 
718 	switch (type) {
719 	case MOD_LOAD:
720 		bridge_dn_p = bridge_dummynet;
721 		bridge_same_p = bridge_same;
722 		bridge_get_softc_p = bridge_get_softc;
723 		bridge_member_ifaddrs_p = bridge_member_ifaddrs;
724 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
725 		    ifnet_departure_event, bridge_ifdetach, NULL,
726 		    EVENTHANDLER_PRI_ANY);
727 		break;
728 	case MOD_UNLOAD:
729 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
730 		    bridge_detach_cookie);
731 		bridge_dn_p = NULL;
732 		bridge_same_p = NULL;
733 		bridge_get_softc_p = NULL;
734 		bridge_member_ifaddrs_p = NULL;
735 		break;
736 	default:
737 		return (EOPNOTSUPP);
738 	}
739 	return (0);
740 }
741 
742 static moduledata_t bridge_mod = {
743 	"if_bridge",
744 	bridge_modevent,
745 	0
746 };
747 
748 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
749 MODULE_VERSION(if_bridge, 1);
750 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
751 
752 /*
753  * handler for net.link.bridge.ipfw
754  */
755 static int
sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)756 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
757 {
758 	int enable = V_pfil_ipfw;
759 	int error;
760 
761 	error = sysctl_handle_int(oidp, &enable, 0, req);
762 	enable &= 1;
763 
764 	if (enable != V_pfil_ipfw) {
765 		V_pfil_ipfw = enable;
766 
767 		/*
768 		 * Disable pfil so that ipfw doesnt run twice, if the user
769 		 * really wants both then they can re-enable pfil_bridge and/or
770 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
771 		 * layer2 type.
772 		 */
773 		if (V_pfil_ipfw) {
774 			V_pfil_onlyip = 0;
775 			V_pfil_bridge = 0;
776 			V_pfil_member = 0;
777 		}
778 	}
779 
780 	return (error);
781 }
782 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
783     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT,
784     &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
785     "Layer2 filter with IPFW");
786 
787 #ifdef VIMAGE
788 static void
bridge_reassign(struct ifnet * ifp,struct vnet * newvnet,char * arg)789 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg)
790 {
791 	struct bridge_softc *sc = ifp->if_softc;
792 	struct bridge_iflist *bif;
793 
794 	BRIDGE_LOCK(sc);
795 
796 	while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
797 		bridge_delete_member(sc, bif, 0);
798 
799 	while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
800 		bridge_delete_span(sc, bif);
801 	}
802 
803 	BRIDGE_UNLOCK(sc);
804 
805 	ether_reassign(ifp, newvnet, arg);
806 }
807 #endif
808 
809 /*
810  * bridge_get_softc:
811  *
812  * Return the bridge softc for an ifnet.
813  */
814 static void *
bridge_get_softc(struct ifnet * ifp)815 bridge_get_softc(struct ifnet *ifp)
816 {
817 	struct bridge_iflist *bif;
818 
819 	NET_EPOCH_ASSERT();
820 
821 	bif = ifp->if_bridge;
822 	if (bif == NULL)
823 		return (NULL);
824 	return (bif->bif_sc);
825 }
826 
827 /*
828  * bridge_same:
829  *
830  * Return true if two interfaces are in the same bridge.  This is only used by
831  * bridgestp via bridge_same_p.
832  */
833 static bool
bridge_same(const void * bifap,const void * bifbp)834 bridge_same(const void *bifap, const void *bifbp)
835 {
836 	const struct bridge_iflist *bifa = bifap, *bifb = bifbp;
837 
838 	NET_EPOCH_ASSERT();
839 
840 	if (bifa == NULL || bifb == NULL)
841 		return (false);
842 
843 	return (bifa->bif_sc == bifb->bif_sc);
844 }
845 
846 /*
847  * bridge_clone_create:
848  *
849  *	Create a new bridge instance.
850  */
851 static int
bridge_clone_create(struct if_clone * ifc,char * name,size_t len,struct ifc_data * ifd,struct ifnet ** ifpp)852 bridge_clone_create(struct if_clone *ifc, char *name, size_t len,
853     struct ifc_data *ifd, struct ifnet **ifpp)
854 {
855 	struct bridge_softc *sc;
856 	struct ifnet *ifp;
857 
858 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
859 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
860 
861 	BRIDGE_LOCK_INIT(sc);
862 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
863 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
864 
865 	/* Initialize our routing table. */
866 	bridge_rtable_init(sc);
867 
868 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
869 
870 	CK_LIST_INIT(&sc->sc_iflist);
871 	CK_LIST_INIT(&sc->sc_spanlist);
872 
873 	ifp->if_softc = sc;
874 	if_initname(ifp, bridge_name, ifd->unit);
875 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
876 	ifp->if_capabilities = ifp->if_capenable = IFCAP_VLAN_HWTAGGING;
877 	ifp->if_ioctl = bridge_ioctl;
878 #ifdef ALTQ
879 	ifp->if_start = bridge_altq_start;
880 	ifp->if_transmit = bridge_altq_transmit;
881 	IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
882 	ifp->if_snd.ifq_drv_maxlen = 0;
883 	IFQ_SET_READY(&ifp->if_snd);
884 #else
885 	ifp->if_transmit = bridge_transmit;
886 #endif
887 	ifp->if_qflush = bridge_qflush;
888 	ifp->if_init = bridge_init;
889 	ifp->if_type = IFT_BRIDGE;
890 
891 	ether_gen_addr(ifp, &sc->sc_defaddr);
892 
893 	bstp_attach(&sc->sc_stp, &bridge_ops);
894 	ether_ifattach(ifp, sc->sc_defaddr.octet);
895 	/* Now undo some of the damage... */
896 	ifp->if_baudrate = 0;
897 #ifdef VIMAGE
898 	ifp->if_reassign = bridge_reassign;
899 #endif
900 	sc->sc_if_input = ifp->if_input;	/* ether_input */
901 	ifp->if_input = bridge_inject;
902 
903 	/*
904 	 * Allow BRIDGE_INPUT() to pass in packets originating from the bridge
905 	 * itself via bridge_inject().  This is required for netmap but
906 	 * otherwise has no effect.
907 	 */
908 	ifp->if_bridge_input = bridge_input;
909 
910 	BRIDGE_LIST_LOCK();
911 	LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
912 	BRIDGE_LIST_UNLOCK();
913 	*ifpp = ifp;
914 
915 	return (0);
916 }
917 
918 static void
bridge_clone_destroy_cb(struct epoch_context * ctx)919 bridge_clone_destroy_cb(struct epoch_context *ctx)
920 {
921 	struct bridge_softc *sc;
922 
923 	sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
924 
925 	BRIDGE_LOCK_DESTROY(sc);
926 	free(sc, M_DEVBUF);
927 }
928 
929 /*
930  * bridge_clone_destroy:
931  *
932  *	Destroy a bridge instance.
933  */
934 static int
bridge_clone_destroy(struct if_clone * ifc,struct ifnet * ifp,uint32_t flags)935 bridge_clone_destroy(struct if_clone *ifc, struct ifnet *ifp, uint32_t flags)
936 {
937 	struct bridge_softc *sc = ifp->if_softc;
938 	struct bridge_iflist *bif;
939 	struct epoch_tracker et;
940 
941 	BRIDGE_LOCK(sc);
942 
943 	bridge_stop(ifp, 1);
944 	ifp->if_flags &= ~IFF_UP;
945 
946 	while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
947 		bridge_delete_member(sc, bif, 0);
948 
949 	while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
950 		bridge_delete_span(sc, bif);
951 	}
952 
953 	/* Tear down the routing table. */
954 	bridge_rtable_fini(sc);
955 
956 	BRIDGE_UNLOCK(sc);
957 
958 	NET_EPOCH_ENTER(et);
959 
960 	callout_drain(&sc->sc_brcallout);
961 
962 	BRIDGE_LIST_LOCK();
963 	LIST_REMOVE(sc, sc_list);
964 	BRIDGE_LIST_UNLOCK();
965 
966 	bstp_detach(&sc->sc_stp);
967 #ifdef ALTQ
968 	IFQ_PURGE(&ifp->if_snd);
969 #endif
970 	NET_EPOCH_EXIT(et);
971 
972 	ether_ifdetach(ifp);
973 	if_free(ifp);
974 
975 	NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx);
976 
977 	return (0);
978 }
979 
980 /*
981  * bridge_ioctl:
982  *
983  *	Handle a control request from the operator.
984  */
985 static int
bridge_ioctl(struct ifnet * ifp,u_long cmd,caddr_t data)986 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
987 {
988 	struct bridge_softc *sc = ifp->if_softc;
989 	struct ifreq *ifr = (struct ifreq *)data;
990 	struct bridge_iflist *bif;
991 	struct thread *td = curthread;
992 	union {
993 		struct ifbreq ifbreq;
994 		struct ifbifconf ifbifconf;
995 		struct ifbareq ifbareq;
996 		struct ifbaconf ifbaconf;
997 		struct ifbrparam ifbrparam;
998 		struct ifbropreq ifbropreq;
999 		struct ifbif_vlan_req ifvlanreq;
1000 	} args;
1001 	struct ifdrv *ifd = (struct ifdrv *) data;
1002 	const struct bridge_control *bc;
1003 	int error = 0, oldmtu;
1004 
1005 	BRIDGE_LOCK(sc);
1006 
1007 	switch (cmd) {
1008 	case SIOCADDMULTI:
1009 	case SIOCDELMULTI:
1010 		break;
1011 
1012 	case SIOCGDRVSPEC:
1013 	case SIOCSDRVSPEC:
1014 		if (ifd->ifd_cmd >= bridge_control_table_size) {
1015 			error = EXTERROR(EINVAL, "Invalid control command");
1016 			break;
1017 		}
1018 		bc = &bridge_control_table[ifd->ifd_cmd];
1019 
1020 		if (cmd == SIOCGDRVSPEC &&
1021 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
1022 			error = EXTERROR(EINVAL,
1023 			    "Inappropriate ioctl for command "
1024 			    "(expected SIOCSDRVSPEC)");
1025 			break;
1026 		}
1027 		else if (cmd == SIOCSDRVSPEC &&
1028 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
1029 			error = EXTERROR(EINVAL,
1030 			    "Inappropriate ioctl for command "
1031 			    "(expected SIOCGDRVSPEC)");
1032 			break;
1033 		}
1034 
1035 		if (bc->bc_flags & BC_F_SUSER) {
1036 			error = priv_check(td, PRIV_NET_BRIDGE);
1037 			if (error) {
1038 				EXTERROR(error, "PRIV_NET_BRIDGE required");
1039 				break;
1040 			}
1041 		}
1042 
1043 		if (ifd->ifd_len != bc->bc_argsize ||
1044 		    ifd->ifd_len > sizeof(args)) {
1045 			error = EXTERROR(EINVAL, "Invalid argument size");
1046 			break;
1047 		}
1048 
1049 		bzero(&args, sizeof(args));
1050 		if (bc->bc_flags & BC_F_COPYIN) {
1051 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
1052 			if (error)
1053 				break;
1054 		}
1055 
1056 		oldmtu = ifp->if_mtu;
1057 		error = (*bc->bc_func)(sc, &args);
1058 		if (error)
1059 			break;
1060 
1061 		/*
1062 		 * Bridge MTU may change during addition of the first port.
1063 		 * If it did, do network layer specific procedure.
1064 		 */
1065 		if (ifp->if_mtu != oldmtu)
1066 			if_notifymtu(ifp);
1067 
1068 		if (bc->bc_flags & BC_F_COPYOUT)
1069 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
1070 
1071 		break;
1072 
1073 	case SIOCSIFFLAGS:
1074 		if (!(ifp->if_flags & IFF_UP) &&
1075 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
1076 			/*
1077 			 * If interface is marked down and it is running,
1078 			 * then stop and disable it.
1079 			 */
1080 			bridge_stop(ifp, 1);
1081 		} else if ((ifp->if_flags & IFF_UP) &&
1082 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
1083 			/*
1084 			 * If interface is marked up and it is stopped, then
1085 			 * start it.
1086 			 */
1087 			BRIDGE_UNLOCK(sc);
1088 			(*ifp->if_init)(sc);
1089 			BRIDGE_LOCK(sc);
1090 		}
1091 		break;
1092 
1093 	case SIOCSIFMTU:
1094 		oldmtu = sc->sc_ifp->if_mtu;
1095 
1096 		if (ifr->ifr_mtu < IF_MINMTU) {
1097 			error = EXTERROR(EINVAL,
1098 			    "Requested MTU is lower than IF_MINMTU");
1099 			break;
1100 		}
1101 		if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1102 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
1103 			break;
1104 		}
1105 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1106 			error = (*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
1107 			    SIOCSIFMTU, (caddr_t)ifr);
1108 			if (error != 0) {
1109 				log(LOG_NOTICE, "%s: invalid MTU: %u for"
1110 				    " member %s\n", sc->sc_ifp->if_xname,
1111 				    ifr->ifr_mtu,
1112 				    bif->bif_ifp->if_xname);
1113 				error = EINVAL;
1114 				break;
1115 			}
1116 		}
1117 		if (error) {
1118 			/* Restore the previous MTU on all member interfaces. */
1119 			ifr->ifr_mtu = oldmtu;
1120 			CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1121 				(*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
1122 				    SIOCSIFMTU, (caddr_t)ifr);
1123 			}
1124 			EXTERROR(error,
1125 			    "Failed to set MTU on member interface");
1126 		} else {
1127 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
1128 		}
1129 		break;
1130 	default:
1131 		/*
1132 		 * drop the lock as ether_ioctl() will call bridge_start() and
1133 		 * cause the lock to be recursed.
1134 		 */
1135 		BRIDGE_UNLOCK(sc);
1136 		error = ether_ioctl(ifp, cmd, data);
1137 		BRIDGE_LOCK(sc);
1138 		break;
1139 	}
1140 
1141 	BRIDGE_UNLOCK(sc);
1142 
1143 	return (error);
1144 }
1145 
1146 /*
1147  * bridge_mutecaps:
1148  *
1149  *	Clear or restore unwanted capabilities on the member interface
1150  */
1151 static void
bridge_mutecaps(struct bridge_softc * sc)1152 bridge_mutecaps(struct bridge_softc *sc)
1153 {
1154 	struct bridge_iflist *bif;
1155 	int enabled, mask;
1156 
1157 	BRIDGE_LOCK_ASSERT(sc);
1158 
1159 	/* Initial bitmask of capabilities to test */
1160 	mask = BRIDGE_IFCAPS_MASK;
1161 
1162 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1163 		/* Every member must support it or it's disabled */
1164 		mask &= bif->bif_savedcaps;
1165 	}
1166 
1167 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1168 		enabled = bif->bif_ifp->if_capenable;
1169 		enabled &= ~BRIDGE_IFCAPS_STRIP;
1170 		/* Strip off mask bits and enable them again if allowed */
1171 		enabled &= ~BRIDGE_IFCAPS_MASK;
1172 		enabled |= mask;
1173 		bridge_set_ifcap(sc, bif, enabled);
1174 	}
1175 }
1176 
1177 static void
bridge_set_ifcap(struct bridge_softc * sc,struct bridge_iflist * bif,int set)1178 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1179 {
1180 	struct ifnet *ifp = bif->bif_ifp;
1181 	struct ifreq ifr;
1182 	int error, mask, stuck;
1183 
1184 	bzero(&ifr, sizeof(ifr));
1185 	ifr.ifr_reqcap = set;
1186 
1187 	if (ifp->if_capenable != set) {
1188 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1189 		if (error)
1190 			if_printf(sc->sc_ifp,
1191 			    "error setting capabilities on %s: %d\n",
1192 			    ifp->if_xname, error);
1193 		mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1194 		stuck = ifp->if_capenable & mask & ~set;
1195 		if (stuck != 0)
1196 			if_printf(sc->sc_ifp,
1197 			    "can't disable some capabilities on %s: 0x%x\n",
1198 			    ifp->if_xname, stuck);
1199 	}
1200 }
1201 
1202 /*
1203  * bridge_lookup_member:
1204  *
1205  *	Lookup a bridge member interface.
