sys/net/if_bridge.c

   1 /*      $NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $       */
   2
   3 /*
   4  * Copyright 2001 Wasabi Systems, Inc.
   5  * All rights reserved.
   6  *
   7  * Written by Jason R. Thorpe for Wasabi Systems, Inc.
   8  *
   9  * Redistribution and use in source and binary forms, with or without
  10  * modification, are permitted provided that the following conditions
  11  * are met:
  12  * 1. Redistributions of source code must retain the above copyright
  13  *    notice, this list of conditions and the following disclaimer.
  14  * 2. Redistributions in binary form must reproduce the above copyright
  15  *    notice, this list of conditions and the following disclaimer in the
  16  *    documentation and/or other materials provided with the distribution.
  17  * 3. All advertising materials mentioning features or use of this software
  18  *    must display the following acknowledgement:
  19  *      This product includes software developed for the NetBSD Project by
  20  *      Wasabi Systems, Inc.
  21  * 4. The name of Wasabi Systems, Inc. may not be used to endorse
  22  *    or promote products derived from this software without specific prior
  23  *    written permission.
  24  *
  25  * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
  26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
  27  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
  28  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL WASABI SYSTEMS, INC
  29  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
  30  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
  31  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
  32  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
  33  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
  34  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
  35  * POSSIBILITY OF SUCH DAMAGE.
  36  */
  37
  38 /*
  39  * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
  40  * All rights reserved.
  41  *
  42  * Redistribution and use in source and binary forms, with or without
  43  * modification, are permitted provided that the following conditions
  44  * are met:
  45  * 1. Redistributions of source code must retain the above copyright
  46  *    notice, this list of conditions and the following disclaimer.
  47  * 2. Redistributions in binary form must reproduce the above copyright
  48  *    notice, this list of conditions and the following disclaimer in the
  49  *    documentation and/or other materials provided with the distribution.
  50  *
  51  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
  52  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
  53  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
  54  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
  55  * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
  56  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
  57  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  58  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
  59  * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
  60  * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
  61  * POSSIBILITY OF SUCH DAMAGE.
  62  *
  63  * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
  64  */
  65
  66 /*
  67  * Network interface bridge support.
  68  *
  69  * TODO:
  70  *
  71  *      - Currently only supports Ethernet-like interfaces (Ethernet,
  72  *        802.11, VLANs on Ethernet, etc.)  Figure out a nice way
  73  *        to bridge other types of interfaces (FDDI-FDDI, and maybe
  74  *        consider heterogenous bridges).
  75  */
  76
  77 #include <sys/cdefs.h>
  78 __FBSDID("$FreeBSD$");
  79
  80 #include "opt_inet.h"
  81 #include "opt_inet6.h"
  82
  83 #include <sys/param.h>
  84 #include <sys/mbuf.h>
  85 #include <sys/malloc.h>
  86 #include <sys/protosw.h>
  87 #include <sys/systm.h>
  88 #include <sys/jail.h>
  89 #include <sys/time.h>
  90 #include <sys/socket.h> /* for net/if.h */
  91 #include <sys/sockio.h>
  92 #include <sys/ctype.h>  /* string functions */
  93 #include <sys/kernel.h>
  94 #include <sys/random.h>
  95 #include <sys/syslog.h>
  96 #include <sys/sysctl.h>
  97 #include <vm/uma.h>
  98 #include <sys/module.h>
  99 #include <sys/priv.h>
 100 #include <sys/proc.h>
 101 #include <sys/lock.h>
 102 #include <sys/mutex.h>
 103
 104 #include <net/bpf.h>
 105 #include <net/if.h>
 106 #include <net/if_clone.h>
 107 #include <net/if_dl.h>
 108 #include <net/if_types.h>
 109 #include <net/if_var.h>
 110 #include <net/pfil.h>
 111 #include <net/vnet.h>
 112
 113 #include <netinet/in.h> /* for struct arpcom */
 114 #include <netinet/in_systm.h>
 115 #include <netinet/in_var.h>
 116 #include <netinet/ip.h>
 117 #include <netinet/ip_var.h>
 118 #ifdef INET6
 119 #include <netinet/ip6.h>
 120 #include <netinet6/ip6_var.h>
 121 #endif
 122 #if defined(INET) || defined(INET6)
 123 #include <netinet/ip_carp.h>
 124 #endif
 125 #include <machine/in_cksum.h>
 126 #include <netinet/if_ether.h> /* for struct arpcom */
 127 #include <net/bridgestp.h>
 128 #include <net/if_bridgevar.h>
 129 #include <net/if_llc.h>
 130 #include <net/if_vlan_var.h>
 131
 132 #include <net/route.h>
 133
 134 /*
 135  * Size of the route hash table.  Must be a power of two.
 136  */
 137 #ifndef BRIDGE_RTHASH_SIZE
 138 #define BRIDGE_RTHASH_SIZE              1024
 139 #endif
 140
 141 #define BRIDGE_RTHASH_MASK              (BRIDGE_RTHASH_SIZE - 1)
 142
 143 /*
 144  * Default maximum number of addresses to cache.
 145  */
 146 #ifndef BRIDGE_RTABLE_MAX
 147 #define BRIDGE_RTABLE_MAX               2000
 148 #endif
 149
 150 /*
 151  * Timeout (in seconds) for entries learned dynamically.
 152  */
 153 #ifndef BRIDGE_RTABLE_TIMEOUT
 154 #define BRIDGE_RTABLE_TIMEOUT           (20 * 60)       /* same as ARP */
 155 #endif
 156
 157 /*
 158  * Number of seconds between walks of the route list.
 159  */
 160 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
 161 #define BRIDGE_RTABLE_PRUNE_PERIOD      (5 * 60)
 162 #endif
 163
 164 /*
 165  * List of capabilities to possibly mask on the member interface.
 166  */
 167 #define BRIDGE_IFCAPS_MASK              (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
 168
 169 /*
 170  * List of capabilities to strip
 171  */
 172 #define BRIDGE_IFCAPS_STRIP             IFCAP_LRO
 173
 174 /*
 175  * Bridge interface list entry.
 176  */
 177 struct bridge_iflist {
 178         LIST_ENTRY(bridge_iflist) bif_next;
 179         struct ifnet            *bif_ifp;       /* member if */
 180         struct bstp_port        bif_stp;        /* STP state */
 181         uint32_t                bif_flags;      /* member if flags */
 182         int                     bif_savedcaps;  /* saved capabilities */
 183         uint32_t                bif_addrmax;    /* max # of addresses */
 184         uint32_t                bif_addrcnt;    /* cur. # of addresses */
 185         uint32_t                bif_addrexceeded;/* # of address violations */
 186 };
 187
 188 /*
 189  * Bridge route node.
 190  */
 191 struct bridge_rtnode {
 192         LIST_ENTRY(bridge_rtnode) brt_hash;     /* hash table linkage */
 193         LIST_ENTRY(bridge_rtnode) brt_list;     /* list linkage */
 194         struct bridge_iflist    *brt_dst;       /* destination if */
 195         unsigned long           brt_expire;     /* expiration time */
 196         uint8_t                 brt_flags;      /* address flags */
 197         uint8_t                 brt_addr[ETHER_ADDR_LEN];
 198         uint16_t                brt_vlan;       /* vlan id */
 199 };
 200 #define brt_ifp                 brt_dst->bif_ifp
 201
 202 /*
 203  * Software state for each bridge.