1206  */
1207 static struct bridge_iflist *
bridge_lookup_member(struct bridge_softc * sc,const char * name)1208 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1209 {
1210 	struct bridge_iflist *bif;
1211 	struct ifnet *ifp;
1212 
1213 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1214 
1215 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1216 		ifp = bif->bif_ifp;
1217 		if (strcmp(ifp->if_xname, name) == 0)
1218 			return (bif);
1219 	}
1220 
1221 	return (NULL);
1222 }
1223 
1224 /*
1225  * bridge_lookup_member_if:
1226  *
1227  *	Lookup a bridge member interface by ifnet*.
1228  */
1229 static struct bridge_iflist *
bridge_lookup_member_if(struct bridge_softc * sc,struct ifnet * member_ifp)1230 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1231 {
1232 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1233 	return (member_ifp->if_bridge);
1234 }
1235 
1236 static void
bridge_delete_member_cb(struct epoch_context * ctx)1237 bridge_delete_member_cb(struct epoch_context *ctx)
1238 {
1239 	struct bridge_iflist *bif;
1240 
1241 	bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1242 
1243 	free(bif, M_DEVBUF);
1244 }
1245 
1246 /*
1247  * bridge_delete_member:
1248  *
1249  *	Delete the specified member interface.
1250  */
1251 static void
bridge_delete_member(struct bridge_softc * sc,struct bridge_iflist * bif,int gone)1252 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1253     int gone)
1254 {
1255 	struct ifnet *ifs = bif->bif_ifp;
1256 	struct ifnet *fif = NULL;
1257 	struct bridge_iflist *bifl;
1258 
1259 	BRIDGE_LOCK_ASSERT(sc);
1260 
1261 	if (bif->bif_flags & IFBIF_STP)
1262 		bstp_disable(&bif->bif_stp);
1263 
1264 	ifs->if_bridge = NULL;
1265 	CK_LIST_REMOVE(bif, bif_next);
1266 
1267 	/*
1268 	 * If removing the interface that gave the bridge its mac address, set
1269 	 * the mac address of the bridge to the address of the next member, or
1270 	 * to its default address if no members are left.
1271 	 */
1272 	if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1273 		if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1274 			bcopy(&sc->sc_defaddr,
1275 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1276 			sc->sc_ifaddr = NULL;
1277 		} else {
1278 			bifl = CK_LIST_FIRST(&sc->sc_iflist);
1279 			fif = bifl->bif_ifp;
1280 			bcopy(IF_LLADDR(fif),
1281 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1282 			sc->sc_ifaddr = fif;
1283 		}
1284 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1285 	}
1286 
1287 	bridge_linkcheck(sc);
1288 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
1289 	BRIDGE_RT_LOCK(sc);
1290 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1291 	BRIDGE_RT_UNLOCK(sc);
1292 	KASSERT(bif->bif_addrcnt == 0,
1293 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1294 
1295 	ifs->if_bridge_output = NULL;
1296 	ifs->if_bridge_input = NULL;
1297 	ifs->if_bridge_linkstate = NULL;
1298 	if (!gone) {
1299 		switch (ifs->if_type) {
1300 		case IFT_ETHER:
1301 		case IFT_L2VLAN:
1302 			/*
1303 			 * Take the interface out of promiscuous mode, but only
1304 			 * if it was promiscuous in the first place. It might
1305 			 * not be if we're in the bridge_ioctl_add() error path.
1306 			 */
1307 			if (ifs->if_flags & IFF_PROMISC)
1308 				(void) ifpromisc(ifs, 0);
1309 			break;
1310 
1311 		case IFT_GIF:
1312 			break;
1313 
1314 		default:
1315 #ifdef DIAGNOSTIC
1316 			panic("bridge_delete_member: impossible");
1317 #endif
1318 			break;
1319 		}
1320 		/* Re-enable any interface capabilities */
1321 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1322 	}
1323 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1324 
1325 	NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1326 }
1327 
1328 /*
1329  * bridge_delete_span:
1330  *
1331  *	Delete the specified span interface.
1332  */
1333 static void
bridge_delete_span(struct bridge_softc * sc,struct bridge_iflist * bif)1334 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1335 {
1336 	BRIDGE_LOCK_ASSERT(sc);
1337 
1338 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1339 	    ("%s: not a span interface", __func__));
1340 
1341 	CK_LIST_REMOVE(bif, bif_next);
1342 
1343 	NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1344 }
1345 
1346 static int
bridge_ioctl_add(struct bridge_softc * sc,void * arg)1347 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1348 {
1349 	struct ifbreq *req = arg;
1350 	struct bridge_iflist *bif = NULL;
1351 	struct ifnet *ifs;
1352 	int error = 0;
1353 
1354 	ifs = ifunit(req->ifbr_ifsname);
1355 	if (ifs == NULL)
1356 		return (EXTERROR(ENOENT, "No such interface",
1357 		    req->ifbr_ifsname));
1358 	if (ifs->if_ioctl == NULL)	/* must be supported */
1359 		return (EXTERROR(EINVAL, "Interface must support ioctl(2)"));
1360 
1361 	/*
1362 	 * If the new interface is a vlan(4), it could be a bridge SVI.
1363 	 * Don't allow such things to be added to bridges.
1364 	 */
1365 	if (ifs->if_type == IFT_L2VLAN) {
1366 		struct ifnet *parent;
1367 		struct epoch_tracker et;
1368 		bool is_bridge;
1369 
1370 		/*
1371 		 * Entering NET_EPOCH with BRIDGE_LOCK held, but this is okay
1372 		 * since we don't sleep here.
1373 		 */
1374 		NET_EPOCH_ENTER(et);
1375 		parent = VLAN_TRUNKDEV(ifs);
1376 		is_bridge = (parent != NULL && parent->if_type == IFT_BRIDGE);
1377 		NET_EPOCH_EXIT(et);
1378 
1379 		if (is_bridge)
1380 			return (EXTERROR(EINVAL,
1381 			    "Bridge SVI cannot be added to a bridge"));
1382 	}
1383 
1384 	/* If it's in the span list, it can't be a member. */
1385 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1386 		if (ifs == bif->bif_ifp)
1387 			return (EXTERROR(EBUSY,
1388 			    "Span interface cannot be a member"));
1389 
1390 	if (ifs->if_bridge) {
1391 		struct bridge_iflist *sbif = ifs->if_bridge;
1392 		if (sbif->bif_sc == sc)
1393 			return (EXTERROR(EEXIST,
1394 			    "Interface is already a member of this bridge"));
1395 
1396 		return (EXTERROR(EBUSY,
1397 		    "Interface is already a member of another bridge"));
1398 	}
1399 
1400 	switch (ifs->if_type) {
1401 	case IFT_ETHER:
1402 	case IFT_L2VLAN:
1403 	case IFT_GIF:
1404 		/* permitted interface types */
1405 		break;
1406 	default:
1407 		return (EXTERROR(EINVAL, "Unsupported interface type"));
1408 	}
1409 
1410 #ifdef INET6
1411 	/*
1412 	 * Two valid inet6 addresses with link-local scope must not be
1413 	 * on the parent interface and the member interfaces at the
1414 	 * same time.  This restriction is needed to prevent violation
1415 	 * of link-local scope zone.  Attempts to add a member
1416 	 * interface which has inet6 addresses when the parent has
1417 	 * inet6 triggers removal of all inet6 addresses on the member
1418 	 * interface.
1419 	 */
1420 
1421 	/* Check if the parent interface has a link-local scope addr. */
1422 	if (V_allow_llz_overlap == 0 &&
1423 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1424 		/*
1425 		 * If any, remove all inet6 addresses from the member
1426 		 * interfaces.
1427 		 */
1428 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1429  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1430 				in6_ifdetach(bif->bif_ifp);
1431 				if_printf(sc->sc_ifp,
1432 				    "IPv6 addresses on %s have been removed "
1433 				    "before adding it as a member to prevent "
1434 				    "IPv6 address scope violation.\n",
1435 				    bif->bif_ifp->if_xname);
1436 			}
1437 		}
1438 		if (in6ifa_llaonifp(ifs)) {
1439 			in6_ifdetach(ifs);
1440 			if_printf(sc->sc_ifp,
1441 			    "IPv6 addresses on %s have been removed "
1442 			    "before adding it as a member to prevent "
1443 			    "IPv6 address scope violation.\n",
1444 			    ifs->if_xname);
1445 		}
1446 	}
1447 #endif
1448 
1449 	/*
1450 	 * If member_ifaddrs is disabled, do not allow an interface with
1451 	 * assigned IP addresses to be added to a bridge.  Skip this check
1452 	 * for gif interfaces, because the IP address assigned to a gif
1453 	 * interface is separate from the bridge's Ethernet segment.
1454 	 */
1455 	if (ifs->if_type != IFT_GIF) {
1456 		struct ifaddr *ifa;
1457 
1458 		CK_STAILQ_FOREACH(ifa, &ifs->if_addrhead, ifa_link) {
1459 			if (ifa->ifa_addr->sa_family != AF_INET &&
1460 			    ifa->ifa_addr->sa_family != AF_INET6)
1461 				continue;
1462 
1463 			if (V_member_ifaddrs) {
1464 				if_printf(sc->sc_ifp,
1465 				    "WARNING: Adding member interface %s which "
1466 				    "has an IP address assigned is deprecated "
1467 				    "and will be unsupported in a future "
1468 				    "release.\n", ifs->if_xname);
1469 				break;
1470 			} else {
1471 				return (EXTERROR(EINVAL,
1472 				    "Member interface may not have "
1473 				    "an IP address assigned"));
1474 			}
1475 		}
1476 	}
1477 
1478 	/* Allow the first Ethernet member to define the MTU */
1479 	if (CK_LIST_EMPTY(&sc->sc_iflist))
1480 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1481 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1482 		struct ifreq ifr;
1483 
1484 		snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s",
1485 		    ifs->if_xname);
1486 		ifr.ifr_mtu = sc->sc_ifp->if_mtu;
1487 
1488 		error = (*ifs->if_ioctl)(ifs,
1489 		    SIOCSIFMTU, (caddr_t)&ifr);
1490 		if (error != 0) {
1491 			log(LOG_NOTICE, "%s: invalid MTU: %u for"
1492 			    " new member %s\n", sc->sc_ifp->if_xname,
1493 			    ifr.ifr_mtu,
1494 			    ifs->if_xname);
1495 			return (EXTERROR(EINVAL,
1496 			    "Failed to set MTU on new member"));
1497 		}
1498 	}
1499 
1500 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1501 	if (bif == NULL)
1502 		return (ENOMEM);
1503 
1504 	bif->bif_sc = sc;
1505 	bif->bif_ifp = ifs;
1506 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1507 	bif->bif_savedcaps = ifs->if_capenable;
1508 	bif->bif_vlanproto = ETHERTYPE_VLAN;
1509 	bif->bif_pvid = sc->sc_defpvid;
1510 	if (sc->sc_flags & IFBRF_DEFQINQ)
1511 		bif->bif_flags |= IFBIF_QINQ;
1512 
1513 	/*
1514 	 * Assign the interface's MAC address to the bridge if it's the first
1515 	 * member and the MAC address of the bridge has not been changed from
1516 	 * the default randomly generated one.
1517 	 */
1518 	if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1519 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1520 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1521 		sc->sc_ifaddr = ifs;
1522 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1523 	}
1524 
1525 	ifs->if_bridge = bif;
1526 	ifs->if_bridge_output = bridge_output;
1527 	ifs->if_bridge_input = bridge_input;
1528 	ifs->if_bridge_linkstate = bridge_linkstate;
1529 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1530 	/*
1531 	 * XXX: XLOCK HERE!?!