 204  */
 205 struct bridge_softc {
 206         struct ifnet            *sc_ifp;        /* make this an interface */
 207         LIST_ENTRY(bridge_softc) sc_list;
 208         struct mtx              sc_mtx;
 209         struct cv               sc_cv;
 210         uint32_t                sc_brtmax;      /* max # of addresses */
 211         uint32_t                sc_brtcnt;      /* cur. # of addresses */
 212         uint32_t                sc_brttimeout;  /* rt timeout in seconds */
 213         struct callout          sc_brcallout;   /* bridge callout */
 214         uint32_t                sc_iflist_ref;  /* refcount for sc_iflist */
 215         uint32_t                sc_iflist_xcnt; /* refcount for sc_iflist */
 216         LIST_HEAD(, bridge_iflist) sc_iflist;   /* member interface list */
 217         LIST_HEAD(, bridge_rtnode) *sc_rthash;  /* our forwarding table */
 218         LIST_HEAD(, bridge_rtnode) sc_rtlist;   /* list version of above */
 219         uint32_t                sc_rthash_key;  /* key for hash */
 220         LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
 221         struct bstp_state       sc_stp;         /* STP state */
 222         uint32_t                sc_brtexceeded; /* # of cache drops */
 223         struct ifnet            *sc_ifaddr;     /* member mac copied from */
 224         u_char                  sc_defaddr[6];  /* Default MAC address */
 225 };
 226
 227 static struct mtx       bridge_list_mtx;
 228 eventhandler_tag        bridge_detach_cookie = NULL;
 229
 230 int     bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
 231
 232 uma_zone_t bridge_rtnode_zone;
 233
 234 static int      bridge_clone_create(struct if_clone *, int, caddr_t);
 235 static void     bridge_clone_destroy(struct ifnet *);
 236
 237 static int      bridge_ioctl(struct ifnet *, u_long, caddr_t);
 238 static void     bridge_mutecaps(struct bridge_softc *);
 239 static void     bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
 240                     int);
 241 static void     bridge_ifdetach(void *arg __unused, struct ifnet *);
 242 static void     bridge_init(void *);
 243 static void     bridge_dummynet(struct mbuf *, struct ifnet *);
 244 static void     bridge_stop(struct ifnet *, int);
 245 static int      bridge_transmit(struct ifnet *, struct mbuf *);
 246 static void     bridge_qflush(struct ifnet *);
 247 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
 248 static int      bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
 249                     struct rtentry *);
 250 static int      bridge_enqueue(struct bridge_softc *, struct ifnet *,
 251                     struct mbuf *);
 252 static void     bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
 253
 254 static void     bridge_forward(struct bridge_softc *, struct bridge_iflist *,
 255                     struct mbuf *m);
 256
 257 static void     bridge_timer(void *);
 258
 259 static void     bridge_broadcast(struct bridge_softc *, struct ifnet *,
 260                     struct mbuf *, int);
 261 static void     bridge_span(struct bridge_softc *, struct mbuf *);
 262
 263 static int      bridge_rtupdate(struct bridge_softc *, const uint8_t *,
 264                     uint16_t, struct bridge_iflist *, int, uint8_t);
 265 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
 266                     uint16_t);
 267 static void     bridge_rttrim(struct bridge_softc *);
 268 static void     bridge_rtage(struct bridge_softc *);
 269 static void     bridge_rtflush(struct bridge_softc *, int);
 270 static int      bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
 271                     uint16_t);
 272
 273 static void     bridge_rtable_init(struct bridge_softc *);
 274 static void     bridge_rtable_fini(struct bridge_softc *);
 275
 276 static int      bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
 277 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
 278                     const uint8_t *, uint16_t);
 279 static int      bridge_rtnode_insert(struct bridge_softc *,
 280                     struct bridge_rtnode *);
 281 static void     bridge_rtnode_destroy(struct bridge_softc *,
 282                     struct bridge_rtnode *);
 283 static void     bridge_rtable_expire(struct ifnet *, int);
 284 static void     bridge_state_change(struct ifnet *, int);
 285
 286 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
 287                     const char *name);
 288 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
 289                     struct ifnet *ifp);
 290 static void     bridge_delete_member(struct bridge_softc *,
 291                     struct bridge_iflist *, int);
 292 static void     bridge_delete_span(struct bridge_softc *,
 293                     struct bridge_iflist *);
 294
 295 static int      bridge_ioctl_add(struct bridge_softc *, void *);
 296 static int      bridge_ioctl_del(struct bridge_softc *, void *);
 297 static int      bridge_ioctl_gifflags(struct bridge_softc *, void *);
 298 static int      bridge_ioctl_sifflags(struct bridge_softc *, void *);
 299 static int      bridge_ioctl_scache(struct bridge_softc *, void *);
 300 static int      bridge_ioctl_gcache(struct bridge_softc *, void *);
 301 static int      bridge_ioctl_gifs(struct bridge_softc *, void *);
 302 static int      bridge_ioctl_rts(struct bridge_softc *, void *);
 303 static int      bridge_ioctl_saddr(struct bridge_softc *, void *);
 304 static int      bridge_ioctl_sto(struct bridge_softc *, void *);
 305 static int      bridge_ioctl_gto(struct bridge_softc *, void *);
 306 static int      bridge_ioctl_daddr(struct bridge_softc *, void *);
 307 static int      bridge_ioctl_flush(struct bridge_softc *, void *);
 308 static int      bridge_ioctl_gpri(struct bridge_softc *, void *);
 309 static int      bridge_ioctl_spri(struct bridge_softc *, void *);
 310 static int      bridge_ioctl_ght(struct bridge_softc *, void *);
 311 static int      bridge_ioctl_sht(struct bridge_softc *, void *);
 312 static int      bridge_ioctl_gfd(struct bridge_softc *, void *);
 313 static int      bridge_ioctl_sfd(struct bridge_softc *, void *);
 314 static int      bridge_ioctl_gma(struct bridge_softc *, void *);
 315 static int      bridge_ioctl_sma(struct bridge_softc *, void *);
 316 static int      bridge_ioctl_sifprio(struct bridge_softc *, void *);
 317 static int      bridge_ioctl_sifcost(struct bridge_softc *, void *);
 318 static int      bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
 319 static int      bridge_ioctl_addspan(struct bridge_softc *, void *);
 320 static int      bridge_ioctl_delspan(struct bridge_softc *, void *);
 321 static int      bridge_ioctl_gbparam(struct bridge_softc *, void *);
 322 static int      bridge_ioctl_grte(struct bridge_softc *, void *);
 323 static int      bridge_ioctl_gifsstp(struct bridge_softc *, void *);
 324 static int      bridge_ioctl_sproto(struct bridge_softc *, void *);
 325 static int      bridge_ioctl_stxhc(struct bridge_softc *, void *);
 326 static int      bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
 327                     int);
 328 static int      bridge_ip_checkbasic(struct mbuf **mp);
 329 #ifdef INET6
 330 static int      bridge_ip6_checkbasic(struct mbuf **mp);
 331 #endif /* INET6 */
 332 static int      bridge_fragment(struct ifnet *, struct mbuf *,
 333                     struct ether_header *, int, struct llc *);
 334 static void     bridge_linkstate(struct ifnet *ifp);
 335 static void     bridge_linkcheck(struct bridge_softc *sc);
 336
 337 extern void (*bridge_linkstate_p)(struct ifnet *ifp);
 338
 339 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
 340 #define VLANTAGOF(_m)   \
 341     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
 342
 343 static struct bstp_cb_ops bridge_ops = {
 344         .bcb_state = bridge_state_change,
 345         .bcb_rtage = bridge_rtable_expire
 346 };
 347
 348 SYSCTL_DECL(_net_link);
 349 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
 350
 351 static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */
 352 static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */
 353 static int pfil_member = 1; /* run pfil hooks on the member interface */
 354 static int pfil_ipfw = 0;   /* layer2 filter with ipfw */
 355 static int pfil_ipfw_arp = 0;   /* layer2 filter with ipfw */
 356 static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for
 357                                    locally destined packets */
 358 static int log_stp   = 0;   /* log STP state changes */
 359 static int bridge_inherit_mac = 0;   /* share MAC with first bridge member */
 360 TUNABLE_INT("net.link.bridge.pfil_onlyip", &pfil_onlyip);
 361 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
 362     &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
 363 TUNABLE_INT("net.link.bridge.ipfw_arp", &pfil_ipfw_arp);
 364 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
 365     &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
 366 TUNABLE_INT("net.link.bridge.pfil_bridge", &pfil_bridge);
 367 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
 368     &pfil_bridge, 0, "Packet filter on the bridge interface");
 369 TUNABLE_INT("net.link.bridge.pfil_member", &pfil_member);
 370 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
 371     &pfil_member, 0, "Packet filter on the member interface");
 372 TUNABLE_INT("net.link.bridge.pfil_local_phys", &pfil_local_phys);
 373 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
 374     &pfil_local_phys, 0,
 375     "Packet filter on the physical interface for locally destined packets");
 376 TUNABLE_INT("net.link.bridge.log_stp", &log_stp);
 377 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
 378     &log_stp, 0, "Log STP state changes");
 379 TUNABLE_INT("net.link.bridge.inherit_mac", &bridge_inherit_mac);
 380 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, CTLFLAG_RW,
 381     &bridge_inherit_mac, 0,
 382     "Inherit MAC address from the first bridge member");
 383
 384 struct bridge_control {
 385         int     (*bc_func)(struct bridge_softc *, void *);
 386         int     bc_argsize;
 387         int     bc_flags;
 388 };
 389
 390 #define BC_F_COPYIN             0x01    /* copy arguments in */
 391 #define BC_F_COPYOUT            0x02    /* copy arguments out */
 392 #define BC_F_SUSER              0x04    /* do super-user check */
 393
 394 const struct bridge_control bridge_control_table[] = {
 395         { bridge_ioctl_add,             sizeof(struct ifbreq),
 396           BC_F_COPYIN|BC_F_SUSER },
 397         { bridge_ioctl_del,             sizeof(struct ifbreq),
 398           BC_F_COPYIN|BC_F_SUSER },
 399
 400         { bridge_ioctl_gifflags,        sizeof(struct ifbreq),
 401           BC_F_COPYIN|BC_F_COPYOUT },
 402         { bridge_ioctl_sifflags,        sizeof(struct ifbreq),
 403           BC_F_COPYIN|BC_F_SUSER },
 404
 405         { bridge_ioctl_scache,          sizeof(struct ifbrparam),
 406           BC_F_COPYIN|BC_F_SUSER },
 407         { bridge_ioctl_gcache,          sizeof(struct ifbrparam),
 408           BC_F_COPYOUT },
 409
 410         { bridge_ioctl_gifs,            sizeof(struct ifbifconf),
 411           BC_F_COPYIN|BC_F_COPYOUT },
 412         { bridge_ioctl_rts,             sizeof(struct ifbaconf),
 413           BC_F_COPYIN|BC_F_COPYOUT },
 414
 415         { bridge_ioctl_saddr,           sizeof(struct ifbareq),
 416           BC_F_COPYIN|BC_F_SUSER },
 417
 418         { bridge_ioctl_sto,             sizeof(struct ifbrparam),
 419           BC_F_COPYIN|BC_F_SUSER },
 420         { bridge_ioctl_gto,             sizeof(struct ifbrparam),
 421           BC_F_COPYOUT },
 422
 423         { bridge_ioctl_daddr,           sizeof(struct ifbareq),
 424           BC_F_COPYIN|BC_F_SUSER },
 425
 426         { bridge_ioctl_flush,           sizeof(struct ifbreq),
 427           BC_F_COPYIN|BC_F_SUSER },
 428
 429         { bridge_ioctl_gpri,            sizeof(struct ifbrparam),
 430           BC_F_COPYOUT },
 431         { bridge_ioctl_spri,            sizeof(struct ifbrparam),
 432           BC_F_COPYIN|BC_F_SUSER },
 433
 434         { bridge_ioctl_ght,             sizeof(struct ifbrparam),
 435           BC_F_COPYOUT },
 436         { bridge_ioctl_sht,             sizeof(struct ifbrparam),
 437           BC_F_COPYIN|BC_F_SUSER },
 438
 439         { bridge_ioctl_gfd,             sizeof(struct ifbrparam),
 440           BC_F_COPYOUT },
 441         { bridge_ioctl_sfd,             sizeof(struct ifbrparam),
 442           BC_F_COPYIN|BC_F_SUSER },
 443
 444         { bridge_ioctl_gma,             sizeof(struct ifbrparam),
 445           BC_F_COPYOUT },
 446         { bridge_ioctl_sma,             sizeof(struct ifbrparam),
 447           BC_F_COPYIN|BC_F_SUSER },
 448
 449         { bridge_ioctl_sifprio,         sizeof(struct ifbreq),
 450           BC_F_COPYIN|BC_F_SUSER },
 451
 452         { bridge_ioctl_sifcost,         sizeof(struct ifbreq),
 453           BC_F_COPYIN|BC_F_SUSER },
 454
 455         { bridge_ioctl_addspan,         sizeof(struct ifbreq),
 456           BC_F_COPYIN|BC_F_SUSER },
 457         { bridge_ioctl_delspan,         sizeof(struct ifbreq),
 458           BC_F_COPYIN|BC_F_SUSER },
 459
 460         { bridge_ioctl_gbparam,         sizeof(struct ifbropreq),
 461           BC_F_COPYOUT },
 462
 463         { bridge_ioctl_grte,            sizeof(struct ifbrparam),
 464           BC_F_COPYOUT },
 465
 466         { bridge_ioctl_gifsstp,         sizeof(struct ifbpstpconf),
 467           BC_F_COPYIN|BC_F_COPYOUT },
 468
 469         { bridge_ioctl_sproto,          sizeof(struct ifbrparam),
 470           BC_F_COPYIN|BC_F_SUSER },
 471
 472         { bridge_ioctl_stxhc,           sizeof(struct ifbrparam),
 473           BC_F_COPYIN|BC_F_SUSER },
 474
 475         { bridge_ioctl_sifmaxaddr,      sizeof(struct ifbreq),
 476           BC_F_COPYIN|BC_F_SUSER },
 477
 478 };
 479 const int bridge_control_table_size =
 480     sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
 481
 482 LIST_HEAD(, bridge_softc) bridge_list;
 483
 484 static struct if_clone *bridge_cloner;
 485 static const char bridge_name[] = "bridge";
 486
 487 static int
 488 bridge_modevent(module_t mod, int type, void *data)
 489 {
 490
 491         switch (type) {
 492         case MOD_LOAD:
 493                 mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
 494                 bridge_cloner = if_clone_simple(bridge_name,
 495                     bridge_clone_create, bridge_clone_destroy, 0);
 496                 bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
 497                     sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
 498                     UMA_ALIGN_PTR, 0);
 499                 LIST_INIT(&bridge_list);
 500                 bridge_input_p = bridge_input;
 501                 bridge_output_p = bridge_output;
 502                 bridge_dn_p = bridge_dummynet;
 503                 bridge_linkstate_p = bridge_linkstate;
 504                 bridge_detach_cookie = EVENTHANDLER_REGISTER(
 505                     ifnet_departure_event, bridge_ifdetach, NULL,
 506                     EVENTHANDLER_PRI_ANY);
 507                 break;
 508         case MOD_UNLOAD:
 509                 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
 510                     bridge_detach_cookie);
 511                 if_clone_detach(bridge_cloner);
 512                 uma_zdestroy(bridge_rtnode_zone);
 513                 bridge_input_p = NULL;
 514                 bridge_output_p = NULL;
 515                 bridge_dn_p = NULL;
 516                 bridge_linkstate_p = NULL;
 517                 mtx_destroy(&bridge_list_mtx);
 518                 break;
 519         default:
 520                 return (EOPNOTSUPP);
 521         }
 522         return (0);
 523 }
 524
 525 static moduledata_t bridge_mod = {
 526         "if_bridge",
 527         bridge_modevent,
 528         0
 529 };
 530
 531 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
 532 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
 533
 534 /*
 535  * handler for net.link.bridge.pfil_ipfw
 536  */
 537 static int
 538 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
 539 {
 540         int enable = pfil_ipfw;
 541         int error;
 542
 543         error = sysctl_handle_int(oidp, &enable, 0, req);
 544         enable = (enable) ? 1 : 0;
 545
 546         if (enable != pfil_ipfw) {
 547                 pfil_ipfw = enable;
 548
 549                 /*
 550                  * Disable pfil so that ipfw doesnt run twice, if the user
 551                  * really wants both then they can re-enable pfil_bridge and/or
 552                  * pfil_member. Also allow non-ip packets as ipfw can filter by
 553                  * layer2 type.
 554                  */
 555                 if (pfil_ipfw) {
 556                         pfil_onlyip = 0;
 557                         pfil_bridge = 0;
 558                         pfil_member = 0;
 559                 }
 560         }
 561
 562         return (error);
 563 }
 564 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
 565             &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
 566
 567 /*
 568  * bridge_clone_create:
 569  *
 570  *      Create a new bridge instance.
 571  */
 572 static int
 573 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
 574 {
 575         struct bridge_softc *sc, *sc2;
 576         struct ifnet *bifp, *ifp;
 577         int fb, retry;
 578         unsigned long hostid;
 579
 580         sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
 581         ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
 582         if (ifp == NULL) {
 583                 free(sc, M_DEVBUF);
 584                 return (ENOSPC);
 585         }
 586
 587         BRIDGE_LOCK_INIT(sc);
 588         sc->sc_brtmax = BRIDGE_RTABLE_MAX;
 589         sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
 590
 591         /* Initialize our routing table. */
 592         bridge_rtable_init(sc);
 593
 594         callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
 595
 596         LIST_INIT(&sc->sc_iflist);
 597         LIST_INIT(&sc->sc_spanlist);
 598
 599         ifp->if_softc = sc;
 600         if_initname(ifp, bridge_name, unit);
 601         ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
 602         ifp->if_ioctl = bridge_ioctl;
 603         ifp->if_transmit = bridge_transmit;
 604         ifp->if_qflush = bridge_qflush;
 605         ifp->if_init = bridge_init;
 606         ifp->if_type = IFT_BRIDGE;
 607
 608         /*
 609          * Generate an ethernet address with a locally administered address.
 610          *
 611          * Since we are using random ethernet addresses for the bridge, it is
 612          * possible that we might have address collisions, so make sure that
 613          * this hardware address isn't already in use on another bridge.
 614          * The first try uses the hostid and falls back to arc4rand().
 615          */
 616         fb = 0;
 617         getcredhostid(curthread->td_ucred, &hostid);
 618         for (retry = 1; retry != 0;) {
 619                 if (fb || hostid == 0) {
 620                         arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
 621                         sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
 622                         sc->sc_defaddr[0] |= 2; /* set the LAA bit */
 623                 } else {
 624                         sc->sc_defaddr[0] = 0x2;
 625                         sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
 626                         sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
 627                         sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
 628                         sc->sc_defaddr[4] =  hostid        & 0xff;
 629                         sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
 630                 }
 631
 632                 fb = 1;
 633                 retry = 0;
 634                 mtx_lock(&bridge_list_mtx);
 635                 LIST_FOREACH(sc2, &bridge_list, sc_list) {
 636                         bifp = sc2->sc_ifp;
 637                         if (memcmp(sc->sc_defaddr,
 638                             IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0)
 639                                 retry = 1;
 640                 }
 641                 mtx_unlock(&bridge_list_mtx);
 642         }
 643
 644         bstp_attach(&sc->sc_stp, &bridge_ops);
 645         ether_ifattach(ifp, sc->sc_defaddr);
 646         /* Now undo some of the damage... */
 647         ifp->if_baudrate = 0;
 648         ifp->if_type = IFT_BRIDGE;
 649
 650         mtx_lock(&bridge_list_mtx);
 651         LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
 652         mtx_unlock(&bridge_list_mtx);
 653
 654         return (0);
 655 }
 656
 657 /*
 658  * bridge_clone_destroy:
 659  *
 660  *      Destroy a bridge instance.