1532 	 *
1533 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1534 	 */
1535 	CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1536 
1537 	/* Set interface capabilities to the intersection set of all members */
1538 	bridge_mutecaps(sc);
1539 	bridge_linkcheck(sc);
1540 
1541 	/* Place the interface into promiscuous mode */
1542 	switch (ifs->if_type) {
1543 		case IFT_ETHER:
1544 		case IFT_L2VLAN:
1545 			error = ifpromisc(ifs, 1);
1546 			break;
1547 	}
1548 
1549 	if (error)
1550 		bridge_delete_member(sc, bif, 0);
1551 	return (error);
1552 }
1553 
1554 static int
bridge_ioctl_del(struct bridge_softc * sc,void * arg)1555 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1556 {
1557 	struct ifbreq *req = arg;
1558 	struct bridge_iflist *bif;
1559 
1560 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1561 	if (bif == NULL)
1562 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1563 
1564 	bridge_delete_member(sc, bif, 0);
1565 
1566 	return (0);
1567 }
1568 
1569 static int
bridge_ioctl_gifflags(struct bridge_softc * sc,void * arg)1570 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1571 {
1572 	struct ifbreq *req = arg;
1573 	struct bridge_iflist *bif;
1574 	struct bstp_port *bp;
1575 
1576 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1577 	if (bif == NULL)
1578 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1579 
1580 	bp = &bif->bif_stp;
1581 	req->ifbr_ifsflags = bif->bif_flags;
1582 	req->ifbr_state = bp->bp_state;
1583 	req->ifbr_priority = bp->bp_priority;
1584 	req->ifbr_path_cost = bp->bp_path_cost;
1585 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1586 	req->ifbr_proto = bp->bp_protover;
1587 	req->ifbr_role = bp->bp_role;
1588 	req->ifbr_stpflags = bp->bp_flags;
1589 	req->ifbr_addrcnt = bif->bif_addrcnt;
1590 	req->ifbr_addrmax = bif->bif_addrmax;
1591 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1592 	req->ifbr_pvid = bif->bif_pvid;
1593 	req->ifbr_vlanproto = bif->bif_vlanproto;
1594 
1595 	/* Copy STP state options as flags */
1596 	if (bp->bp_operedge)
1597 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1598 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1599 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1600 	if (bp->bp_ptp_link)
1601 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1602 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1603 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1604 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1605 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1606 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1607 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1608 	return (0);
1609 }
1610 
1611 static int
bridge_ioctl_sifflags(struct bridge_softc * sc,void * arg)1612 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1613 {
1614 	struct epoch_tracker et;
1615 	struct ifbreq *req = arg;
1616 	struct bridge_iflist *bif;
1617 	struct bstp_port *bp;
1618 	int error;
1619 
1620 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1621 	if (bif == NULL)
1622 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1623 	bp = &bif->bif_stp;
1624 
1625 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1626 		/* SPAN is readonly */
1627 		return (EXTERROR(EINVAL, "Span interface cannot be modified"));
1628 
1629 	NET_EPOCH_ENTER(et);
1630 
1631 	if (req->ifbr_ifsflags & IFBIF_STP) {
1632 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1633 			error = bstp_enable(&bif->bif_stp);
1634 			if (error) {
1635 				NET_EPOCH_EXIT(et);
1636 				return (EXTERROR(error,
1637 				    "Failed to enable STP"));
1638 			}
1639 		}
1640 	} else {
1641 		if ((bif->bif_flags & IFBIF_STP) != 0)
1642 			bstp_disable(&bif->bif_stp);
1643 	}
1644 
1645 	/* Pass on STP flags */
1646 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1647 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1648 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1649 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1650 
1651 	/* Save the bits relating to the bridge */
1652 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1653 
1654 	NET_EPOCH_EXIT(et);
1655 
1656 	return (0);
1657 }
1658 
1659 static int
bridge_ioctl_scache(struct bridge_softc * sc,void * arg)1660 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1661 {
1662 	struct ifbrparam *param = arg;
1663 
1664 	sc->sc_brtmax = param->ifbrp_csize;
1665 	bridge_rttrim(sc);
1666 
1667 	return (0);
1668 }
1669 
1670 static int
bridge_ioctl_gcache(struct bridge_softc * sc,void * arg)1671 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1672 {
1673 	struct ifbrparam *param = arg;
1674 
1675 	param->ifbrp_csize = sc->sc_brtmax;
1676 
1677 	return (0);
1678 }
1679 
1680 static int
bridge_ioctl_gifs(struct bridge_softc * sc,void * arg)1681 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1682 {
1683 	struct ifbifconf *bifc = arg;
1684 	struct bridge_iflist *bif;
1685 	struct ifbreq breq;
1686 	char *buf, *outbuf;
1687 	int count, buflen, len, error = 0;
1688 
1689 	count = 0;
1690 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1691 		count++;
1692 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1693 		count++;
1694 
1695 	buflen = sizeof(breq) * count;
1696 	if (bifc->ifbic_len == 0) {
1697 		bifc->ifbic_len = buflen;
1698 		return (0);
1699 	}
1700 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1701 	if (outbuf == NULL)
1702 		return (ENOMEM);
1703 
1704 	count = 0;
1705 	buf = outbuf;
1706 	len = min(bifc->ifbic_len, buflen);
1707 	bzero(&breq, sizeof(breq));
1708 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1709 		if (len < sizeof(breq))
1710 			break;
1711 
1712 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1713 		    sizeof(breq.ifbr_ifsname));
1714 		/* Fill in the ifbreq structure */
1715 		error = bridge_ioctl_gifflags(sc, &breq);
1716 		if (error)
1717 			break;
1718 		memcpy(buf, &breq, sizeof(breq));
1719 		count++;
1720 		buf += sizeof(breq);
1721 		len -= sizeof(breq);
1722 	}
1723 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1724 		if (len < sizeof(breq))
1725 			break;
1726 
1727 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1728 		    sizeof(breq.ifbr_ifsname));
1729 		breq.ifbr_ifsflags = bif->bif_flags;
1730 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1731 		memcpy(buf, &breq, sizeof(breq));
1732 		count++;
1733 		buf += sizeof(breq);
1734 		len -= sizeof(breq);
1735 	}
1736 
1737 	bifc->ifbic_len = sizeof(breq) * count;
1738 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1739 	free(outbuf, M_TEMP);
1740 	return (error);
1741 }
1742 
1743 static int
bridge_ioctl_rts(struct bridge_softc * sc,void * arg)1744 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1745 {
1746 	struct ifbaconf *bac = arg;
1747 	struct bridge_rtnode *brt;
1748 	struct ifbareq bareq;
1749 	char *buf, *outbuf;
1750 	int count, buflen, len, error = 0;
1751 
1752 	if (bac->ifbac_len == 0)
1753 		return (0);
1754 
1755 	count = 0;
1756 	CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1757 		count++;
1758 	buflen = sizeof(bareq) * count;
1759 
1760 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1761 	if (outbuf == NULL)
1762 		return (ENOMEM);
1763 
1764 	count = 0;
1765 	buf = outbuf;
1766 	len = min(bac->ifbac_len, buflen);
1767 	bzero(&bareq, sizeof(bareq));
1768 	CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1769 		if (len < sizeof(bareq))
1770 			goto out;
1771 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1772 		    sizeof(bareq.ifba_ifsname));
1773 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1774 		bareq.ifba_vlan = brt->brt_vlan;
1775 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1776 				time_uptime < brt->brt_expire)
1777 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1778 		else
1779 			bareq.ifba_expire = 0;
1780 		bareq.ifba_flags = brt->brt_flags;
1781 
1782 		memcpy(buf, &bareq, sizeof(bareq));
1783 		count++;
1784 		buf += sizeof(bareq);
1785 		len -= sizeof(bareq);
1786 	}
1787 out:
1788 	bac->ifbac_len = sizeof(bareq) * count;
1789 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1790 	free(outbuf, M_TEMP);
1791 	return (error);
1792 }
1793 
1794 static int
bridge_ioctl_saddr(struct bridge_softc * sc,void * arg)1795 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1796 {
1797 	struct ifbareq *req = arg;
1798 	struct bridge_iflist *bif;
1799 	struct epoch_tracker et;
1800 	int error;
1801 
1802 	NET_EPOCH_ENTER(et);
1803 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1804 	if (bif == NULL) {
1805 		NET_EPOCH_EXIT(et);
1806 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1807 	}
1808 
1809 	/* bridge_rtupdate() may acquire the lock. */
1810 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1811 	    req->ifba_flags);
1812 	NET_EPOCH_EXIT(et);
1813 
1814 	return (error);
1815 }
1816 
1817 static int
bridge_ioctl_sto(struct bridge_softc * sc,void * arg)1818 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1819 {
1820 	struct ifbrparam *param = arg;
1821 
1822 	sc->sc_brttimeout = param->ifbrp_ctime;
1823 	return (0);
1824 }
1825 
1826 static int
bridge_ioctl_gto(struct bridge_softc * sc,void * arg)1827 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1828 {
1829 	struct ifbrparam *param = arg;
1830 
1831 	param->ifbrp_ctime = sc->sc_brttimeout;
1832 	return (0);
1833 }
1834 
1835 static int
bridge_ioctl_daddr(struct bridge_softc * sc,void * arg)1836 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1837 {
1838 	struct ifbareq *req = arg;
1839 	int vlan = req->ifba_vlan;
1840 
1841 	/* Userspace uses '0' to mean 'any vlan' */
1842 	if (vlan == 0)
1843 		vlan = DOT1Q_VID_RSVD_IMPL;
1844 
1845 	return (bridge_rtdaddr(sc, req->ifba_dst, vlan));
1846 }
1847 
1848 static int
bridge_ioctl_flush(struct bridge_softc * sc,void * arg)1849 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1850 {
1851 	struct ifbreq *req = arg;
1852 
1853 	BRIDGE_RT_LOCK(sc);
1854 	bridge_rtflush(sc, req->ifbr_ifsflags);
1855 	BRIDGE_RT_UNLOCK(sc);
1856 
1857 	return (0);
1858 }
1859 
1860 static int
bridge_ioctl_gpri(struct bridge_softc * sc,void * arg)1861 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1862 {
1863 	struct ifbrparam *param = arg;
1864 	struct bstp_state *bs = &sc->sc_stp;
1865 
1866 	param->ifbrp_prio = bs->bs_bridge_priority;
1867 	return (0);
1868 }
1869 
1870 static int
bridge_ioctl_spri(struct bridge_softc * sc,void * arg)1871 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1872 {
1873 	struct ifbrparam *param = arg;
1874 
1875 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1876 }
1877 
1878 static int
bridge_ioctl_ght(struct bridge_softc * sc,void * arg)1879 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1880 {
1881 	struct ifbrparam *param = arg;
1882 	struct bstp_state *bs = &sc->sc_stp;
1883 
1884 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1885 	return (0);
1886 }
1887 
1888 static int
bridge_ioctl_sht(struct bridge_softc * sc,void * arg)1889 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1890 {
1891 	struct ifbrparam *param = arg;
1892 
1893 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1894 }
1895 
1896 static int
bridge_ioctl_gfd(struct bridge_softc * sc,void * arg)1897 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1898 {
1899 	struct ifbrparam *param = arg;
1900 	struct bstp_state *bs = &sc->sc_stp;
1901 
1902 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1903 	return (0);
1904 }
1905 
1906 static int
bridge_ioctl_sfd(struct bridge_softc * sc,void * arg)1907 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1908 {
1909 	struct ifbrparam *param = arg;
1910 
1911 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1912 }
1913 
1914 static int
bridge_ioctl_gma(struct bridge_softc * sc,void * arg)1915 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1916 {
1917 	struct ifbrparam *param = arg;
1918 	struct bstp_state *bs = &sc->sc_stp;
1919 
1920 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1921 	return (0);
1922 }
1923 
1924 static int
bridge_ioctl_sma(struct bridge_softc * sc,void * arg)1925 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1926 {
1927 	struct ifbrparam *param = arg;
1928 
1929 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1930 }
1931 
1932 static int
bridge_ioctl_sifprio(struct bridge_softc * sc,void * arg)1933 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1934 {
1935 	struct ifbreq *req = arg;
1936 	struct bridge_iflist *bif;
1937 
1938 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1939 	if (bif == NULL)
1940 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1941 
1942 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1943 }
1944 
1945 static int
bridge_ioctl_sifcost(struct bridge_softc * sc,void * arg)1946 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1947 {
1948 	struct ifbreq *req = arg;
1949 	struct bridge_iflist *bif;
1950 
1951 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1952 	if (bif == NULL)
1953 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1954 
1955 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1956 }
1957 
1958 static int
bridge_ioctl_sifmaxaddr(struct bridge_softc * sc,void * arg)1959 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1960 {
1961 	struct ifbreq *req = arg;
1962 	struct bridge_iflist *bif;
1963 
1964 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1965 	if (bif == NULL)
1966 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1967 
1968 	bif->bif_addrmax = req->ifbr_addrmax;
1969 	return (0);
1970 }
1971 
1972 static int
bridge_ioctl_sifpvid(struct bridge_softc * sc,void * arg)1973 bridge_ioctl_sifpvid(struct bridge_softc *sc, void *arg)
1974 {
1975 	struct ifbreq *req = arg;
1976 	struct bridge_iflist *bif;
1977 
1978 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1979 	if (bif == NULL)
1980 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1981 
1982 	if (req->ifbr_pvid > DOT1Q_VID_MAX)
1983 		return (EXTERROR(EINVAL, "Invalid VLAN ID"));
1984 
1985 	bif->bif_pvid = req->ifbr_pvid;
1986 	return (0);
1987 }
1988 
1989 static int
bridge_ioctl_sifvlanset(struct bridge_softc * sc,void * arg)1990 bridge_ioctl_sifvlanset(struct bridge_softc *sc, void *arg)
1991 {
1992 	struct ifbif_vlan_req *req = arg;
1993 	struct bridge_iflist *bif;
1994 
1995 	if ((sc->sc_flags & IFBRF_VLANFILTER) == 0)
1996 		return (EXTERROR(EINVAL, "VLAN filtering not enabled"));
1997 
1998 	bif = bridge_lookup_member(sc, req->bv_ifname);
1999 	if (bif == NULL)
2000 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
2001 
2002 	/* Reject invalid VIDs. */
2003 	if (BRVLAN_TEST(&req->bv_set, DOT1Q_VID_NULL) ||
2004 	    BRVLAN_TEST(&req->bv_set, DOT1Q_VID_RSVD_IMPL))
2005 		return (EXTERROR(EINVAL, "Invalid VLAN ID in set"));
2006 
2007 	switch (req->bv_op) {
2008 		/* Replace the existing vlan set with the new set */
2009 	case BRDG_VLAN_OP_SET:
2010 		BIT_COPY(BRVLAN_SETSIZE, &req->bv_set, &bif->bif_vlan_set);
2011 		break;
2012 
2013 		/* Modify the existing vlan set to add the given vlans */
2014 	case BRDG_VLAN_OP_ADD:
2015 		BIT_OR(BRVLAN_SETSIZE, &bif->bif_vlan_set, &req->bv_set);
2016 		break;
2017 
2018 		/* Modify the existing vlan set to remove the given vlans */
2019 	case BRDG_VLAN_OP_DEL:
2020 		BIT_ANDNOT(BRVLAN_SETSIZE, &bif->bif_vlan_set, &req->bv_set);
2021 		break;
2022 
2023 		/* Invalid or unknown operation */
2024 	default:
2025 		return (EXTERROR(EINVAL,
2026 		    "Unsupported BRDGSIFVLANSET operation"));
2027 	}
2028 
2029 	return (0);
2030 }
2031 
2032 static int
bridge_ioctl_gifvlanset(struct bridge_softc * sc,void * arg)2033 bridge_ioctl_gifvlanset(struct bridge_softc *sc, void *arg)
2034 {
2035 	struct ifbif_vlan_req *req = arg;
2036 	struct bridge_iflist *bif;
2037 
2038 	bif = bridge_lookup_member(sc, req->bv_ifname);
2039 	if (bif == NULL)
2040 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
2041 
2042 	BIT_COPY(BRVLAN_SETSIZE, &bif->bif_vlan_set, &req->bv_set);
2043 	return (0);
2044 }
2045 
2046 static int
bridge_ioctl_addspan(struct bridge_softc * sc,void * arg)2047 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
2048 {
2049 	struct ifbreq *req = arg;
2050 	struct bridge_iflist *bif = NULL;
2051 	struct ifnet *ifs;
2052 
2053 	ifs = ifunit(req->ifbr_ifsname);
2054 	if (ifs == NULL)
2055 		return (EXTERROR(ENOENT, "No such interface"));
2056 
2057 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
2058 		if (ifs == bif->bif_ifp)
2059 			return (EXTERROR(EBUSY,
2060 			    "Interface is already a span port"));
2061 
2062 	if (ifs->if_bridge != NULL)
2063 		return (EXTERROR(EEXIST,
2064 		    "Interface is already a bridge member"));
2065 
2066 	switch (ifs->if_type) {
2067 		case IFT_ETHER:
2068 		case IFT_GIF:
2069 		case IFT_L2VLAN:
2070 			break;
2071 		default:
2072 			return (EXTERROR(EINVAL, "Unsupported interface type"));
2073 	}
2074 
2075 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
2076 	if (bif == NULL)
2077 		return (ENOMEM);
2078 
2079 	bif->bif_ifp = ifs;
2080 	bif->bif_flags = IFBIF_SPAN;
2081 
2082 	CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
2083 
2084 	return (0);
2085 }
2086 
2087 static int
bridge_ioctl_delspan(struct bridge_softc * sc,void * arg)2088 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
2089 {
2090 	struct ifbreq *req = arg;
2091 	struct bridge_iflist *bif;
2092 	struct ifnet *ifs;
2093 
2094 	ifs = ifunit(req->ifbr_ifsname);
2095 	if (ifs == NULL)
2096 		return (EXTERROR(ENOENT, "No such interface"));
2097 
2098 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
2099 		if (ifs == bif->bif_ifp)
2100 			break;
2101 
2102 	if (bif == NULL)
2103 		return (EXTERROR(ENOENT, "Interface is not a span port"));
2104 
2105 	bridge_delete_span(sc, bif);
2106 
2107 	return (0);
2108 }
2109 
2110 static int
bridge_ioctl_gbparam(struct bridge_softc * sc,void * arg)2111 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
2112 {
2113 	struct ifbropreq *req = arg;
2114 	struct bstp_state *bs = &sc->sc_stp;
2115 	struct bstp_port *root_port;
2116 
2117 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
2118 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
2119 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
2120 
2121 	root_port = bs->bs_root_port;
2122 	if (root_port == NULL)
2123 		req->ifbop_root_port = 0;
2124 	else
2125 		req->ifbop_root_port = root_port->bp_ifp->if_index;
2126 
2127 	req->ifbop_holdcount = bs->bs_txholdcount;
2128 	req->ifbop_priority = bs->bs_bridge_priority;
2129 	req->ifbop_protocol = bs->bs_protover;
2130 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
2131 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
2132 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
2133 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
2134 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
2135 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
2136 
2137 	return (0);
2138 }
2139 
2140 static int
bridge_ioctl_grte(struct bridge_softc * sc,void * arg)2141 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
2142 {
2143 	struct ifbrparam *param = arg;
2144 
2145 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
2146 	return (0);
2147 }
2148 
2149 static int
bridge_ioctl_gifsstp(struct bridge_softc * sc,void * arg)2150 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
2151 {
2152 	struct ifbpstpconf *bifstp = arg;
2153 	struct bridge_iflist *bif;
2154 	struct bstp_port *bp;
2155 	struct ifbpstpreq bpreq;
2156 	char *buf, *outbuf;
2157 	int count, buflen, len, error = 0;
2158 
2159 	count = 0;
2160 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2161 		if ((bif->bif_flags & IFBIF_STP) != 0)
2162 			count++;
2163 	}
2164 
2165 	buflen = sizeof(bpreq) * count;
2166 	if (bifstp->ifbpstp_len == 0) {
2167 		bifstp->ifbpstp_len = buflen;
2168 		return (0);
2169 	}
2170 
2171 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
2172 	if (outbuf == NULL)
2173 		return (ENOMEM);
2174 
2175 	count = 0;
2176 	buf = outbuf;
2177 	len = min(bifstp->ifbpstp_len, buflen);
2178 	bzero(&bpreq, sizeof(bpreq));
2179 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2180 		if (len < sizeof(bpreq))
2181 			break;
2182 
2183 		if ((bif->bif_flags & IFBIF_STP) == 0)
2184 			continue;
2185 
2186 		bp = &bif->bif_stp;
2187 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
2188 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
2189 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
2190 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
2191 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
2192 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
2193 
2194 		memcpy(buf, &bpreq, sizeof(bpreq));
2195 		count++;
2196 		buf += sizeof(bpreq);
2197 		len -= sizeof(bpreq);
2198 	}
2199 
2200 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
2201 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
2202 	free(outbuf, M_TEMP);
2203 	return (error);
2204 }
2205 
2206 static int
bridge_ioctl_sproto(struct bridge_softc * sc,void * arg)2207 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
2208 {
2209 	struct ifbrparam *param = arg;
2210 
2211 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
2212 }
2213 
2214 static int
bridge_ioctl_stxhc(struct bridge_softc * sc,void * arg)2215 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
2216 {
2217 	struct ifbrparam *param = arg;
2218 
2219 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
2220 }
2221 
2222 static int
bridge_ioctl_gflags(struct bridge_softc * sc,void * arg)2223 bridge_ioctl_gflags(struct bridge_softc *sc, void *arg)
2224 {
2225 	struct ifbrparam *param = arg;
2226 
2227 	param->ifbrp_flags = sc->sc_flags;
2228 
2229 	return (0);
2230 }
2231 
2232 static int
bridge_ioctl_sflags(struct bridge_softc * sc,void * arg)2233 bridge_ioctl_sflags(struct bridge_softc *sc, void *arg)
2234 {
2235 	struct ifbrparam *param = arg;
2236 
2237 	sc->sc_flags = param->ifbrp_flags;
2238 
2239 	return (0);
2240 }
2241 
2242 static int
bridge_ioctl_gdefpvid(struct bridge_softc * sc,void * arg)2243 bridge_ioctl_gdefpvid(struct bridge_softc *sc, void *arg)
2244 {
2245 	struct ifbrparam *param = arg;
2246 
2247 	param->ifbrp_defpvid = sc->sc_defpvid;
2248 
2249 	return (0);
2250 }
2251 
2252 static int
bridge_ioctl_sdefpvid(struct bridge_softc * sc,void * arg)2253 bridge_ioctl_sdefpvid(struct bridge_softc *sc, void *arg)
2254 {
2255 	struct ifbrparam *param = arg;
2256 
2257 	/* Reject invalid VIDs, but allow 0 to mean 'none'. */
2258 	if (param->ifbrp_defpvid > DOT1Q_VID_MAX)
2259 		return (EINVAL);
2260 
2261 	sc->sc_defpvid = param->ifbrp_defpvid;
2262 
2263 	return (0);
2264 }
2265 
2266 static int
bridge_ioctl_svlanproto(struct bridge_softc * sc,void * arg)2267 bridge_ioctl_svlanproto(struct bridge_softc *sc, void *arg)
2268 {
2269 	struct ifbreq *req = arg;
2270 	struct bridge_iflist *bif;
2271 
2272 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
2273 	if (bif == NULL)
2274 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
2275 
2276 	if (req->ifbr_vlanproto != ETHERTYPE_VLAN &&
2277 	    req->ifbr_vlanproto != ETHERTYPE_QINQ)
2278 		return (EXTERROR(EINVAL, "Invalid VLAN protocol"));
2279 
2280 	bif->bif_vlanproto = req->ifbr_vlanproto;
2281 
2282 	return (0);
2283 }
2284 /*
2285  * bridge_ifdetach:
2286  *
2287  *	Detach an interface from a bridge.  Called when a member
2288  *	interface is detaching.