 661  */
 662 static void
 663 bridge_clone_destroy(struct ifnet *ifp)
 664 {
 665         struct bridge_softc *sc = ifp->if_softc;
 666         struct bridge_iflist *bif;
 667
 668         BRIDGE_LOCK(sc);
 669
 670         bridge_stop(ifp, 1);
 671         ifp->if_flags &= ~IFF_UP;
 672
 673         while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
 674                 bridge_delete_member(sc, bif, 0);
 675
 676         while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
 677                 bridge_delete_span(sc, bif);
 678         }
 679
 680         BRIDGE_UNLOCK(sc);
 681
 682         callout_drain(&sc->sc_brcallout);
 683
 684         mtx_lock(&bridge_list_mtx);
 685         LIST_REMOVE(sc, sc_list);
 686         mtx_unlock(&bridge_list_mtx);
 687
 688         bstp_detach(&sc->sc_stp);
 689         ether_ifdetach(ifp);
 690         if_free(ifp);
 691
 692         /* Tear down the routing table. */
 693         bridge_rtable_fini(sc);
 694
 695         BRIDGE_LOCK_DESTROY(sc);
 696         free(sc, M_DEVBUF);
 697 }
 698
 699 /*
 700  * bridge_ioctl:
 701  *
 702  *      Handle a control request from the operator.
 703  */
 704 static int
 705 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
 706 {
 707         struct bridge_softc *sc = ifp->if_softc;
 708         struct ifreq *ifr = (struct ifreq *)data;
 709         struct bridge_iflist *bif;
 710         struct thread *td = curthread;
 711         union {
 712                 struct ifbreq ifbreq;
 713                 struct ifbifconf ifbifconf;
 714                 struct ifbareq ifbareq;
 715                 struct ifbaconf ifbaconf;
 716                 struct ifbrparam ifbrparam;
 717                 struct ifbropreq ifbropreq;
 718         } args;
 719         struct ifdrv *ifd = (struct ifdrv *) data;
 720         const struct bridge_control *bc;
 721         int error = 0;
 722
 723         switch (cmd) {
 724
 725         case SIOCADDMULTI:
 726         case SIOCDELMULTI:
 727                 break;
 728
 729         case SIOCGDRVSPEC:
 730         case SIOCSDRVSPEC:
 731                 if (ifd->ifd_cmd >= bridge_control_table_size) {
 732                         error = EINVAL;
 733                         break;
 734                 }
 735                 bc = &bridge_control_table[ifd->ifd_cmd];
 736
 737                 if (cmd == SIOCGDRVSPEC &&
 738                     (bc->bc_flags & BC_F_COPYOUT) == 0) {
 739                         error = EINVAL;
 740                         break;
 741                 }
 742                 else if (cmd == SIOCSDRVSPEC &&
 743                     (bc->bc_flags & BC_F_COPYOUT) != 0) {
 744                         error = EINVAL;
 745                         break;
 746                 }
 747
 748                 if (bc->bc_flags & BC_F_SUSER) {
 749                         error = priv_check(td, PRIV_NET_BRIDGE);
 750                         if (error)
 751                                 break;
 752                 }
 753
 754                 if (ifd->ifd_len != bc->bc_argsize ||
 755                     ifd->ifd_len > sizeof(args)) {
 756                         error = EINVAL;
 757                         break;
 758                 }
 759
 760                 bzero(&args, sizeof(args));
 761                 if (bc->bc_flags & BC_F_COPYIN) {
 762                         error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
 763                         if (error)
 764                                 break;
 765                 }
 766
 767                 BRIDGE_LOCK(sc);
 768                 error = (*bc->bc_func)(sc, &args);
 769                 BRIDGE_UNLOCK(sc);
 770                 if (error)
 771                         break;
 772
 773                 if (bc->bc_flags & BC_F_COPYOUT)
 774                         error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
 775
 776                 break;
 777
 778         case SIOCSIFFLAGS:
 779                 if (!(ifp->if_flags & IFF_UP) &&
 780                     (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
 781                         /*
 782                          * If interface is marked down and it is running,
 783                          * then stop and disable it.
 784                          */
 785                         BRIDGE_LOCK(sc);
 786                         bridge_stop(ifp, 1);
 787                         BRIDGE_UNLOCK(sc);
 788                 } else if ((ifp->if_flags & IFF_UP) &&
 789                     !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
 790                         /*
 791                          * If interface is marked up and it is stopped, then
 792                          * start it.
 793                          */
 794                         (*ifp->if_init)(sc);
 795                 }
 796                 break;
 797
 798         case SIOCSIFMTU:
 799                 if (ifr->ifr_mtu < 576) {
 800                         error = EINVAL;
 801                         break;
 802                 }
 803                 if (LIST_EMPTY(&sc->sc_iflist)) {
 804                         sc->sc_ifp->if_mtu = ifr->ifr_mtu;
 805                         break;
 806                 }
 807                 BRIDGE_LOCK(sc);
 808                 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
 809                         if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
 810                                 log(LOG_NOTICE, "%s: invalid MTU: %lu(%s)"
 811                                     " != %d\n", sc->sc_ifp->if_xname,
 812                                     bif->bif_ifp->if_mtu,
 813                                     bif->bif_ifp->if_xname, ifr->ifr_mtu);
 814                                 error = EINVAL;
 815                                 break;
 816                         }
 817                 }
 818                 if (!error)
 819                         sc->sc_ifp->if_mtu = ifr->ifr_mtu;
 820                 BRIDGE_UNLOCK(sc);
 821                 break;
 822         default:
 823                 /*
 824                  * drop the lock as ether_ioctl() will call bridge_start() and
 825                  * cause the lock to be recursed.
 826                  */
 827                 error = ether_ioctl(ifp, cmd, data);
 828                 break;
 829         }
 830
 831         return (error);
 832 }
 833
 834 /*
 835  * bridge_mutecaps:
 836  *
 837  *      Clear or restore unwanted capabilities on the member interface
 838  */
 839 static void
 840 bridge_mutecaps(struct bridge_softc *sc)
 841 {
 842         struct bridge_iflist *bif;
 843         int enabled, mask;
 844
 845         /* Initial bitmask of capabilities to test */
 846         mask = BRIDGE_IFCAPS_MASK;
 847
 848         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
 849                 /* Every member must support it or its disabled */
 850                 mask &= bif->bif_savedcaps;
 851         }
 852
 853         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
 854                 enabled = bif->bif_ifp->if_capenable;
 855                 enabled &= ~BRIDGE_IFCAPS_STRIP;
 856                 /* strip off mask bits and enable them again if allowed */
 857                 enabled &= ~BRIDGE_IFCAPS_MASK;
 858                 enabled |= mask;
 859                 bridge_set_ifcap(sc, bif, enabled);
 860         }
 861
 862 }
 863
 864 static void
 865 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
 866 {
 867         struct ifnet *ifp = bif->bif_ifp;
 868         struct ifreq ifr;
 869         int error;
 870
 871         bzero(&ifr, sizeof(ifr));
 872         ifr.ifr_reqcap = set;
 873
 874         if (ifp->if_capenable != set) {
 875                 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
 876                 if (error)
 877                         if_printf(sc->sc_ifp,
 878                             "error setting interface capabilities on %s\n",
 879                             ifp->if_xname);
 880         }
 881 }
 882
 883 /*
 884  * bridge_lookup_member:
 885  *
 886  *      Lookup a bridge member interface.
 887  */
 888 static struct bridge_iflist *
 889 bridge_lookup_member(struct bridge_softc *sc, const char *name)
 890 {
 891         struct bridge_iflist *bif;
 892         struct ifnet *ifp;
 893
 894         BRIDGE_LOCK_ASSERT(sc);
 895
 896         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
 897                 ifp = bif->bif_ifp;
 898                 if (strcmp(ifp->if_xname, name) == 0)
 899                         return (bif);
 900         }
 901
 902         return (NULL);
 903 }
 904
 905 /*
 906  * bridge_lookup_member_if:
 907  *
 908  *      Lookup a bridge member interface by ifnet*.
 909  */
 910 static struct bridge_iflist *
 911 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
 912 {
 913         struct bridge_iflist *bif;
 914
 915         BRIDGE_LOCK_ASSERT(sc);
 916
 917         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
 918                 if (bif->bif_ifp == member_ifp)
 919                         return (bif);
 920         }
 921
 922         return (NULL);
 923 }
 924
 925 /*
 926  * bridge_delete_member:
 927  *
 928  *      Delete the specified member interface.
 929  */
 930 static void
 931 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
 932     int gone)
 933 {
 934         struct ifnet *ifs = bif->bif_ifp;
 935         struct ifnet *fif = NULL;
 936
 937         BRIDGE_LOCK_ASSERT(sc);
 938
 939         if (bif->bif_flags & IFBIF_STP)
 940                 bstp_disable(&bif->bif_stp);
 941
 942         ifs->if_bridge = NULL;
 943         BRIDGE_XLOCK(sc);
 944         LIST_REMOVE(bif, bif_next);
 945         BRIDGE_XDROP(sc);
 946
 947         /*
 948          * If removing the interface that gave the bridge its mac address, set
 949          * the mac address of the bridge to the address of the next member, or
 950          * to its default address if no members are left.
 951          */
 952         if (bridge_inherit_mac && sc->sc_ifaddr == ifs) {
 953                 if (LIST_EMPTY(&sc->sc_iflist)) {
 954                         bcopy(sc->sc_defaddr,
 955                             IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
 956                         sc->sc_ifaddr = NULL;
 957                 } else {
 958                         fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
 959                         bcopy(IF_LLADDR(fif),
 960                             IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
 961                         sc->sc_ifaddr = fif;
 962                 }
 963                 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
 964         }
 965
 966         bridge_linkcheck(sc);
 967         bridge_mutecaps(sc);    /* recalcuate now this interface is removed */
 968         bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
 969         KASSERT(bif->bif_addrcnt == 0,
 970             ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
 971
 972         BRIDGE_UNLOCK(sc);
 973         if (!gone) {
 974                 switch (ifs->if_type) {
 975                 case IFT_ETHER:
 976                 case IFT_L2VLAN:
 977                         /*
 978                          * Take the interface out of promiscuous mode.
 979                          */
 980                         (void) ifpromisc(ifs, 0);
 981                         break;
 982
 983                 case IFT_GIF:
 984                         break;
 985
 986                 default:
 987 #ifdef DIAGNOSTIC
 988                         panic("bridge_delete_member: impossible");
 989 #endif
 990                         break;
 991                 }
 992                 /* reneable any interface capabilities */
 993                 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
 994         }
 995         bstp_destroy(&bif->bif_stp);    /* prepare to free */
 996         BRIDGE_LOCK(sc);
 997         free(bif, M_DEVBUF);
 998 }
 999
1000 /*
1001  * bridge_delete_span:
1002  *
1003  *      Delete the specified span interface.
1004  */
1005 static void
1006 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1007 {
1008         BRIDGE_LOCK_ASSERT(sc);
1009
1010         KASSERT(bif->bif_ifp->if_bridge == NULL,
1011             ("%s: not a span interface", __func__));
1012
1013         LIST_REMOVE(bif, bif_next);
1014         free(bif, M_DEVBUF);
1015 }
1016
1017 static int
1018 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1019 {
1020         struct ifbreq *req = arg;
1021         struct bridge_iflist *bif = NULL;
1022         struct ifnet *ifs;
1023         int error = 0;
1024
1025         ifs = ifunit(req->ifbr_ifsname);
1026         if (ifs == NULL)
1027                 return (ENOENT);
1028         if (ifs->if_ioctl == NULL)      /* must be supported */
1029                 return (EINVAL);
1030
1031         /* If it's in the span list, it can't be a member. */
1032         LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1033                 if (ifs == bif->bif_ifp)
1034                         return (EBUSY);
1035
1036         if (ifs->if_bridge == sc)
1037                 return (EEXIST);
1038
1039         if (ifs->if_bridge != NULL)
1040                 return (EBUSY);
1041
1042         bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1043         if (bif == NULL)
1044                 return (ENOMEM);
1045
1046         bif->bif_ifp = ifs;
1047         bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1048         bif->bif_savedcaps = ifs->if_capenable;
1049
1050         switch (ifs->if_type) {
1051         case IFT_ETHER:
1052         case IFT_L2VLAN:
1053         case IFT_GIF:
1054                 /* permitted interface types */
1055                 break;
1056         default:
1057                 error = EINVAL;
1058                 goto out;
1059         }
1060
1061         /* Allow the first Ethernet member to define the MTU */
1062         if (LIST_EMPTY(&sc->sc_iflist))
1063                 sc->sc_ifp->if_mtu = ifs->if_mtu;
1064         else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1065                 if_printf(sc->sc_ifp, "invalid MTU: %lu(%s) != %lu\n",
1066                     ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1067                 error = EINVAL;
1068                 goto out;
1069         }
1070
1071         /*
1072          * Assign the interface's MAC address to the bridge if it's the first
1073          * member and the MAC address of the bridge has not been changed from
1074          * the default randomly generated one.
1075          */
1076         if (bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1077             !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1078                 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1079                 sc->sc_ifaddr = ifs;
1080                 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1081         }
1082
1083         ifs->if_bridge = sc;
1084         bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1085         /*
1086          * XXX: XLOCK HERE!?!
1087          *
1088          * NOTE: insert_***HEAD*** should be safe for the traversals.
1089          */
1090         LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1091
1092         /* Set interface capabilities to the intersection set of all members */
1093         bridge_mutecaps(sc);
1094         bridge_linkcheck(sc);
1095
1096         /* Place the interface into promiscuous mode */
1097         switch (ifs->if_type) {
1098                 case IFT_ETHER:
1099                 case IFT_L2VLAN:
1100                         BRIDGE_UNLOCK(sc);
1101                         error = ifpromisc(ifs, 1);
1102                         BRIDGE_LOCK(sc);
1103                         break;
1104         }
1105         if (error)
1106                 bridge_delete_member(sc, bif, 0);
1107 out:
1108         if (error) {
1109                 if (bif != NULL)
1110                         free(bif, M_DEVBUF);
1111         }
1112         return (error);
1113 }
1114
1115 static int
1116 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1117 {
1118         struct ifbreq *req = arg;
1119         struct bridge_iflist *bif;
1120
1121         bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1122         if (bif == NULL)
1123                 return (ENOENT);
1124
1125         bridge_delete_member(sc, bif, 0);
1126
1127         return (0);
1128 }
1129
1130 static int
1131 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1132 {
1133         struct ifbreq *req = arg;
1134         struct bridge_iflist *bif;
1135         struct bstp_port *bp;
1136
1137         bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1138         if (bif == NULL)
1139                 return (ENOENT);
1140
1141         bp = &bif->bif_stp;
1142         req->ifbr_ifsflags = bif->bif_flags;
1143         req->ifbr_state = bp->bp_state;
1144         req->ifbr_priority = bp->bp_priority;
1145         req->ifbr_path_cost = bp->bp_path_cost;
1146         req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1147         req->ifbr_proto = bp->bp_protover;
1148         req->ifbr_role = bp->bp_role;
1149         req->ifbr_stpflags = bp->bp_flags;
1150         req->ifbr_addrcnt = bif->bif_addrcnt;
1151         req->ifbr_addrmax = bif->bif_addrmax;
1152         req->ifbr_addrexceeded = bif->bif_addrexceeded;
1153
1154         /* Copy STP state options as flags */
1155         if (bp->bp_operedge)
1156                 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1157         if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1158                 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1159         if (bp->bp_ptp_link)
1160                 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1161         if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1162                 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1163         if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1164                 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1165         if (bp->bp_flags & BSTP_PORT_ADMCOST)
1166                 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1167         return (0);
1168 }
1169
1170 static int
1171 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1172 {
1173         struct ifbreq *req = arg;
1174         struct bridge_iflist *bif;
1175         struct bstp_port *bp;
1176         int error;
1177
1178         bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1179         if (bif == NULL)
1180                 return (ENOENT);
1181         bp = &bif->bif_stp;
1182
1183         if (req->ifbr_ifsflags & IFBIF_SPAN)
1184                 /* SPAN is readonly */
1185                 return (EINVAL);
1186
1187         if (req->ifbr_ifsflags & IFBIF_STP) {
1188                 if ((bif->bif_flags & IFBIF_STP) == 0) {
1189                         error = bstp_enable(&bif->bif_stp);
1190                         if (error)
1191                                 return (error);
1192                 }
1193         } else {
1194                 if ((bif->bif_flags & IFBIF_STP) != 0)
1195                         bstp_disable(&bif->bif_stp);
1196         }
1197
1198         /* Pass on STP flags */
1199         bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1200         bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1201         bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1202         bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1203
1204         /* Save the bits relating to the bridge */
1205         bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1206
1207         return (0);
1208 }
1209
1210 static int
1211 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1212 {
1213         struct ifbrparam *param = arg;
1214
1215         sc->sc_brtmax = param->ifbrp_csize;
1216         bridge_rttrim(sc);
1217
1218         return (0);
1219 }
1220
1221 static int
1222 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1223 {
1224         struct ifbrparam *param = arg;
1225
1226         param->ifbrp_csize = sc->sc_brtmax;
1227
1228         return (0);
1229 }
1230
1231 static int
1232 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1233 {
1234         struct ifbifconf *bifc = arg;
1235         struct bridge_iflist *bif;
1236         struct ifbreq breq;
1237         char *buf, *outbuf;
1238         int count, buflen, len, error = 0;
1239
1240         count = 0;
1241         LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1242                 count++;
1243         LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1244                 count++;
1245
1246         buflen = sizeof(breq) * count;
1247         if (bifc->ifbic_len == 0) {
1248                 bifc->ifbic_len = buflen;
1249                 return (0);
1250         }
1251         BRIDGE_UNLOCK(sc);
1252         outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1253         BRIDGE_LOCK(sc);
1254
1255         count = 0;
1256         buf = outbuf;
1257         len = min(bifc->ifbic_len, buflen);
1258         bzero(&breq, sizeof(breq));
1259         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1260                 if (len < sizeof(breq))
1261                         break;
1262
1263                 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1264                     sizeof(breq.ifbr_ifsname));
1265                 /* Fill in the ifbreq structure */
1266                 error = bridge_ioctl_gifflags(sc, &breq);
1267                 if (error)
1268                         break;
1269                 memcpy(buf, &breq, sizeof(breq));
1270                 count++;
1271                 buf += sizeof(breq);
1272                 len -= sizeof(breq);
1273         }
1274         LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1275                 if (len < sizeof(breq))
1276                         break;
1277
1278                 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1279                     sizeof(breq.ifbr_ifsname));
1280                 breq.ifbr_ifsflags = bif->bif_flags;
1281                 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1282                 memcpy(buf, &breq, sizeof(breq));
1283                 count++;
1284                 buf += sizeof(breq);
1285                 len -= sizeof(breq);
1286         }
1287
1288         BRIDGE_UNLOCK(sc);
1289         bifc->ifbic_len = sizeof(breq) * count;
1290         error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1291         BRIDGE_LOCK(sc);
1292         free(outbuf, M_TEMP);