2289  */
2290 static void
bridge_ifdetach(void * arg __unused,struct ifnet * ifp)2291 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
2292 {
2293 	struct bridge_iflist *bif = ifp->if_bridge;
2294 	struct bridge_softc *sc = NULL;
2295 
2296 	if (bif)
2297 		sc = bif->bif_sc;
2298 
2299 	if (ifp->if_flags & IFF_RENAMING)
2300 		return;
2301 	if (V_bridge_cloner == NULL) {
2302 		/*
2303 		 * This detach handler can be called after
2304 		 * vnet_bridge_uninit().  Just return in that case.
2305 		 */
2306 		return;
2307 	}
2308 	/* Check if the interface is a bridge member */
2309 	if (sc != NULL) {
2310 		BRIDGE_LOCK(sc);
2311 		bridge_delete_member(sc, bif, 1);
2312 		BRIDGE_UNLOCK(sc);
2313 		return;
2314 	}
2315 
2316 	/* Check if the interface is a span port */
2317 	BRIDGE_LIST_LOCK();
2318 	LIST_FOREACH(sc, &V_bridge_list, sc_list) {
2319 		BRIDGE_LOCK(sc);
2320 		CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
2321 			if (ifp == bif->bif_ifp) {
2322 				bridge_delete_span(sc, bif);
2323 				break;
2324 			}
2325 
2326 		BRIDGE_UNLOCK(sc);
2327 	}
2328 	BRIDGE_LIST_UNLOCK();
2329 }
2330 
2331 /*
2332  * bridge_init:
2333  *
2334  *	Initialize a bridge interface.
2335  */
2336 static void
bridge_init(void * xsc)2337 bridge_init(void *xsc)
2338 {
2339 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
2340 	struct ifnet *ifp = sc->sc_ifp;
2341 
2342 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
2343 		return;
2344 
2345 	BRIDGE_LOCK(sc);
2346 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
2347 	    bridge_timer, sc);
2348 
2349 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
2350 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
2351 
2352 	BRIDGE_UNLOCK(sc);
2353 }
2354 
2355 /*
2356  * bridge_stop:
2357  *
2358  *	Stop the bridge interface.
2359  */
2360 static void
bridge_stop(struct ifnet * ifp,int disable)2361 bridge_stop(struct ifnet *ifp, int disable)
2362 {
2363 	struct bridge_softc *sc = ifp->if_softc;
2364 
2365 	BRIDGE_LOCK_ASSERT(sc);
2366 
2367 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2368 		return;
2369 
2370 	BRIDGE_RT_LOCK(sc);
2371 	callout_stop(&sc->sc_brcallout);
2372 
2373 	bstp_stop(&sc->sc_stp);
2374 
2375 	bridge_rtflush(sc, IFBF_FLUSHDYN);
2376 	BRIDGE_RT_UNLOCK(sc);
2377 
2378 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
2379 }
2380 
2381 /*
2382  * bridge_enqueue:
2383  *
2384  *	Enqueue a packet on a bridge member interface.
2385  *
2386  */
2387 static int
bridge_enqueue(struct bridge_softc * sc,struct ifnet * dst_ifp,struct mbuf * m,struct bridge_iflist * bif)2388 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m,
2389     struct bridge_iflist *bif)
2390 {
2391 	int len, err = 0;
2392 	short mflags;
2393 	struct mbuf *m0;
2394 
2395 	/*
2396 	 * Find the bridge member port this packet is being sent on, if the
2397 	 * caller didn't already provide it.
2398 	 */
2399 	if (bif == NULL)
2400 		bif = bridge_lookup_member_if(sc, dst_ifp);
2401 	if (bif == NULL) {
2402 		/* Perhaps the interface was removed from the bridge */
2403 		m_freem(m);
2404 		return (EINVAL);
2405 	}
2406 
2407 	/* Do VLAN filtering. */
2408 	if (!bridge_vfilter_out(bif, m)) {
2409 		m_freem(m);
2410 		return (0);
2411 	}
2412 
2413 	/* We may be sending a fragment so traverse the mbuf */
2414 	for (; m; m = m0) {
2415 		m0 = m->m_nextpkt;
2416 		m->m_nextpkt = NULL;
2417 		len = m->m_pkthdr.len;
2418 		mflags = m->m_flags;
2419 
2420 		/*
2421 		 * If the native VLAN ID of the outgoing interface matches the
2422 		 * VLAN ID of the frame, remove the VLAN tag.
2423 		 */
2424 		if (bif->bif_pvid != DOT1Q_VID_NULL &&
2425 		    VLANTAGOF(m) == bif->bif_pvid) {
2426 			m->m_flags &= ~M_VLANTAG;
2427 			m->m_pkthdr.ether_vtag = 0;
2428 		}
2429 
2430 		/*
2431 		 * There are two cases where we have to insert our own tag:
2432 		 * if the member interface doesn't support hardware tagging,
2433 		 * or if the tag proto is not 802.1q.
2434 		 */
2435 		if ((m->m_flags & M_VLANTAG) &&
2436 		    ((dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0 ||
2437 		      bif->bif_vlanproto != ETHERTYPE_VLAN)) {
2438 			m = ether_vlanencap_proto(m, m->m_pkthdr.ether_vtag,
2439 			    bif->bif_vlanproto);
2440 			if (m == NULL) {
2441 				if_printf(dst_ifp,
2442 				    "unable to prepend VLAN header\n");
2443 				if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2444 				continue;
2445 			}
2446 			m->m_flags &= ~M_VLANTAG;
2447 		}
2448 
2449 		M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2450 		/*
2451 		 * XXXZL: gif(4) requires the af to be saved in csum_data field
2452 		 * so that gif_transmit() routine can pull it back.
2453 		 */
2454 		if (dst_ifp->if_type == IFT_GIF)
2455 			m->m_pkthdr.csum_data = AF_LINK;
2456 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2457 			int n;
2458 
2459 			for (m = m0, n = 1; m != NULL; m = m0, n++) {
2460 				m0 = m->m_nextpkt;
2461 				m_freem(m);
2462 			}
2463 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, n);
2464 			break;
2465 		}
2466 
2467 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2468 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2469 		if (mflags & M_MCAST)
2470 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2471 	}
2472 
2473 	return (err);
2474 }
2475 
2476 /*
2477  * bridge_dummynet:
2478  *
2479  * 	Receive a queued packet from dummynet and pass it on to the output
2480  * 	interface.
2481  *
2482  *	The mbuf has the Ethernet header already attached.
2483  */
2484 static void
bridge_dummynet(struct mbuf * m,struct ifnet * ifp)2485 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2486 {
2487 	struct bridge_iflist *bif = ifp->if_bridge;
2488 	struct bridge_softc *sc = NULL;
2489 
2490 	if (bif)
2491 		sc = bif->bif_sc;
2492 
2493 	/*
2494 	 * The packet didnt originate from a member interface. This should only
2495 	 * ever happen if a member interface is removed while packets are
2496 	 * queued for it.
2497 	 */
2498 	if (sc == NULL) {
2499 		m_freem(m);
2500 		return;
2501 	}
2502 
2503 	if (PFIL_HOOKED_OUT_46) {
2504 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2505 			return;
2506 		if (m == NULL)
2507 			return;
2508 	}
2509 
2510 	bridge_enqueue(sc, ifp, m, NULL);
2511 }
2512 
2513 /*
2514  * bridge_output:
2515  *
2516  *	Send output from a bridge member interface.  This
2517  *	performs the bridging function for locally originated
2518  *	packets.
2519  *
2520  *	The mbuf has the Ethernet header already attached.  We must
2521  *	enqueue or free the mbuf before returning.
2522  */
2523 static int
bridge_output(struct ifnet * ifp,struct mbuf * m,struct sockaddr * sa,struct rtentry * rt)2524 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2525     struct rtentry *rt)
2526 {
2527 	struct ether_header *eh;
2528 	struct bridge_iflist *sbif;
2529 	struct ifnet *bifp, *dst_if;
2530 	struct bridge_softc *sc;
2531 	ether_vlanid_t vlan;
2532 
2533 	NET_EPOCH_ASSERT();
2534 
2535 	if (m->m_len < ETHER_HDR_LEN) {
2536 		m = m_pullup(m, ETHER_HDR_LEN);
2537 		if (m == NULL)
2538 			return (0);
2539 	}
2540 
2541 	sbif = ifp->if_bridge;
2542 	sc = sbif->bif_sc;
2543 	bifp = sc->sc_ifp;
2544 
2545 	eh = mtod(m, struct ether_header *);
2546 	vlan = VLANTAGOF(m);
2547 
2548 	/*
2549 	 * If bridge is down, but the original output interface is up,
2550 	 * go ahead and send out that interface.  Otherwise, the packet
2551 	 * is dropped below.
2552 	 */
2553 	if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2554 		dst_if = ifp;
2555 		goto sendunicast;
2556 	}
2557 
2558 	/*
2559 	 * If the packet is a multicast, or we don't know a better way to
2560 	 * get there, send to all interfaces.
2561 	 */
2562 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
2563 		dst_if = NULL;
2564 	else
2565 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2566 	/* Tap any traffic not passing back out the originating interface */
2567 	if (dst_if != ifp)
2568 		ETHER_BPF_MTAP(bifp, m);
2569 	if (dst_if == NULL) {
2570 		struct bridge_iflist *bif;
2571 		struct mbuf *mc;
2572 		int used = 0;
2573 
2574 		bridge_span(sc, m);
2575 
2576 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2577 			dst_if = bif->bif_ifp;
2578 
2579 			if (dst_if->if_type == IFT_GIF)
2580 				continue;
2581 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2582 				continue;
2583 
2584 			/*
2585 			 * If this is not the original output interface,
2586 			 * and the interface is participating in spanning
2587 			 * tree, make sure the port is in a state that
2588 			 * allows forwarding.
2589 			 */
2590 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2591 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2592 				continue;
2593 
2594 			if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2595 				used = 1;
2596 				mc = m;
2597 			} else {
2598 				mc = m_dup(m, M_NOWAIT);
2599 				if (mc == NULL) {
2600 					if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2601 					continue;
2602 				}
2603 			}
2604 
2605 			bridge_enqueue(sc, dst_if, mc, bif);
2606 		}
2607 		if (used == 0)
2608 			m_freem(m);
2609 		return (0);
2610 	}
2611 
2612 sendunicast:
2613 	/*
2614 	 * XXX Spanning tree consideration here?
2615 	 */
2616 
2617 	bridge_span(sc, m);
2618 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2619 		m_freem(m);
2620 		return (0);
2621 	}
2622 
2623 	bridge_enqueue(sc, dst_if, m, NULL);
2624 	return (0);
2625 }
2626 
2627 /*
2628  * bridge_transmit:
2629  *
2630  *	Do output on a bridge.