1293         return (error);
1294 }
1295
1296 static int
1297 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1298 {
1299         struct ifbaconf *bac = arg;
1300         struct bridge_rtnode *brt;
1301         struct ifbareq bareq;
1302         char *buf, *outbuf;
1303         int count, buflen, len, error = 0;
1304
1305         if (bac->ifbac_len == 0)
1306                 return (0);
1307
1308         count = 0;
1309         LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1310                 count++;
1311         buflen = sizeof(bareq) * count;
1312
1313         BRIDGE_UNLOCK(sc);
1314         outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1315         BRIDGE_LOCK(sc);
1316
1317         count = 0;
1318         buf = outbuf;
1319         len = min(bac->ifbac_len, buflen);
1320         bzero(&bareq, sizeof(bareq));
1321         LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1322                 if (len < sizeof(bareq))
1323                         goto out;
1324                 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1325                     sizeof(bareq.ifba_ifsname));
1326                 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1327                 bareq.ifba_vlan = brt->brt_vlan;
1328                 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1329                                 time_uptime < brt->brt_expire)
1330                         bareq.ifba_expire = brt->brt_expire - time_uptime;
1331                 else
1332                         bareq.ifba_expire = 0;
1333                 bareq.ifba_flags = brt->brt_flags;
1334
1335                 memcpy(buf, &bareq, sizeof(bareq));
1336                 count++;
1337                 buf += sizeof(bareq);
1338                 len -= sizeof(bareq);
1339         }
1340 out:
1341         BRIDGE_UNLOCK(sc);
1342         bac->ifbac_len = sizeof(bareq) * count;
1343         error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1344         BRIDGE_LOCK(sc);
1345         free(outbuf, M_TEMP);
1346         return (error);
1347 }
1348
1349 static int
1350 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1351 {
1352         struct ifbareq *req = arg;
1353         struct bridge_iflist *bif;
1354         int error;
1355
1356         bif = bridge_lookup_member(sc, req->ifba_ifsname);
1357         if (bif == NULL)
1358                 return (ENOENT);
1359
1360         error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1361             req->ifba_flags);
1362
1363         return (error);
1364 }
1365
1366 static int
1367 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1368 {
1369         struct ifbrparam *param = arg;
1370
1371         sc->sc_brttimeout = param->ifbrp_ctime;
1372         return (0);
1373 }
1374
1375 static int
1376 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1377 {
1378         struct ifbrparam *param = arg;
1379
1380         param->ifbrp_ctime = sc->sc_brttimeout;
1381         return (0);
1382 }
1383
1384 static int
1385 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1386 {
1387         struct ifbareq *req = arg;
1388
1389         return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1390 }
1391
1392 static int
1393 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1394 {
1395         struct ifbreq *req = arg;
1396
1397         bridge_rtflush(sc, req->ifbr_ifsflags);
1398         return (0);
1399 }
1400
1401 static int
1402 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1403 {
1404         struct ifbrparam *param = arg;
1405         struct bstp_state *bs = &sc->sc_stp;
1406
1407         param->ifbrp_prio = bs->bs_bridge_priority;
1408         return (0);
1409 }
1410
1411 static int
1412 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1413 {
1414         struct ifbrparam *param = arg;
1415
1416         return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1417 }
1418
1419 static int
1420 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1421 {
1422         struct ifbrparam *param = arg;
1423         struct bstp_state *bs = &sc->sc_stp;
1424
1425         param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1426         return (0);
1427 }
1428
1429 static int
1430 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1431 {
1432         struct ifbrparam *param = arg;
1433
1434         return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1435 }
1436
1437 static int
1438 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1439 {
1440         struct ifbrparam *param = arg;
1441         struct bstp_state *bs = &sc->sc_stp;
1442
1443         param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1444         return (0);
1445 }
1446
1447 static int
1448 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1449 {
1450         struct ifbrparam *param = arg;
1451
1452         return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1453 }
1454
1455 static int
1456 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1457 {
1458         struct ifbrparam *param = arg;
1459         struct bstp_state *bs = &sc->sc_stp;
1460
1461         param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1462         return (0);
1463 }
1464
1465 static int
1466 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1467 {
1468         struct ifbrparam *param = arg;
1469
1470         return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1471 }
1472
1473 static int
1474 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1475 {
1476         struct ifbreq *req = arg;
1477         struct bridge_iflist *bif;
1478
1479         bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1480         if (bif == NULL)
1481                 return (ENOENT);
1482
1483         return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1484 }
1485
1486 static int
1487 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1488 {
1489         struct ifbreq *req = arg;
1490         struct bridge_iflist *bif;
1491
1492         bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1493         if (bif == NULL)
1494                 return (ENOENT);
1495
1496         return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1497 }
1498
1499 static int
1500 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1501 {
1502         struct ifbreq *req = arg;
1503         struct bridge_iflist *bif;
1504
1505         bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1506         if (bif == NULL)
1507                 return (ENOENT);
1508
1509         bif->bif_addrmax = req->ifbr_addrmax;
1510         return (0);
1511 }
1512
1513 static int
1514 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1515 {
1516         struct ifbreq *req = arg;
1517         struct bridge_iflist *bif = NULL;
1518         struct ifnet *ifs;
1519
1520         ifs = ifunit(req->ifbr_ifsname);
1521         if (ifs == NULL)
1522                 return (ENOENT);
1523
1524         LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1525                 if (ifs == bif->bif_ifp)
1526                         return (EBUSY);
1527
1528         if (ifs->if_bridge != NULL)
1529                 return (EBUSY);
1530
1531         switch (ifs->if_type) {
1532                 case IFT_ETHER:
1533                 case IFT_GIF:
1534                 case IFT_L2VLAN:
1535                         break;
1536                 default:
1537                         return (EINVAL);
1538         }
1539
1540         bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1541         if (bif == NULL)
1542                 return (ENOMEM);
1543
1544         bif->bif_ifp = ifs;
1545         bif->bif_flags = IFBIF_SPAN;
1546
1547         LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1548
1549         return (0);
1550 }
1551
1552 static int
1553 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1554 {
1555         struct ifbreq *req = arg;
1556         struct bridge_iflist *bif;
1557         struct ifnet *ifs;
1558
1559         ifs = ifunit(req->ifbr_ifsname);
1560         if (ifs == NULL)
1561                 return (ENOENT);
1562
1563         LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1564                 if (ifs == bif->bif_ifp)
1565                         break;
1566
1567         if (bif == NULL)
1568                 return (ENOENT);
1569
1570         bridge_delete_span(sc, bif);
1571
1572         return (0);
1573 }
1574
1575 static int
1576 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1577 {
1578         struct ifbropreq *req = arg;
1579         struct bstp_state *bs = &sc->sc_stp;
1580         struct bstp_port *root_port;
1581
1582         req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1583         req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1584         req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1585
1586         root_port = bs->bs_root_port;
1587         if (root_port == NULL)
1588                 req->ifbop_root_port = 0;
1589         else
1590                 req->ifbop_root_port = root_port->bp_ifp->if_index;
1591
1592         req->ifbop_holdcount = bs->bs_txholdcount;
1593         req->ifbop_priority = bs->bs_bridge_priority;
1594         req->ifbop_protocol = bs->bs_protover;
1595         req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1596         req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1597         req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1598         req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1599         req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1600         req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1601
1602         return (0);
1603 }
1604
1605 static int
1606 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1607 {
1608         struct ifbrparam *param = arg;
1609
1610         param->ifbrp_cexceeded = sc->sc_brtexceeded;
1611         return (0);
1612 }
1613
1614 static int
1615 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1616 {
1617         struct ifbpstpconf *bifstp = arg;
1618         struct bridge_iflist *bif;
1619         struct bstp_port *bp;
1620         struct ifbpstpreq bpreq;
1621         char *buf, *outbuf;
1622         int count, buflen, len, error = 0;
1623
1624         count = 0;
1625         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1626                 if ((bif->bif_flags & IFBIF_STP) != 0)
1627                         count++;
1628         }
1629
1630         buflen = sizeof(bpreq) * count;
1631         if (bifstp->ifbpstp_len == 0) {
1632                 bifstp->ifbpstp_len = buflen;
1633                 return (0);
1634         }
1635
1636         BRIDGE_UNLOCK(sc);
1637         outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1638         BRIDGE_LOCK(sc);
1639
1640         count = 0;
1641         buf = outbuf;
1642         len = min(bifstp->ifbpstp_len, buflen);
1643         bzero(&bpreq, sizeof(bpreq));
1644         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1645                 if (len < sizeof(bpreq))
1646                         break;
1647
1648                 if ((bif->bif_flags & IFBIF_STP) == 0)
1649                         continue;
1650
1651                 bp = &bif->bif_stp;
1652                 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1653                 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1654                 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1655                 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1656                 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1657                 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1658
1659                 memcpy(buf, &bpreq, sizeof(bpreq));
1660                 count++;
1661                 buf += sizeof(bpreq);
1662                 len -= sizeof(bpreq);
1663         }
1664
1665         BRIDGE_UNLOCK(sc);
1666         bifstp->ifbpstp_len = sizeof(bpreq) * count;
1667         error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1668         BRIDGE_LOCK(sc);
1669         free(outbuf, M_TEMP);
1670         return (error);
1671 }
1672
1673 static int
1674 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1675 {
1676         struct ifbrparam *param = arg;
1677
1678         return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1679 }
1680
1681 static int
1682 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1683 {
1684         struct ifbrparam *param = arg;
1685
1686         return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1687 }
1688
1689 /*
1690  * bridge_ifdetach:
1691  *
1692  *      Detach an interface from a bridge.  Called when a member
1693  *      interface is detaching.
1694  */
1695 static void
1696 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1697 {
1698         struct bridge_softc *sc = ifp->if_bridge;
1699         struct bridge_iflist *bif;
1700
1701         /* Check if the interface is a bridge member */
1702         if (sc != NULL) {
1703                 BRIDGE_LOCK(sc);
1704
1705                 bif = bridge_lookup_member_if(sc, ifp);
1706                 if (bif != NULL)
1707                         bridge_delete_member(sc, bif, 1);
1708
1709                 BRIDGE_UNLOCK(sc);
1710                 return;
1711         }
1712
1713         /* Check if the interface is a span port */
1714         mtx_lock(&bridge_list_mtx);
1715         LIST_FOREACH(sc, &bridge_list, sc_list) {
1716                 BRIDGE_LOCK(sc);
1717                 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1718                         if (ifp == bif->bif_ifp) {
1719                                 bridge_delete_span(sc, bif);
1720                                 break;
1721                         }
1722
1723                 BRIDGE_UNLOCK(sc);
1724         }
1725         mtx_unlock(&bridge_list_mtx);
1726 }
1727
1728 /*
1729  * bridge_init:
1730  *
1731  *      Initialize a bridge interface.
1732  */
1733 static void
1734 bridge_init(void *xsc)
1735 {
1736         struct bridge_softc *sc = (struct bridge_softc *)xsc;
1737         struct ifnet *ifp = sc->sc_ifp;
1738
1739         if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1740                 return;
1741
1742         BRIDGE_LOCK(sc);
1743         callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1744             bridge_timer, sc);
1745
1746         ifp->if_drv_flags |= IFF_DRV_RUNNING;
1747         bstp_init(&sc->sc_stp);         /* Initialize Spanning Tree */
1748
1749         BRIDGE_UNLOCK(sc);
1750 }
1751
1752 /*
1753  * bridge_stop:
1754  *
1755  *      Stop the bridge interface.
1756  */
1757 static void
1758 bridge_stop(struct ifnet *ifp, int disable)
1759 {
1760         struct bridge_softc *sc = ifp->if_softc;
1761
1762         BRIDGE_LOCK_ASSERT(sc);
1763
1764         if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1765                 return;
1766
1767         callout_stop(&sc->sc_brcallout);
1768         bstp_stop(&sc->sc_stp);
1769
1770         bridge_rtflush(sc, IFBF_FLUSHDYN);
1771
1772         ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1773 }
1774
1775 /*
1776  * bridge_enqueue:
1777  *
1778  *      Enqueue a packet on a bridge member interface.
1779  *
1780  */
1781 static int
1782 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1783 {
1784         int len, err = 0;
1785         short mflags;
1786         struct mbuf *m0;
1787
1788         /* We may be sending a fragment so traverse the mbuf */
1789         for (; m; m = m0) {
1790                 m0 = m->m_nextpkt;
1791                 m->m_nextpkt = NULL;
1792                 len = m->m_pkthdr.len;
1793                 mflags = m->m_flags;
1794
1795                 /*
1796                  * If underlying interface can not do VLAN tag insertion itself
1797                  * then attach a packet tag that holds it.
1798                  */
1799                 if ((m->m_flags & M_VLANTAG) &&
1800                     (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1801                         m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1802                         if (m == NULL) {
1803                                 if_printf(dst_ifp,
1804                                     "unable to prepend VLAN header\n");
1805                                 dst_ifp->if_oerrors++;
1806                                 continue;
1807                         }
1808                         m->m_flags &= ~M_VLANTAG;
1809                 }
1810
1811                 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1812                         m_freem(m0);
1813                         sc->sc_ifp->if_oerrors++;
1814                         break;
1815                 }
1816
1817                 sc->sc_ifp->if_opackets++;
1818                 sc->sc_ifp->if_obytes += len;
1819                 if (mflags & M_MCAST)
1820                         sc->sc_ifp->if_omcasts++;
1821         }
1822
1823         return (err);
1824 }
1825
1826 /*
1827  * bridge_dummynet:
1828  *
1829  *      Receive a queued packet from dummynet and pass it on to the output
1830  *      interface.
1831  *
1832  *      The mbuf has the Ethernet header already attached.
1833  */
1834 static void
1835 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1836 {
1837         struct bridge_softc *sc;
1838
1839         sc = ifp->if_bridge;
1840
1841         /*
1842          * The packet didnt originate from a member interface. This should only
1843          * ever happen if a member interface is removed while packets are
1844          * queued for it.
1845          */
1846         if (sc == NULL) {
1847                 m_freem(m);
1848                 return;
1849         }
1850
1851         if (PFIL_HOOKED(&V_inet_pfil_hook)
1852 #ifdef INET6
1853             || PFIL_HOOKED(&V_inet6_pfil_hook)
1854 #endif
1855             ) {
1856                 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1857                         return;
1858                 if (m == NULL)
1859                         return;
1860         }
1861
1862         bridge_enqueue(sc, ifp, m);
1863 }
1864
1865 /*
1866  * bridge_output:
1867  *
1868  *      Send output from a bridge member interface.  This
1869  *      performs the bridging function for locally originated
1870  *      packets.
1871  *
1872  *      The mbuf has the Ethernet header already attached.  We must
1873  *      enqueue or free the mbuf before returning.
1874  */
1875 static int
1876 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1877     struct rtentry *rt)
1878 {
1879         struct ether_header *eh;
1880         struct ifnet *dst_if;
1881         struct bridge_softc *sc;
1882         uint16_t vlan;
1883
1884         if (m->m_len < ETHER_HDR_LEN) {
1885                 m = m_pullup(m, ETHER_HDR_LEN);
1886                 if (m == NULL)
1887                         return (0);
1888         }
1889
1890         eh = mtod(m, struct ether_header *);
1891         sc = ifp->if_bridge;
1892         vlan = VLANTAGOF(m);
1893
1894         BRIDGE_LOCK(sc);
1895
1896         /*
1897          * If bridge is down, but the original output interface is up,
1898          * go ahead and send out that interface.  Otherwise, the packet
1899          * is dropped below.
1900          */
1901         if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1902                 dst_if = ifp;
1903                 goto sendunicast;
1904         }
1905
1906         /*
1907          * If the packet is a multicast, or we don't know a better way to
1908          * get there, send to all interfaces.
1909          */
1910         if (ETHER_IS_MULTICAST(eh->ether_dhost))
1911                 dst_if = NULL;
1912         else
1913                 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1914         if (dst_if == NULL) {
1915                 struct bridge_iflist *bif;
1916                 struct mbuf *mc;
1917                 int error = 0, used = 0;
1918
1919                 bridge_span(sc, m);
1920
1921                 BRIDGE_LOCK2REF(sc, error);
1922                 if (error) {
1923                         m_freem(m);
1924                         return (0);
1925                 }
1926
1927                 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1928                         dst_if = bif->bif_ifp;
1929
1930                         if (dst_if->if_type == IFT_GIF)
1931                                 continue;
1932                         if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1933                                 continue;
1934
1935                         /*
1936                          * If this is not the original output interface,
1937                          * and the interface is participating in spanning
1938                          * tree, make sure the port is in a state that
1939                          * allows forwarding.
1940                          */
1941                         if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
1942                             bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1943                                 continue;
1944
1945                         if (LIST_NEXT(bif, bif_next) == NULL) {
1946                                 used = 1;
1947                                 mc = m;
1948                         } else {
1949                                 mc = m_copypacket(m, M_DONTWAIT);
1950                                 if (mc == NULL) {
1951                                         sc->sc_ifp->if_oerrors++;
1952                                         continue;
1953                                 }
1954                         }
1955
1956                         bridge_enqueue(sc, dst_if, mc);
1957                 }
1958                 if (used == 0)
1959                         m_freem(m);
1960                 BRIDGE_UNREF(sc);
1961                 return (0);
1962         }
1963
1964 sendunicast:
1965         /*
1966          * XXX Spanning tree consideration here?