2631  *
2632  */
2633 static int
bridge_transmit(struct ifnet * ifp,struct mbuf * m)2634 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2635 {
2636 	struct bridge_softc *sc;
2637 	struct ether_header *eh;
2638 	struct ifnet *dst_if;
2639 	int error = 0;
2640 	ether_vlanid_t vlan;
2641 
2642 	sc = ifp->if_softc;
2643 
2644 	ETHER_BPF_MTAP(ifp, m);
2645 
2646 	eh = mtod(m, struct ether_header *);
2647 	vlan = VLANTAGOF(m);
2648 
2649 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2650 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan)) != NULL) {
2651 		error = bridge_enqueue(sc, dst_if, m, NULL);
2652 	} else
2653 		bridge_broadcast(sc, ifp, m, 0);
2654 
2655 	return (error);
2656 }
2657 
2658 #ifdef ALTQ
2659 static void
bridge_altq_start(if_t ifp)2660 bridge_altq_start(if_t ifp)
2661 {
2662 	struct ifaltq *ifq = &ifp->if_snd;
2663 	struct mbuf *m;
2664 
2665 	IFQ_LOCK(ifq);
2666 	IFQ_DEQUEUE_NOLOCK(ifq, m);
2667 	while (m != NULL) {
2668 		bridge_transmit(ifp, m);
2669 		IFQ_DEQUEUE_NOLOCK(ifq, m);
2670 	}
2671 	IFQ_UNLOCK(ifq);
2672 }
2673 
2674 static int
bridge_altq_transmit(if_t ifp,struct mbuf * m)2675 bridge_altq_transmit(if_t ifp, struct mbuf *m)
2676 {
2677 	int err;
2678 
2679 	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
2680 		IFQ_ENQUEUE(&ifp->if_snd, m, err);
2681 		if (err == 0)
2682 			bridge_altq_start(ifp);
2683 	} else
2684 		err = bridge_transmit(ifp, m);
2685 
2686 	return (err);
2687 }
2688 #endif	/* ALTQ */
2689 
2690 /*
2691  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2692  */
2693 static void
bridge_qflush(struct ifnet * ifp __unused)2694 bridge_qflush(struct ifnet *ifp __unused)
2695 {
2696 }
2697 
2698 /*
2699  * bridge_forward:
2700  *
2701  *	The forwarding function of the bridge.
2702  *
2703  *	NOTE: Releases the lock on return.
2704  */
2705 static void
bridge_forward(struct bridge_softc * sc,struct bridge_iflist * sbif,struct mbuf * m)2706 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2707     struct mbuf *m)
2708 {
2709 	struct bridge_iflist *dbif;
2710 	struct ifnet *src_if, *dst_if, *ifp;
2711 	struct ether_header *eh;
2712 	uint8_t *dst;
2713 	int error;
2714 	ether_vlanid_t vlan;
2715 
2716 	NET_EPOCH_ASSERT();
2717 
2718 	src_if = m->m_pkthdr.rcvif;
2719 	ifp = sc->sc_ifp;
2720 	vlan = VLANTAGOF(m);
2721 
2722 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2723 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2724 
2725 	if ((sbif->bif_flags & IFBIF_STP) &&
2726 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2727 		goto drop;
2728 
2729 	eh = mtod(m, struct ether_header *);
2730 	dst = eh->ether_dhost;
2731 
2732 	/* If the interface is learning, record the address. */
2733 	if (sbif->bif_flags & IFBIF_LEARNING) {
2734 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2735 		    sbif, 0, IFBAF_DYNAMIC);
2736 		/*
2737 		 * If the interface has addresses limits then deny any source
2738 		 * that is not in the cache.
2739 		 */
2740 		if (error && sbif->bif_addrmax)
2741 			goto drop;
2742 	}
2743 
2744 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2745 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2746 		goto drop;
2747 
2748 #ifdef DEV_NETMAP
2749 	/*
2750 	 * Hand the packet to netmap only if it wasn't injected by netmap
2751 	 * itself.
2752 	 */
2753 	if ((m->m_flags & M_BRIDGE_INJECT) == 0 &&
2754 	    (if_getcapenable(ifp) & IFCAP_NETMAP) != 0) {
2755 		ifp->if_input(ifp, m);
2756 		return;
2757 	}
2758 	m->m_flags &= ~M_BRIDGE_INJECT;
2759 #endif
2760 
2761 	/*
2762 	 * At this point, the port either doesn't participate
2763 	 * in spanning tree or it is in the forwarding state.
2764 	 */
2765 
2766 	/*
2767 	 * If the packet is unicast, destined for someone on
2768 	 * "this" side of the bridge, drop it.
2769 	 */
2770 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2771 		dst_if = bridge_rtlookup(sc, dst, vlan);
2772 		if (src_if == dst_if)
2773 			goto drop;
2774 	} else {
2775 		/*
2776 		 * Check if its a reserved multicast address, any address
2777 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2778 		 * bridge.
2779 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2780 		 */
2781 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2782 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2783 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2784 			goto drop;
2785 
2786 		/* ...forward it to all interfaces. */
2787 		if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2788 		dst_if = NULL;
2789 	}
2790 
2791 	/*
2792 	 * If we have a destination interface which is a member of our bridge,
2793 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2794 	 * For broadcast or multicast packets, don't bother because it will
2795 	 * be reinjected into ether_input. We do this before we pass the packets
2796 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2797 	 * drop the packet, or possibly modify it, making it difficult to debug
2798 	 * firewall issues on the bridge.
2799 	 */
2800 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2801 		ETHER_BPF_MTAP(ifp, m);
2802 
2803 	/* run the packet filter */
2804 	if (PFIL_HOOKED_IN_46) {
2805 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2806 			return;
2807 		if (m == NULL)
2808 			return;
2809 	}
2810 
2811 	if (dst_if == NULL) {
2812 		bridge_broadcast(sc, src_if, m, 1);
2813 		return;
2814 	}
2815 
2816 	/*
2817 	 * At this point, we're dealing with a unicast frame
2818 	 * going to a different interface.
2819 	 */
2820 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2821 		goto drop;
2822 
2823 	dbif = bridge_lookup_member_if(sc, dst_if);
2824 	if (dbif == NULL)
2825 		/* Not a member of the bridge (anymore?) */
2826 		goto drop;
2827 
2828 	/* Private segments can not talk to each other */
2829 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2830 		goto drop;
2831 
2832 	if ((dbif->bif_flags & IFBIF_STP) &&
2833 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2834 		goto drop;
2835 
2836 	if (PFIL_HOOKED_OUT_46) {
2837 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2838 			return;
2839 		if (m == NULL)
2840 			return;
2841 	}
2842 
2843 	bridge_enqueue(sc, dst_if, m, dbif);
2844 	return;
2845 
2846 drop:
2847 	m_freem(m);
2848 }
2849 
2850 /*
2851  * bridge_input:
2852  *
2853  *	Receive input from a member interface.  Queue the packet for
2854  *	bridging if it is not for us.
2855  */
2856 static struct mbuf *
bridge_input(struct ifnet * ifp,struct mbuf * m)2857 bridge_input(struct ifnet *ifp, struct mbuf *m)
2858 {
2859 	struct bridge_softc *sc = NULL;
2860 	struct bridge_iflist *bif, *bif2;
2861 	struct ifnet *bifp;
2862 	struct ether_header *eh;
2863 	struct mbuf *mc, *mc2;
2864 	ether_vlanid_t vlan;
2865 	int error;
2866 
2867 	NET_EPOCH_ASSERT();
2868 
2869 	/* We need the Ethernet header later, so make sure we have it now. */
2870 	if (m->m_len < ETHER_HDR_LEN) {
2871 		m = m_pullup(m, ETHER_HDR_LEN);
2872 		if (m == NULL) {
2873 			if_inc_counter(sc->sc_ifp, IFCOUNTER_IERRORS, 1);
2874 			m_freem(m);
2875 			return (NULL);
2876 		}
2877 	}
2878 
2879 	eh = mtod(m, struct ether_header *);
2880 	vlan = VLANTAGOF(m);
2881 
2882 	/*
2883 	 * If this frame has a VLAN tag and the receiving interface has a
2884 	 * vlan(4) trunk, then it is is destined for vlan(4), not for us.
2885 	 * This means if vlan(4) and bridge(4) are configured on the same
2886 	 * interface, vlan(4) is preferred, which is what users typically
2887 	 * expect.
2888 	 */
2889 	if (vlan != DOT1Q_VID_NULL && ifp->if_vlantrunk != NULL)
2890 		return (m);
2891 
2892 	bif = ifp->if_bridge;
2893 	if (bif)
2894 		sc = bif->bif_sc;
2895 
2896 	if (sc == NULL) {
2897 		/*
2898 		 * This packet originated from the bridge itself, so it must
2899 		 * have been transmitted by netmap.  Derive the "source"
2900 		 * interface from the source address and drop the packet if the
2901 		 * source address isn't known.
2902 		 */
2903 		KASSERT((m->m_flags & M_BRIDGE_INJECT) != 0,
2904 		    ("%s: ifnet %p missing a bridge softc", __func__, ifp));
2905 		sc = if_getsoftc(ifp);
2906 		ifp = bridge_rtlookup(sc, eh->ether_shost, vlan);
2907 		if (ifp == NULL) {
2908 			if_inc_counter(sc->sc_ifp, IFCOUNTER_IERRORS, 1);
2909 			m_freem(m);
2910 			return (NULL);
2911 		}
2912 		m->m_pkthdr.rcvif = ifp;
2913 	}
2914 	bifp = sc->sc_ifp;
2915 	if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2916 		return (m);
2917 
2918 	/*
2919 	 * Implement support for bridge monitoring. If this flag has been
2920 	 * set on this interface, discard the packet once we push it through
2921 	 * the bpf(4) machinery, but before we do, increment the byte and
2922 	 * packet counters associated with this interface.
2923 	 */
2924 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2925 		m->m_pkthdr.rcvif  = bifp;
2926 		ETHER_BPF_MTAP(bifp, m);
2927 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2928 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2929 		m_freem(m);
2930 		return (NULL);
2931 	}
2932 
2933 	/* Do VLAN filtering. */
2934 	if (!bridge_vfilter_in(bif, m)) {
2935 		if_inc_counter(sc->sc_ifp, IFCOUNTER_IERRORS, 1);
2936 		m_freem(m);
2937 		return (NULL);
2938 	}
2939 	/* bridge_vfilter_in() may add a tag */
2940 	vlan = VLANTAGOF(m);
2941 
2942 	bridge_span(sc, m);
2943 
2944 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2945 		/* Tap off 802.1D packets; they do not get forwarded. */
2946 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2947 		    ETHER_ADDR_LEN) == 0) {
2948 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2949 			return (NULL);
2950 		}
2951 
2952 		if ((bif->bif_flags & IFBIF_STP) &&
2953 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2954 			return (m);
2955 		}
2956 
2957 		/*
2958 		 * Make a deep copy of the packet and enqueue the copy
2959 		 * for bridge processing; return the original packet for
2960 		 * local processing.
2961 		 */
2962 		mc = m_dup(m, M_NOWAIT);
2963 		if (mc == NULL) {
2964 			return (m);
2965 		}
2966 
2967 		/* Perform the bridge forwarding function with the copy. */
2968 		bridge_forward(sc, bif, mc);
2969 
2970 #ifdef DEV_NETMAP
2971 		/*
2972 		 * If netmap is enabled and has not already seen this packet,
2973 		 * then it will be consumed by bridge_forward().
2974 		 */
2975 		if ((if_getcapenable(bifp) & IFCAP_NETMAP) != 0 &&
2976 		    (m->m_flags & M_BRIDGE_INJECT) == 0) {
2977 			m_freem(m);
2978 			return (NULL);
2979 		}
2980 #endif
2981 
2982 		/*
2983 		 * Reinject the mbuf as arriving on the bridge so we have a
2984 		 * chance at claiming multicast packets. We can not loop back
2985 		 * here from ether_input as a bridge is never a member of a
2986 		 * bridge.
2987 		 */
2988 		KASSERT(bifp->if_bridge == NULL,
2989 		    ("loop created in bridge_input"));
2990 		mc2 = m_dup(m, M_NOWAIT);
2991 		if (mc2 != NULL) {
2992 			/* Keep the layer3 header aligned */
2993 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2994 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2995 		}
2996 		if (mc2 != NULL) {
2997 			mc2->m_pkthdr.rcvif = bifp;
2998 			mc2->m_flags &= ~M_BRIDGE_INJECT;
2999 			sc->sc_if_input(bifp, mc2);
3000 		}
3001 
3002 		/* Return the original packet for local processing. */
3003 		return (m);
3004 	}
3005 
3006 	if ((bif->bif_flags & IFBIF_STP) &&
3007 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
3008 		return (m);
3009 	}
3010 
3011 #if defined(INET) || defined(INET6)
3012 #define	CARP_CHECK_WE_ARE_DST(iface) \
3013 	((iface)->if_carp && (*carp_forus_p)((iface), eh->ether_dhost))
3014 #define	CARP_CHECK_WE_ARE_SRC(iface) \
3015 	((iface)->if_carp && (*carp_forus_p)((iface), eh->ether_shost))
3016 #else
3017 #define	CARP_CHECK_WE_ARE_DST(iface)	false
3018 #define	CARP_CHECK_WE_ARE_SRC(iface)	false
3019 #endif
3020 
3021 #ifdef DEV_NETMAP
3022 #define	GRAB_FOR_NETMAP(ifp, m) do {					\
3023 	if ((if_getcapenable(ifp) & IFCAP_NETMAP) != 0 &&		\
3024 	    ((m)->m_flags & M_BRIDGE_INJECT) == 0) {			\
3025 		(ifp)->if_input(ifp, m);				\
3026 		return (NULL);						\
3027 	}								\
3028 } while (0)
3029 #else
3030 #define	GRAB_FOR_NETMAP(ifp, m)
3031 #endif
3032 
3033 #define GRAB_OUR_PACKETS(iface)						\
3034 	if ((iface)->if_type == IFT_GIF)				\
3035 		continue;						\
3036 	/* It is destined for us. */					\
3037 	if (memcmp(IF_LLADDR(iface), eh->ether_dhost, ETHER_ADDR_LEN) == 0 || \
3038 	    CARP_CHECK_WE_ARE_DST(iface)) {				\
3039 		if (bif->bif_flags & IFBIF_LEARNING) {			\
3040 			error = bridge_rtupdate(sc, eh->ether_shost,	\
3041 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
3042 			if (error && bif->bif_addrmax) {		\
3043 				m_freem(m);				\
3044 				return (NULL);				\
3045 			}						\
3046 		}							\
3047 		m->m_pkthdr.rcvif = iface;				\
3048 		if ((iface) == ifp) {					\
3049 			/* Skip bridge processing... src == dest */	\
3050 			return (m);					\
3051 		}							\
3052 		/* It's passing over or to the bridge, locally. */	\
3053 		ETHER_BPF_MTAP(bifp, m);				\
3054 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);		\
3055 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);\
3056 		/* Hand the packet over to netmap if necessary. */	\
3057 		GRAB_FOR_NETMAP(bifp, m);				\
3058 		/* Filter on the physical interface. */			\
3059 		if (V_pfil_local_phys && PFIL_HOOKED_IN_46) {		\
3060 			if (bridge_pfil(&m, NULL, ifp,			\
3061 			    PFIL_IN) != 0 || m == NULL) {		\
3062 				return (NULL);				\
3063 			}						\
3064 		}							\
3065 		if ((iface) != bifp)					\
3066 			ETHER_BPF_MTAP(iface, m);			\
3067 		/* Pass tagged packets to if_vlan, if it's loaded */	\
3068 		if (VLANTAGOF(m) != 0) {				\
3069 			if (bifp->if_vlantrunk == NULL) {		\
3070 				m_freem(m);				\
3071 				return (NULL);				\
3072 			}						\
3073 			(*vlan_input_p)(bifp, m);			\
3074 			return (NULL);					\
3075 		}							\
3076 		return (m);						\
3077 	}								\
3078 									\
3079 	/* We just received a packet that we sent out. */		\
3080 	if (memcmp(IF_LLADDR(iface), eh->ether_shost, ETHER_ADDR_LEN) == 0 || \
3081 	    CARP_CHECK_WE_ARE_SRC(iface)) {				\
3082 		m_freem(m);						\
3083 		return (NULL);						\
3084 	}
3085 
3086 	/*
3087 	 * Unicast.  Make sure it's not for the bridge.
3088 	 */
3089 	do { GRAB_OUR_PACKETS(bifp) } while (0);
3090 
3091 	/*
3092 	 * If member_ifaddrs is enabled, see if the packet is destined for
3093 	 * one of the members' addresses.
3094 	 */
3095 	if (V_member_ifaddrs) {
3096 		/* Check the interface the packet arrived on. */
3097 		do { GRAB_OUR_PACKETS(ifp) } while (0);
3098 
3099 		CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
3100 			GRAB_OUR_PACKETS(bif2->bif_ifp)
3101 		}
3102 	}
3103 
3104 #undef CARP_CHECK_WE_ARE_DST
3105 #undef CARP_CHECK_WE_ARE_SRC
3106 #undef GRAB_FOR_NETMAP
3107 #undef GRAB_OUR_PACKETS
3108 
3109 	/* Perform the bridge forwarding function. */
3110 	bridge_forward(sc, bif, m);
3111 
3112 	return (NULL);
3113 }
3114 
3115 /*
3116  * Inject a packet back into the host ethernet stack.  This will generally only
3117  * be used by netmap when an application writes to the host TX ring.  The
3118  * M_BRIDGE_INJECT flag ensures that the packet is re-routed to the bridge
3119  * interface after ethernet processing.
3120  */
3121 static void
bridge_inject(struct ifnet * ifp,struct mbuf * m)3122 bridge_inject(struct ifnet *ifp, struct mbuf *m)
3123 {
3124 	struct bridge_softc *sc;
3125 
3126 	if (ifp->if_type == IFT_L2VLAN) {
3127 		/*
3128 		 * vlan(4) gives us the vlan ifnet, so we need to get the
3129 		 * bridge softc to get a pointer to ether_input to send the
3130 		 * packet to.
3131 		 */
3132 		struct ifnet *bifp = NULL;
3133 
3134 		if (vlan_trunkdev_p == NULL) {
3135 			m_freem(m);
3136 			return;
3137 		}
3138 
3139 		bifp = vlan_trunkdev_p(ifp);
3140 		if (bifp == NULL) {
3141 			m_freem(m);
3142 			return;
3143 		}
3144 
3145 		sc = if_getsoftc(bifp);
3146 		sc->sc_if_input(ifp, m);
3147 		return;
3148 	}
3149 
3150 	KASSERT((if_getcapenable(ifp) & IFCAP_NETMAP) != 0,
3151 	    ("%s: iface %s is not running in netmap mode",
3152 	    __func__, if_name(ifp)));
3153 	KASSERT((m->m_flags & M_BRIDGE_INJECT) == 0,
3154 	    ("%s: mbuf %p has M_BRIDGE_INJECT set", __func__, m));
3155 
3156 	m->m_flags |= M_BRIDGE_INJECT;
3157 	sc = if_getsoftc(ifp);
3158 	sc->sc_if_input(ifp, m);
3159 }
3160 
3161 /*
3162  * bridge_broadcast:
3163  *
3164  *	Send a frame to all interfaces that are members of
3165  *	the bridge, except for the one on which the packet
3166  *	arrived.
3167  *
3168  *	NOTE: Releases the lock on return.
3169  */
3170 static void
bridge_broadcast(struct bridge_softc * sc,struct ifnet * src_if,struct mbuf * m,int runfilt)3171 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
3172     struct mbuf *m, int runfilt)
3173 {
3174 	struct bridge_iflist *dbif, *sbif;
3175 	struct mbuf *mc;
3176 	struct ifnet *dst_if;
3177 	int used = 0, i;
3178 
3179 	NET_EPOCH_ASSERT();
3180 
3181 	sbif = bridge_lookup_member_if(sc, src_if);
3182 
3183 	/* Filter on the bridge interface before broadcasting */
3184 	if (runfilt && PFIL_HOOKED_OUT_46) {
3185 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
3186 			return;
3187 		if (m == NULL)
3188 			return;
3189 	}
3190 
3191 	CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
3192 		dst_if = dbif->bif_ifp;
3193 		if (dst_if == src_if)
3194 			continue;
3195 
3196 		/* Private segments can not talk to each other */
3197 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
3198 			continue;
3199 
3200 		if ((dbif->bif_flags & IFBIF_STP) &&
3201 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
3202 			continue;
3203 
3204 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
3205 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
3206 			continue;
3207 
3208 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
3209 			continue;
3210 
3211 		if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
3212 			mc = m;
3213 			used = 1;
3214 		} else {
3215 			mc = m_dup(m, M_NOWAIT);
3216 			if (mc == NULL) {
3217 				if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
3218 				continue;
3219 			}
3220 		}
3221 
3222 		/*
3223 		 * Filter on the output interface. Pass a NULL bridge interface
3224 		 * pointer so we do not redundantly filter on the bridge for
3225 		 * each interface we broadcast on.
3226 		 */
3227 		if (runfilt && PFIL_HOOKED_OUT_46) {
3228 			if (used == 0) {
3229 				/* Keep the layer3 header aligned */
3230 				i = min(mc->m_pkthdr.len, max_protohdr);
3231 				mc = m_copyup(mc, i, ETHER_ALIGN);
3232 				if (mc == NULL) {
3233 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
3234 					continue;
3235 				}
3236 			}
3237 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
3238 				continue;
3239 			if (mc == NULL)
3240 				continue;
3241 		}
3242 
3243 		bridge_enqueue(sc, dst_if, mc, dbif);
3244 	}
3245 	if (used == 0)
3246 		m_freem(m);
3247 }
3248 
3249 /*
3250  * bridge_span:
3251  *
3252  *	Duplicate a packet out one or more interfaces that are in span mode,
3253  *	the original mbuf is unmodified.
3254  */
3255 static void
bridge_span(struct bridge_softc * sc,struct mbuf * m)3256 bridge_span(struct bridge_softc *sc, struct mbuf *m)
3257 {
3258 	struct bridge_iflist *bif;
3259 	struct ifnet *dst_if;
3260 	struct mbuf *mc;
3261 
3262 	NET_EPOCH_ASSERT();
3263 
3264 	if (CK_LIST_EMPTY(&sc->sc_spanlist))
3265 		return;
3266 
3267 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
3268 		dst_if = bif->bif_ifp;
3269 
3270 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
3271 			continue;
3272 
3273 		mc = m_dup(m, M_NOWAIT);
3274 		if (mc == NULL) {
3275 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
3276 			continue;
3277 		}
3278 
3279 		bridge_enqueue(sc, dst_if, mc, bif);
3280 	}
3281 }
3282 
3283 /*
3284  * Incoming VLAN filtering.  Given a frame and the member interface it was
3285  * received on, decide whether the port configuration allows it.
3286  */
3287 static bool
bridge_vfilter_in(const struct bridge_iflist * sbif,struct mbuf * m)3288 bridge_vfilter_in(const struct bridge_iflist *sbif, struct mbuf *m)
3289 {
3290 	ether_vlanid_t vlan;
3291 
3292 	vlan = VLANTAGOF(m);
3293 	/* Make sure the vlan id is reasonable. */
3294 	if (vlan > DOT1Q_VID_MAX)
3295 		return (false);
3296 
3297 	/*
3298 	 * If VLAN filtering isn't enabled, pass everything, but add a tag
3299 	 * if the port has a pvid configured.
3300 	 */
3301 	if ((sbif->bif_sc->sc_flags & IFBRF_VLANFILTER) == 0) {
3302 		if (vlan == DOT1Q_VID_NULL &&
3303 		    sbif->bif_pvid != DOT1Q_VID_NULL) {
3304 			m->m_pkthdr.ether_vtag = sbif->bif_pvid;
3305 			m->m_flags |= M_VLANTAG;
3306 		}
3307 
3308 		return (true);
3309 	}
3310 
3311 	/* If Q-in-Q is disabled, check for stacked tags. */
3312 	if ((sbif->bif_flags & IFBIF_QINQ) == 0) {
3313 		struct ether_header *eh;
3314 		uint16_t proto;
3315 
3316 		eh = mtod(m, struct ether_header *);
3317 		proto = ntohs(eh->ether_type);
3318 
3319 		if (proto == ETHERTYPE_VLAN || proto == ETHERTYPE_QINQ)
3320 			return (false);
3321 	}
3322 
3323 	if (vlan == DOT1Q_VID_NULL) {
3324 		/*
3325 		 * The frame doesn't have a tag.  If the interface does not
3326 		 * have an untagged vlan configured, drop the frame.
3327 		 */
3328 		if (sbif->bif_pvid == DOT1Q_VID_NULL)
3329 			return (false);
3330 
3331 		/*
3332 		 * Otherwise, insert a new tag based on the interface's
3333 		 * untagged vlan id.
3334 		 */
3335 		m->m_pkthdr.ether_vtag = sbif->bif_pvid;
3336 		m->m_flags |= M_VLANTAG;
3337 	} else {
3338 		/*
3339 		 * The frame has a tag, so check it matches the interface's
3340 		 * vlan access list.  We explicitly do not accept tagged
3341 		 * frames for the untagged vlan id here (unless it's also
3342 		 * in the access list).
3343 		 */
3344 		if (!BRVLAN_TEST(&sbif->bif_vlan_set, vlan))
3345 			return (false);
3346 	}
3347 
3348 	/* Accept the frame. */
3349 	return (true);
3350 }
3351 
3352 /*
3353  * Outgoing VLAN filtering.  Given a frame, its vlan, and the member interface
3354  * we intend to send it to, decide whether the port configuration allows it to
3355  * be sent.
3356  */
3357 static bool
bridge_vfilter_out(const struct bridge_iflist * dbif,const struct mbuf * m)3358 bridge_vfilter_out(const struct bridge_iflist *dbif, const struct mbuf *m)
3359 {
3360 	struct ether_header *eh;
3361 	ether_vlanid_t vlan;
3362 
3363 	NET_EPOCH_ASSERT();
3364 
3365 	/*
3366 	 * If the interface is in span mode, then bif_sc will be NULL.
3367 	 * Since the purpose of span interfaces is to receive all frames,
3368 	 * pass everything.
3369 	 */
3370 	if (dbif->bif_sc == NULL)
3371 		return (true);
3372 
3373 	/* If VLAN filtering isn't enabled, pass everything. */
3374 	if ((dbif->bif_sc->sc_flags & IFBRF_VLANFILTER) == 0)
3375 		return (true);
3376 
3377 	vlan = VLANTAGOF(m);
3378 
3379 	/*
3380 	 * Always allow untagged 802.1D STP frames, even if they would
3381 	 * otherwise be dropped.  This is required for STP to work on
3382 	 * a filtering bridge.
3383 	 *
3384 	 * Tagged STP (Cisco PVST+) is a non-standard extension, so
3385 	 * handle those frames via the normal filtering path.
3386 	 */
3387 	eh = mtod(m, struct ether_header *);
3388 	if (vlan == DOT1Q_VID_NULL &&
3389 	    memcmp(eh->ether_dhost, bstp_etheraddr, ETHER_ADDR_LEN) == 0)
3390 		return (true);
3391 
3392 	/*
3393 	 * If the frame wasn't assigned to a vlan at ingress, drop it.
3394 	 * We can't forward these frames to filtering ports because we
3395 	 * don't know what VLAN they're supposed to be in.
3396 	 */
3397 	if (vlan == DOT1Q_VID_NULL)
3398 		return (false);
3399 
3400 	/*
3401 	 * If the frame's vlan matches the interfaces's untagged vlan,
3402 	 * allow it.
3403 	 */
3404 	if (vlan == dbif->bif_pvid)
3405 		return (true);
3406 
3407 	/*
3408 	 * If the frame's vlan is on the interface's tagged access list,
3409 	 * allow it.
3410 	 */
3411 	if (BRVLAN_TEST(&dbif->bif_vlan_set, vlan))
3412 		return (true);
3413 
3414 	/* The frame was not permitted, so drop it. */
3415 	return (false);
3416 }
3417 
3418 /*
3419  * bridge_rtupdate:
3420  *
3421  *	Add a bridge routing entry.
3422  */
3423 static int
bridge_rtupdate(struct bridge_softc * sc,const uint8_t * dst,ether_vlanid_t vlan,struct bridge_iflist * bif,int setflags,uint8_t flags)3424 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst,
3425 		ether_vlanid_t vlan, struct bridge_iflist *bif,
3426 		int setflags, uint8_t flags)
3427 {
3428 	struct bridge_rtnode *brt;
3429 	struct bridge_iflist *obif;
3430 	int error;
3431 
3432 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3433 
3434 	/* Check the source address is valid and not multicast. */
3435 	if (ETHER_IS_MULTICAST(dst))
3436 		return (EXTERROR(EINVAL, "Multicast address not permitted"));
3437 	if (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
3438 	    dst[3] == 0 && dst[4] == 0 && dst[5] == 0)
3439 		return (EXTERROR(EINVAL, "Zero address not permitted"));
3440 
3441 	/*
3442 	 * A route for this destination might already exist.  If so,
3443 	 * update it, otherwise create a new one.
3444 	 */
3445 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
3446 		BRIDGE_RT_LOCK(sc);
3447 
3448 		/* Check again, now that we have the lock. There could have
3449 		 * been a race and we only want to insert this once. */
3450 		if (bridge_rtnode_lookup(sc, dst, vlan) != NULL) {
3451 			BRIDGE_RT_UNLOCK(sc);
3452 			return (0);
3453 		}
3454 
3455 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
3456 			sc->sc_brtexceeded++;
3457 			BRIDGE_RT_UNLOCK(sc);
3458 			return (EXTERROR(ENOSPC, "Address table is full"));
3459 		}
3460 		/* Check per interface address limits (if enabled) */
3461 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
3462 			bif->bif_addrexceeded++;
3463 			BRIDGE_RT_UNLOCK(sc);
3464 			return (EXTERROR(ENOSPC,
3465 			    "Interface address limit exceeded"));
3466 		}
3467 
3468 		/*
3469 		 * Allocate a new bridge forwarding node, and
3470 		 * initialize the expiration time and Ethernet
3471 		 * address.
3472 		 */
3473 		brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
3474 		if (brt == NULL) {
3475 			BRIDGE_RT_UNLOCK(sc);
3476 			return (EXTERROR(ENOMEM,
3477 			    "Cannot allocate address node"));
3478 		}
3479 		brt->brt_vnet = curvnet;
3480 
3481 		if (bif->bif_flags & IFBIF_STICKY)
3482 			brt->brt_flags = IFBAF_STICKY;
3483 		else
3484 			brt->brt_flags = IFBAF_DYNAMIC;
3485 
3486 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
3487 		brt->brt_vlan = vlan;
3488 
3489 		brt->brt_dst = bif;
3490 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
3491 			uma_zfree(V_bridge_rtnode_zone, brt);
3492 			BRIDGE_RT_UNLOCK(sc);
3493 			return (error);
3494 		}
3495 		bif->bif_addrcnt++;
3496 
3497 		BRIDGE_RT_UNLOCK(sc);
3498 	}
3499 
3500 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
3501 	    (obif = brt->brt_dst) != bif) {
3502 		MPASS(obif != NULL);
3503 
3504 		BRIDGE_RT_LOCK(sc);
3505 		brt->brt_dst->bif_addrcnt--;
3506 		brt->brt_dst = bif;
3507 		brt->brt_dst->bif_addrcnt++;
3508 		BRIDGE_RT_UNLOCK(sc);
3509 
3510 		if (V_log_mac_flap &&
3511 		    ppsratecheck(&V_log_last, &V_log_count, V_log_interval)) {
3512 			log(LOG_NOTICE,
3513 			    "%s: mac address %6D vlan %d moved from %s to %s\n",
3514 			    sc->sc_ifp->if_xname,
3515 			    &brt->brt_addr[0], ":",
3516 			    brt->brt_vlan,
3517 			    obif->bif_ifp->if_xname,
3518 			    bif->bif_ifp->if_xname);
3519 		}
3520 	}
3521 
3522 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3523 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
3524 	if (setflags)
3525 		brt->brt_flags = flags;
3526 
3527 	return (0);
3528 }
3529 
3530 /*
3531  * bridge_rtlookup:
3532  *
3533  *	Lookup the destination interface for an address.
3534  */
3535 static struct ifnet *
bridge_rtlookup(struct bridge_softc * sc,const uint8_t * addr,ether_vlanid_t vlan)3536 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr,
3537 		ether_vlanid_t vlan)
3538 {
3539 	struct bridge_rtnode *brt;
3540 
3541 	NET_EPOCH_ASSERT();
3542 
3543 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
3544 		return (NULL);
3545 
3546 	return (brt->brt_ifp);
3547 }
3548 
3549 /*
3550  * bridge_rttrim:
3551  *
3552  *	Trim the routine table so that we have a number
3553  *	of routing entries less than or equal to the
3554  *	maximum number.
3555  */
3556 static void
bridge_rttrim(struct bridge_softc * sc)3557 bridge_rttrim(struct bridge_softc *sc)
3558 {
3559 	struct bridge_rtnode *brt, *nbrt;
3560 
3561 	NET_EPOCH_ASSERT();
3562 	BRIDGE_RT_LOCK_ASSERT(sc);
3563 
3564 	/* Make sure we actually need to do this. */
3565 	if (sc->sc_brtcnt <= sc->sc_brtmax)
3566 		return;
3567 
3568 	/* Force an aging cycle; this might trim enough addresses. */
3569 	bridge_rtage(sc);
3570 	if (sc->sc_brtcnt <= sc->sc_brtmax)
3571 		return;
3572 
3573 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3574 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
3575 			bridge_rtnode_destroy(sc, brt);
3576 			if (sc->sc_brtcnt <= sc->sc_brtmax)
3577 				return;
3578 		}
3579 	}
3580 }
3581 
3582 /*
3583  * bridge_timer:
3584  *
3585  *	Aging timer for the bridge.
3586  */
3587 static void
bridge_timer(void * arg)3588 bridge_timer(void *arg)
3589 {
3590 	struct bridge_softc *sc = arg;
3591 
3592 	BRIDGE_RT_LOCK_ASSERT(sc);
3593 
3594 	/* Destruction of rtnodes requires a proper vnet context */
3595 	CURVNET_SET(sc->sc_ifp->if_vnet);
3596 	bridge_rtage(sc);
3597 
3598 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
3599 		callout_reset(&sc->sc_brcallout,
3600 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
3601 	CURVNET_RESTORE();
3602 }
3603 
3604 /*
3605  * bridge_rtage:
3606  *
3607  *	Perform an aging cycle.
3608  */
3609 static void
bridge_rtage(struct bridge_softc * sc)3610 bridge_rtage(struct bridge_softc *sc)
3611 {
3612 	struct bridge_rtnode *brt, *nbrt;
3613 
3614 	BRIDGE_RT_LOCK_ASSERT(sc);
3615 
3616 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3617 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
3618 			if (time_uptime >= brt->brt_expire)
3619 				bridge_rtnode_destroy(sc, brt);
3620 		}
3621 	}
3622 }
3623 
3624 /*
3625  * bridge_rtflush:
3626  *
3627  *	Remove all dynamic addresses from the bridge.
3628  */
3629 static void
bridge_rtflush(struct bridge_softc * sc,int full)3630 bridge_rtflush(struct bridge_softc *sc, int full)
3631 {
3632 	struct bridge_rtnode *brt, *nbrt;
3633 
3634 	BRIDGE_RT_LOCK_ASSERT(sc);
3635 
3636 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3637 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3638 			bridge_rtnode_destroy(sc, brt);
3639 	}
3640 }
3641 
3642 /*
3643  * bridge_rtdaddr:
3644  *
3645  *	Remove an address from the table.
3646  */
3647 static int
bridge_rtdaddr(struct bridge_softc * sc,const uint8_t * addr,ether_vlanid_t vlan)3648 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr,
3649 	       ether_vlanid_t vlan)
3650 {
3651 	struct bridge_rtnode *brt;
3652 	int found = 0;
3653 
3654 	BRIDGE_RT_LOCK(sc);
3655 
3656 	/*
3657 	 * If vlan is DOT1Q_VID_RSVD_IMPL then we want to delete for all vlans
3658 	 * so the lookup may return more than one.
3659 	 */
3660 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
3661 		bridge_rtnode_destroy(sc, brt);
3662 		found = 1;
3663 	}
3664 
3665 	BRIDGE_RT_UNLOCK(sc);
3666 
3667 	return (found ? 0 : ENOENT);
3668 }
3669 
3670 /*
3671  * bridge_rtdelete:
3672  *
3673  *	Delete routes to a speicifc member interface.
3674  */
3675 static void
bridge_rtdelete(struct bridge_softc * sc,struct ifnet * ifp,int full)3676 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
3677 {
3678 	struct bridge_rtnode *brt, *nbrt;
3679 
3680 	BRIDGE_RT_LOCK_ASSERT(sc);
3681 
3682 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3683 		if (brt->brt_ifp == ifp && (full ||
3684 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
3685 			bridge_rtnode_destroy(sc, brt);
3686 	}
3687 }
3688 
3689 /*
3690  * bridge_rtable_init:
3691  *
3692  *	Initialize the route table for this bridge.
3693  */
3694 static void
bridge_rtable_init(struct bridge_softc * sc)3695 bridge_rtable_init(struct bridge_softc *sc)
3696 {
3697 	int i;
3698 
3699 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
3700 	    M_DEVBUF, M_WAITOK);
3701 
3702 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
3703 		CK_LIST_INIT(&sc->sc_rthash[i]);
3704 
3705 	sc->sc_rthash_key = arc4random();
3706 	CK_LIST_INIT(&sc->sc_rtlist);
3707 }
3708 
3709 /*
3710  * bridge_rtable_fini:
3711  *
3712  *	Deconstruct the route table for this bridge.
3713  */
3714 static void
bridge_rtable_fini(struct bridge_softc * sc)3715 bridge_rtable_fini(struct bridge_softc *sc)
3716 {
3717 
3718 	KASSERT(sc->sc_brtcnt == 0,
3719 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3720 	free(sc->sc_rthash, M_DEVBUF);
3721 }
3722 
3723 /*
3724  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3725  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3726  */
3727 #define	mix(a, b, c)							\
3728 do {									\
3729 	a -= b; a -= c; a ^= (c >> 13);					\
3730 	b -= c; b -= a; b ^= (a << 8);					\
3731 	c -= a; c -= b; c ^= (b >> 13);					\
3732 	a -= b; a -= c; a ^= (c >> 12);					\
3733 	b -= c; b -= a; b ^= (a << 16);					\
3734 	c -= a; c -= b; c ^= (b >> 5);					\
3735 	a -= b; a -= c; a ^= (c >> 3);					\
3736 	b -= c; b -= a; b ^= (a << 10);					\
3737 	c -= a; c -= b; c ^= (b >> 15);					\
3738 } while (/*CONSTCOND*/0)
3739 
3740 static __inline uint32_t
bridge_rthash(struct bridge_softc * sc,const uint8_t * addr)3741 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3742 {
3743 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3744 
3745 	b += addr[5] << 8;
3746 	b += addr[4];
3747 	a += addr[3] << 24;
3748 	a += addr[2] << 16;
3749 	a += addr[1] << 8;
3750 	a += addr[0];
3751 
3752 	mix(a, b, c);
3753 
3754 	return (c & BRIDGE_RTHASH_MASK);
3755 }
3756 
3757 #undef mix
3758 
3759 static int
bridge_rtnode_addr_cmp(const uint8_t * a,const uint8_t * b)3760 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3761 {
3762 	int i, d;
3763 
3764 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3765 		d = ((int)a[i]) - ((int)b[i]);
3766 	}
3767 
3768 	return (d);
3769 }
3770 
3771 /*
3772  * bridge_rtnode_lookup:
3773  *
3774  *	Look up a bridge route node for the specified destination. Compare the
3775  *	vlan id or if zero then just return the first match.
3776  */
3777 static struct bridge_rtnode *
bridge_rtnode_lookup(struct bridge_softc * sc,const uint8_t * addr,ether_vlanid_t vlan)3778 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr,
3779 		     ether_vlanid_t vlan)
3780 {
3781 	struct bridge_rtnode *brt;
3782 	uint32_t hash;
3783 	int dir;
3784 
3785 	BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3786 
3787 	hash = bridge_rthash(sc, addr);
3788 	CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3789 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3790 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == DOT1Q_VID_RSVD_IMPL))
3791 			return (brt);
3792 		if (dir > 0)
3793 			return (NULL);
3794 	}
3795 
3796 	return (NULL);
3797 }
3798 
3799 /*
3800  * bridge_rtnode_insert:
3801  *
3802  *	Insert the specified bridge node into the route table.  We
3803  *	assume the entry is not already in the table.
3804  */
3805 static int
bridge_rtnode_insert(struct bridge_softc * sc,struct bridge_rtnode * brt)3806 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3807 {
3808 	struct bridge_rtnode *lbrt;
3809 	uint32_t hash;
3810 	int dir;
3811 
3812 	BRIDGE_RT_LOCK_ASSERT(sc);
3813 
3814 	hash = bridge_rthash(sc, brt->brt_addr);
3815 
3816 	lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3817 	if (lbrt == NULL) {
3818 		CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3819 		goto out;
3820 	}
3821 
3822 	do {
3823 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3824 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3825 			return (EXTERROR(EEXIST, "Address already exists"));
3826 		if (dir > 0) {
3827 			CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3828 			goto out;
3829 		}
3830 		if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3831 			CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3832 			goto out;
3833 		}
3834 		lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3835 	} while (lbrt != NULL);
3836 
3837 #ifdef DIAGNOSTIC
3838 	panic("bridge_rtnode_insert: impossible");
3839 #endif
3840 
3841 out:
3842 	CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3843 	sc->sc_brtcnt++;
3844 
3845 	return (0);
3846 }
3847 
3848 static void
bridge_rtnode_destroy_cb(struct epoch_context * ctx)3849 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3850 {
3851 	struct bridge_rtnode *brt;
3852 
3853 	brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3854 
3855 	CURVNET_SET(brt->brt_vnet);
3856 	uma_zfree(V_bridge_rtnode_zone, brt);
3857 	CURVNET_RESTORE();
3858 }
3859 
3860 /*
3861  * bridge_rtnode_destroy:
3862  *
3863  *	Destroy a bridge rtnode.
3864  */
3865 static void
bridge_rtnode_destroy(struct bridge_softc * sc,struct bridge_rtnode * brt)3866 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3867 {
3868 	BRIDGE_RT_LOCK_ASSERT(sc);
3869 
3870 	CK_LIST_REMOVE(brt, brt_hash);
3871 
3872 	CK_LIST_REMOVE(brt, brt_list);
3873 	sc->sc_brtcnt--;
3874 	brt->brt_dst->bif_addrcnt--;
3875 
3876 	NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx);
3877 }
3878 
3879 /*
3880  * bridge_rtable_expire:
3881  *
3882  *	Set the expiry time for all routes on an interface.
3883  */
3884 static void
bridge_rtable_expire(struct ifnet * ifp,int age)3885 bridge_rtable_expire(struct ifnet *ifp, int age)
3886 {
3887 	struct bridge_iflist *bif = NULL;
3888 	struct bridge_softc *sc = NULL;
3889 	struct bridge_rtnode *brt;
3890 
3891 	CURVNET_SET(ifp->if_vnet);
3892 
3893 	bif = ifp->if_bridge;
3894 	if (bif)
3895 		sc = bif->bif_sc;
3896 	MPASS(sc != NULL);
3897 	BRIDGE_RT_LOCK(sc);
3898 
3899 	/*
3900 	 * If the age is zero then flush, otherwise set all the expiry times to
3901 	 * age for the interface
3902 	 */
3903 	if (age == 0)
3904 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3905 	else {
3906 		CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3907 			/* Cap the expiry time to 'age' */
3908 			if (brt->brt_ifp == ifp &&
3909 			    brt->brt_expire > time_uptime + age &&
3910 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3911 				brt->brt_expire = time_uptime + age;
3912 		}
3913 	}
3914 	BRIDGE_RT_UNLOCK(sc);
3915 	CURVNET_RESTORE();
3916 }
3917 
3918 /*
3919  * bridge_state_change:
3920  *
3921  *	Callback from the bridgestp code when a port changes states.
3922  */
3923 static void
bridge_state_change(struct ifnet * ifp,int state)3924 bridge_state_change(struct ifnet *ifp, int state)
3925 {
3926 	struct bridge_iflist *bif = ifp->if_bridge;
3927 	struct bridge_softc *sc = bif->bif_sc;
3928 	static const char *stpstates[] = {
3929 		"disabled",
3930 		"listening",
3931 		"learning",
3932 		"forwarding",
3933 		"blocking",
3934 		"discarding"
3935 	};
3936 
3937 	CURVNET_SET(ifp->if_vnet);
3938 	if (V_log_stp)
3939 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3940 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3941 	CURVNET_RESTORE();
3942 }
3943 
3944 /*
3945  * Send bridge packets through pfil if they are one of the types pfil can deal
3946  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3947  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3948  * that interface.
3949  */
3950 static int
bridge_pfil(struct mbuf ** mp,struct ifnet * bifp,struct ifnet * ifp,int dir)3951 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3952 {
3953 	int snap, error, i;
3954 	struct ether_header *eh1, eh2;
3955 	struct llc llc1;
3956 	u_int16_t ether_type;
3957 	pfil_return_t rv;
3958 #ifdef INET
3959 	struct ip *ip = NULL;
3960 	int hlen = 0;
3961 #endif
3962 
3963 	snap = 0;
3964 	error = -1;	/* Default error if not error == 0 */
3965 
3966 #if 0
3967 	/* we may return with the IP fields swapped, ensure its not shared */
3968 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3969 #endif
3970 
3971 	if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3972 		return (0); /* filtering is disabled */
3973 
3974 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3975 	if ((*mp)->m_len < i) {
3976 	    *mp = m_pullup(*mp, i);
3977 	    if (*mp == NULL) {
3978 		printf("%s: m_pullup failed\n", __func__);
3979 		return (-1);
3980 	    }
3981 	}
3982 
3983 	eh1 = mtod(*mp, struct ether_header *);
3984 	ether_type = ntohs(eh1->ether_type);
3985 
3986 	/*
3987 	 * Check for SNAP/LLC.
3988 	 */
3989 	if (ether_type < ETHERMTU) {
3990 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3991 
3992 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3993 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3994 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3995 		    llc2->llc_control == LLC_UI) {
3996 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3997 			snap = 1;
3998 		}
3999 	}
4000 
4001 	/*
4002 	 * If we're trying to filter bridge traffic, only look at traffic for
4003 	 * protocols available in the kernel (IPv4 and/or IPv6) to avoid
4004 	 * passing traffic for an unsupported protocol to the filter.  This is
4005 	 * lame since if we really wanted, say, an AppleTalk filter, we are
4006 	 * hosed, but of course we don't have an AppleTalk filter to begin
4007 	 * with.  (Note that since pfil doesn't understand ARP it will pass
4008 	 * *ALL* ARP traffic.)
4009 	 */
4010 	switch (ether_type) {
4011 #ifdef INET
4012 		case ETHERTYPE_ARP:
4013 		case ETHERTYPE_REVARP:
4014 			if (V_pfil_ipfw_arp == 0)
4015 				return (0); /* Automatically pass */
4016 
4017 			/* FALLTHROUGH */
4018 		case ETHERTYPE_IP:
4019 #endif
4020 #ifdef INET6
4021 		case ETHERTYPE_IPV6:
4022 #endif /* INET6 */
4023 			break;
4024 
4025 		default:
4026 			/*
4027 			 * We get here if the packet isn't from a supported
4028 			 * protocol.  Check to see if the user wants to pass
4029 			 * non-IP packets, these will not be checked by pfil(9)
4030 			 * and passed unconditionally so the default is to
4031 			 * drop.
4032 			 */
4033 			if (V_pfil_onlyip)
4034 				goto bad;
4035 	}
4036 
4037 	/* Run the packet through pfil before stripping link headers */
4038 	if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
4039 	    dir == PFIL_OUT && ifp != NULL) {
4040 		switch (pfil_mbuf_out(V_link_pfil_head, mp, ifp, NULL)) {
4041 		case PFIL_DROPPED:
4042 			return (EACCES);
4043 		case PFIL_CONSUMED:
4044 			return (0);
4045 		}
4046 	}
4047 
4048 	/* Strip off the Ethernet header and keep a copy. */
4049 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
4050 	m_adj(*mp, ETHER_HDR_LEN);
4051 
4052 	/* Strip off snap header, if present */
4053 	if (snap) {
4054 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
4055 		m_adj(*mp, sizeof(struct llc));
4056 	}
4057 
4058 	/*
4059 	 * Check the IP header for alignment and errors
4060 	 */
4061 	if (dir == PFIL_IN) {
4062 		switch (ether_type) {
4063 #ifdef INET
4064 			case ETHERTYPE_IP:
4065 				error = bridge_ip_checkbasic(mp);
4066 				break;
4067 #endif
4068 #ifdef INET6
4069 			case ETHERTYPE_IPV6:
4070 				error = bridge_ip6_checkbasic(mp);
4071 				break;
4072 #endif /* INET6 */
4073 			default:
4074 				error = 0;
4075 		}
4076 		if (error)
4077 			goto bad;
4078 	}
4079 
4080 	error = 0;
4081 
4082 	/*
4083 	 * Run the packet through pfil
4084 	 */
4085 	rv = PFIL_PASS;
4086 	switch (ether_type) {
4087 #ifdef INET
4088 	case ETHERTYPE_IP:
4089 		/*
4090 		 * Run pfil on the member interface and the bridge, both can
4091 		 * be skipped by clearing pfil_member or pfil_bridge.
4092 		 *
4093 		 * Keep the order:
4094 		 *   in_if -> bridge_if -> out_if
4095 		 */
4096 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
4097 		    pfil_mbuf_out(V_inet_pfil_head, mp, bifp, NULL)) !=
4098 		    PFIL_PASS)
4099 			break;
4100 
4101 		if (V_pfil_member && ifp != NULL) {
4102 			rv = (dir == PFIL_OUT) ?
4103 			    pfil_mbuf_out(V_inet_pfil_head, mp, ifp, NULL) :
4104 			    pfil_mbuf_in(V_inet_pfil_head, mp, ifp, NULL);
4105 			if (rv != PFIL_PASS)
4106 				break;
4107 		}
4108 
4109 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
4110 		    pfil_mbuf_in(V_inet_pfil_head, mp, bifp, NULL)) !=
4111 		    PFIL_PASS)
4112 			break;
4113 
4114 		/* check if we need to fragment the packet */
4115 		/* bridge_fragment generates a mbuf chain of packets */
4116 		/* that already include eth headers */
4117 		if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
4118 			i = (*mp)->m_pkthdr.len;
4119 			if (i > ifp->if_mtu) {
4120 				error = bridge_fragment(ifp, mp, &eh2, snap,
4121 					    &llc1);
4122 				return (error);
4123 			}
4124 		}
4125 
4126 		/* Recalculate the ip checksum. */
4127 		ip = mtod(*mp, struct ip *);
4128 		hlen = ip->ip_hl << 2;
4129 		if (hlen < sizeof(struct ip))
4130 			goto bad;
4131 		if (hlen > (*mp)->m_len) {
4132 			if ((*mp = m_pullup(*mp, hlen)) == NULL)
4133 				goto bad;
4134 			ip = mtod(*mp, struct ip *);
4135 			if (ip == NULL)
4136 				goto bad;
4137 		}
4138 		ip->ip_sum = 0;
4139 		if (hlen == sizeof(struct ip))
4140 			ip->ip_sum = in_cksum_hdr(ip);
4141 		else
4142 			ip->ip_sum = in_cksum(*mp, hlen);
4143 
4144 		break;
4145 #endif /* INET */
4146 #ifdef INET6
4147 	case ETHERTYPE_IPV6:
4148 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
4149 		    pfil_mbuf_out(V_inet6_pfil_head, mp, bifp, NULL)) !=
4150 		    PFIL_PASS)
4151 			break;
4152 
4153 		if (V_pfil_member && ifp != NULL) {
4154 			rv = (dir == PFIL_OUT) ?
4155 			    pfil_mbuf_out(V_inet6_pfil_head, mp, ifp, NULL) :
4156 			    pfil_mbuf_in(V_inet6_pfil_head, mp, ifp, NULL);
4157 			if (rv != PFIL_PASS)
4158 				break;
4159 		}
4160 
4161 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
4162 		    pfil_mbuf_in(V_inet6_pfil_head, mp, bifp, NULL)) !=
4163 		    PFIL_PASS)
4164 			break;
4165 		break;
4166 #endif
4167 	}
4168 
4169 	switch (rv) {
4170 	case PFIL_CONSUMED:
4171 		return (0);
4172 	case PFIL_DROPPED:
4173 		return (EACCES);
4174 	default:
4175 		break;
4176 	}
4177 
4178 	error = -1;
4179 
4180 	/*
4181 	 * Finally, put everything back the way it was and return
4182 	 */
4183 	if (snap) {
4184 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
4185 		if (*mp == NULL)
4186 			return (error);
4187 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
4188 	}
4189 
4190 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
4191 	if (*mp == NULL)
4192 		return (error);
4193 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
4194 
4195 	return (0);
4196 
4197 bad:
4198 	m_freem(*mp);
4199 	*mp = NULL;
4200 	return (error);
4201 }
4202 
4203 #ifdef INET
4204 /*
4205  * Perform basic checks on header size since
4206  * pfil assumes ip_input has already processed
4207  * it for it.  Cut-and-pasted from ip_input.c.
4208  * Given how simple the IPv6 version is,
4209  * does the IPv4 version really need to be
4210  * this complicated?
4211  *
4212  * XXX Should we update ipstat here, or not?
4213  * XXX Right now we update ipstat but not
4214  * XXX csum_counter.
4215  */
4216 static int
bridge_ip_checkbasic(struct mbuf ** mp)4217 bridge_ip_checkbasic(struct mbuf **mp)
4218 {
4219 	struct mbuf *m = *mp;
4220 	struct ip *ip;
4221 	int len, hlen;
4222 	u_short sum;
4223 
4224 	if (*mp == NULL)
4225 		return (-1);
4226 
4227 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
4228 		if ((m = m_copyup(m, sizeof(struct ip),
4229 			(max_linkhdr + 3) & ~3)) == NULL) {
4230 			/* XXXJRT new stat, please */
4231 			KMOD_IPSTAT_INC(ips_toosmall);
4232 			goto bad;
4233 		}
4234 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
4235 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
4236 			KMOD_IPSTAT_INC(ips_toosmall);
4237 			goto bad;
4238 		}
4239 	}
4240 	ip = mtod(m, struct ip *);
4241 	if (ip == NULL) goto bad;
4242 
4243 	if (ip->ip_v != IPVERSION) {
4244 		KMOD_IPSTAT_INC(ips_badvers);
4245 		goto bad;
4246 	}
4247 	hlen = ip->ip_hl << 2;
4248 	if (hlen < sizeof(struct ip)) { /* minimum header length */
4249 		KMOD_IPSTAT_INC(ips_badhlen);
4250 		goto bad;
4251 	}
4252 	if (hlen > m->m_len) {
4253 		if ((m = m_pullup(m, hlen)) == NULL) {
4254 			KMOD_IPSTAT_INC(ips_badhlen);
4255 			goto bad;
4256 		}
4257 		ip = mtod(m, struct ip *);
4258 		if (ip == NULL) goto bad;
4259 	}
4260 
4261 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
4262 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
4263 	} else {
4264 		if (hlen == sizeof(struct ip)) {
4265 			sum = in_cksum_hdr(ip);
4266 		} else {
4267 			sum = in_cksum(m, hlen);
4268 		}
4269 	}
4270 	if (sum) {
4271 		KMOD_IPSTAT_INC(ips_badsum);
4272 		goto bad;
4273 	}
4274 
4275 	/* Retrieve the packet length. */
4276 	len = ntohs(ip->ip_len);
4277 
4278 	/*
4279 	 * Check for additional length bogosity
4280 	 */
4281 	if (len < hlen) {
4282 		KMOD_IPSTAT_INC(ips_badlen);
4283 		goto bad;
4284 	}
4285 
4286 	/*
4287 	 * Check that the amount of data in the buffers
4288 	 * is as at least much as the IP header would have us expect.
4289 	 * Drop packet if shorter than we expect.
4290 	 */
4291 	if (m->m_pkthdr.len < len) {
4292 		KMOD_IPSTAT_INC(ips_tooshort);
4293 		goto bad;
4294 	}
4295 
4296 	/* Checks out, proceed */
4297 	*mp = m;
4298 	return (0);
4299 
4300 bad:
4301 	*mp = m;
4302 	return (-1);
4303 }
4304 #endif /* INET */
4305 
4306 #ifdef INET6
4307 /*
4308  * Same as above, but for IPv6.
4309  * Cut-and-pasted from ip6_input.c.
4310  * XXX Should we update ip6stat, or not?
4311  */
4312 static int
bridge_ip6_checkbasic(struct mbuf ** mp)4313 bridge_ip6_checkbasic(struct mbuf **mp)
4314 {
4315 	struct mbuf *m = *mp;
4316 	struct ip6_hdr *ip6;
4317 
4318 	/*
4319 	 * If the IPv6 header is not aligned, slurp it up into a new
4320 	 * mbuf with space for link headers, in the event we forward
4321 	 * it.  Otherwise, if it is aligned, make sure the entire base
4322 	 * IPv6 header is in the first mbuf of the chain.
4323 	 */
4324 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
4325 		struct ifnet *inifp = m->m_pkthdr.rcvif;
4326 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
4327 			    (max_linkhdr + 3) & ~3)) == NULL) {
4328 			/* XXXJRT new stat, please */
4329 			IP6STAT_INC(ip6s_toosmall);
4330 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
4331 			goto bad;
4332 		}
4333 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
4334 		struct ifnet *inifp = m->m_pkthdr.rcvif;
4335 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
4336 			IP6STAT_INC(ip6s_toosmall);
4337 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
4338 			goto bad;
4339 		}
4340 	}
4341 
4342 	ip6 = mtod(m, struct ip6_hdr *);
4343 
4344 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
4345 		IP6STAT_INC(ip6s_badvers);
4346 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
4347 		goto bad;
4348 	}
4349 
4350 	/* Checks out, proceed */
4351 	*mp = m;
4352 	return (0);
4353 
4354 bad:
4355 	*mp = m;
4356 	return (-1);
4357 }
4358 #endif /* INET6 */
4359 
4360 #ifdef INET
4361 /*
4362  * bridge_fragment:
4363  *
4364  *	Fragment mbuf chain in multiple packets and prepend ethernet header.
4365  */
4366 static int
bridge_fragment(struct ifnet * ifp,struct mbuf ** mp,struct ether_header * eh,int snap,struct llc * llc)4367 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
4368     int snap, struct llc *llc)
4369 {
4370 	struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
4371 	struct ip *ip;
4372 	int error = -1;
4373 
4374 	if (m->m_len < sizeof(struct ip) &&
4375 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
4376 		goto dropit;
4377 	ip = mtod(m, struct ip *);
4378 
4379 	m->m_pkthdr.csum_flags |= CSUM_IP;
4380 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
4381 	if (error)
4382 		goto dropit;
4383 
4384 	/*
4385 	 * Walk the chain and re-add the Ethernet header for
4386 	 * each mbuf packet.
4387 	 */
4388 	for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
4389 		nextpkt = mcur->m_nextpkt;
4390 		mcur->m_nextpkt = NULL;
4391 		if (snap) {
4392 			M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
4393 			if (mcur == NULL) {
4394 				error = ENOBUFS;
4395 				if (mprev != NULL)
4396 					mprev->m_nextpkt = nextpkt;
4397 				goto dropit;
4398 			}
4399 			bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
4400 		}
4401 
4402 		M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
4403 		if (mcur == NULL) {
4404 			error = ENOBUFS;
4405 			if (mprev != NULL)
4406 				mprev->m_nextpkt = nextpkt;
4407 			goto dropit;
4408 		}
4409 		bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
4410 
4411 		/*
4412 		 * The previous two M_PREPEND could have inserted one or two
4413 		 * mbufs in front so we have to update the previous packet's
4414 		 * m_nextpkt.
4415 		 */
4416 		mcur->m_nextpkt = nextpkt;
4417 		if (mprev != NULL)
4418 			mprev->m_nextpkt = mcur;
4419 		else {
4420 			/* The first mbuf in the original chain needs to be
4421 			 * updated. */
4422 			*mp = mcur;
4423 		}
4424 		mprev = mcur;
4425 	}
4426 
4427 	KMOD_IPSTAT_INC(ips_fragmented);
4428 	return (error);
4429 
4430 dropit:
4431 	for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
4432 		m = mcur->m_nextpkt;
4433 		m_freem(mcur);
4434 	}
4435 	return (error);
4436 }
4437 #endif /* INET */
4438 
4439 static void
bridge_linkstate(struct ifnet * ifp)4440 bridge_linkstate(struct ifnet *ifp)
4441 {
4442 	struct bridge_softc *sc = NULL;
4443 	struct bridge_iflist *bif;
4444 	struct epoch_tracker et;
4445 
4446 	NET_EPOCH_ENTER(et);
4447 
4448 	bif = ifp->if_bridge;
4449 	if (bif)
4450 		sc = bif->bif_sc;
4451 
4452 	if (sc != NULL) {
4453 		bridge_linkcheck(sc);
4454 		bstp_linkstate(&bif->bif_stp);
4455 	}
4456 
4457 	NET_EPOCH_EXIT(et);
4458 }
4459 
4460 static void
bridge_linkcheck(struct bridge_softc * sc)4461 bridge_linkcheck(struct bridge_softc *sc)
4462 {
4463 	struct bridge_iflist *bif;
4464 	int new_link, hasls;
4465 
4466 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
4467 
4468 	new_link = LINK_STATE_DOWN;
4469 	hasls = 0;
4470 	/* Our link is considered up if at least one of our ports is active */
4471 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
4472 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
4473 			hasls++;
4474 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
4475 			new_link = LINK_STATE_UP;
4476 			break;
4477 		}
4478 	}
4479 	if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
4480 		/* If no interfaces support link-state then we default to up */
4481 		new_link = LINK_STATE_UP;
4482 	}
4483 	if_link_state_change(sc->sc_ifp, new_link);
4484 }
4485