1967          */
1968
1969         bridge_span(sc, m);
1970         if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1971                 m_freem(m);
1972                 BRIDGE_UNLOCK(sc);
1973                 return (0);
1974         }
1975
1976         BRIDGE_UNLOCK(sc);
1977         bridge_enqueue(sc, dst_if, m);
1978         return (0);
1979 }
1980
1981 /*
1982  * bridge_transmit:
1983  *
1984  *      Do output on a bridge.
1985  *
1986  */
1987 static int
1988 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
1989 {
1990         struct bridge_softc *sc;
1991         struct ether_header *eh;
1992         struct ifnet *dst_if;
1993         int error = 0;
1994
1995         sc = ifp->if_softc;
1996
1997         ETHER_BPF_MTAP(ifp, m);
1998
1999         eh = mtod(m, struct ether_header *);
2000
2001         BRIDGE_LOCK(sc);
2002         if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2003             (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2004                 BRIDGE_UNLOCK(sc);
2005                 error = bridge_enqueue(sc, dst_if, m);
2006         } else
2007                 bridge_broadcast(sc, ifp, m, 0);
2008
2009         return (error);
2010 }
2011
2012 /*
2013  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2014  */
2015 static void
2016 bridge_qflush(struct ifnet *ifp __unused)
2017 {
2018 }
2019
2020 /*
2021  * bridge_forward:
2022  *
2023  *      The forwarding function of the bridge.
2024  *
2025  *      NOTE: Releases the lock on return.
2026  */
2027 static void
2028 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2029     struct mbuf *m)
2030 {
2031         struct bridge_iflist *dbif;
2032         struct ifnet *src_if, *dst_if, *ifp;
2033         struct ether_header *eh;
2034         uint16_t vlan;
2035         uint8_t *dst;
2036         int error;
2037
2038         src_if = m->m_pkthdr.rcvif;
2039         ifp = sc->sc_ifp;
2040
2041         ifp->if_ipackets++;
2042         ifp->if_ibytes += m->m_pkthdr.len;
2043         vlan = VLANTAGOF(m);
2044
2045         if ((sbif->bif_flags & IFBIF_STP) &&
2046             sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2047                 goto drop;
2048
2049         eh = mtod(m, struct ether_header *);
2050         dst = eh->ether_dhost;
2051
2052         /* If the interface is learning, record the address. */
2053         if (sbif->bif_flags & IFBIF_LEARNING) {
2054                 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2055                     sbif, 0, IFBAF_DYNAMIC);
2056                 /*
2057                  * If the interface has addresses limits then deny any source
2058                  * that is not in the cache.
2059                  */
2060                 if (error && sbif->bif_addrmax)
2061                         goto drop;
2062         }
2063
2064         if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2065             sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2066                 goto drop;
2067
2068         /*
2069          * At this point, the port either doesn't participate
2070          * in spanning tree or it is in the forwarding state.
2071          */
2072
2073         /*
2074          * If the packet is unicast, destined for someone on
2075          * "this" side of the bridge, drop it.
2076          */
2077         if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2078                 dst_if = bridge_rtlookup(sc, dst, vlan);
2079                 if (src_if == dst_if)
2080                         goto drop;
2081         } else {
2082                 /*
2083                  * Check if its a reserved multicast address, any address
2084                  * listed in 802.1D section 7.12.6 may not be forwarded by the
2085                  * bridge.
2086                  * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2087                  */
2088                 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2089                     dst[2] == 0xc2 && dst[3] == 0x00 &&
2090                     dst[4] == 0x00 && dst[5] <= 0x0f)
2091                         goto drop;
2092
2093                 /* ...forward it to all interfaces. */
2094                 ifp->if_imcasts++;
2095                 dst_if = NULL;
2096         }
2097
2098         /*
2099          * If we have a destination interface which is a member of our bridge,
2100          * OR this is a unicast packet, push it through the bpf(4) machinery.
2101          * For broadcast or multicast packets, don't bother because it will
2102          * be reinjected into ether_input. We do this before we pass the packets
2103          * through the pfil(9) framework, as it is possible that pfil(9) will
2104          * drop the packet, or possibly modify it, making it difficult to debug
2105          * firewall issues on the bridge.
2106          */
2107         if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2108                 ETHER_BPF_MTAP(ifp, m);
2109
2110         /* run the packet filter */
2111         if (PFIL_HOOKED(&V_inet_pfil_hook)
2112 #ifdef INET6
2113             || PFIL_HOOKED(&V_inet6_pfil_hook)
2114 #endif
2115             ) {
2116                 BRIDGE_UNLOCK(sc);
2117                 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2118                         return;
2119                 if (m == NULL)
2120                         return;
2121                 BRIDGE_LOCK(sc);
2122         }
2123
2124         if (dst_if == NULL) {
2125                 bridge_broadcast(sc, src_if, m, 1);
2126                 return;
2127         }
2128
2129         /*
2130          * At this point, we're dealing with a unicast frame
2131          * going to a different interface.
2132          */
2133         if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2134                 goto drop;
2135
2136         dbif = bridge_lookup_member_if(sc, dst_if);
2137         if (dbif == NULL)
2138                 /* Not a member of the bridge (anymore?) */
2139                 goto drop;
2140
2141         /* Private segments can not talk to each other */
2142         if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2143                 goto drop;
2144
2145         if ((dbif->bif_flags & IFBIF_STP) &&
2146             dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2147                 goto drop;
2148
2149         BRIDGE_UNLOCK(sc);
2150
2151         if (PFIL_HOOKED(&V_inet_pfil_hook)
2152 #ifdef INET6
2153             || PFIL_HOOKED(&V_inet6_pfil_hook)
2154 #endif
2155             ) {
2156                 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2157                         return;
2158                 if (m == NULL)
2159                         return;
2160         }
2161
2162         bridge_enqueue(sc, dst_if, m);
2163         return;
2164
2165 drop:
2166         BRIDGE_UNLOCK(sc);
2167         m_freem(m);
2168 }
2169
2170 /*
2171  * bridge_input:
2172  *
2173  *      Receive input from a member interface.  Queue the packet for
2174  *      bridging if it is not for us.
2175  */
2176 static struct mbuf *
2177 bridge_input(struct ifnet *ifp, struct mbuf *m)
2178 {
2179         struct bridge_softc *sc = ifp->if_bridge;
2180         struct bridge_iflist *bif, *bif2;
2181         struct ifnet *bifp;
2182         struct ether_header *eh;
2183         struct mbuf *mc, *mc2;
2184         uint16_t vlan;
2185         int error;
2186
2187         if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2188                 return (m);
2189
2190         bifp = sc->sc_ifp;
2191         vlan = VLANTAGOF(m);
2192
2193         /*
2194          * Implement support for bridge monitoring. If this flag has been
2195          * set on this interface, discard the packet once we push it through
2196          * the bpf(4) machinery, but before we do, increment the byte and
2197          * packet counters associated with this interface.
2198          */
2199         if ((bifp->if_flags & IFF_MONITOR) != 0) {
2200                 m->m_pkthdr.rcvif  = bifp;
2201                 ETHER_BPF_MTAP(bifp, m);
2202                 bifp->if_ipackets++;
2203                 bifp->if_ibytes += m->m_pkthdr.len;
2204                 m_freem(m);
2205                 return (NULL);
2206         }
2207         BRIDGE_LOCK(sc);
2208         bif = bridge_lookup_member_if(sc, ifp);
2209         if (bif == NULL) {
2210                 BRIDGE_UNLOCK(sc);
2211                 return (m);
2212         }
2213
2214         eh = mtod(m, struct ether_header *);
2215
2216         bridge_span(sc, m);
2217
2218         if (m->m_flags & (M_BCAST|M_MCAST)) {
2219                 /* Tap off 802.1D packets; they do not get forwarded. */
2220                 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2221                     ETHER_ADDR_LEN) == 0) {
2222                         bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2223                         BRIDGE_UNLOCK(sc);
2224                         return (NULL);
2225                 }
2226
2227                 if ((bif->bif_flags & IFBIF_STP) &&
2228                     bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2229                         BRIDGE_UNLOCK(sc);
2230                         return (m);
2231                 }
2232
2233                 /*
2234                  * Make a deep copy of the packet and enqueue the copy
2235                  * for bridge processing; return the original packet for
2236                  * local processing.
2237                  */
2238                 mc = m_dup(m, M_DONTWAIT);
2239                 if (mc == NULL) {
2240                         BRIDGE_UNLOCK(sc);
2241                         return (m);
2242                 }
2243
2244                 /* Perform the bridge forwarding function with the copy. */
2245                 bridge_forward(sc, bif, mc);
2246
2247                 /*
2248                  * Reinject the mbuf as arriving on the bridge so we have a
2249                  * chance at claiming multicast packets. We can not loop back
2250                  * here from ether_input as a bridge is never a member of a
2251                  * bridge.
2252                  */
2253                 KASSERT(bifp->if_bridge == NULL,
2254                     ("loop created in bridge_input"));
2255                 mc2 = m_dup(m, M_DONTWAIT);
2256                 if (mc2 != NULL) {
2257                         /* Keep the layer3 header aligned */
2258                         int i = min(mc2->m_pkthdr.len, max_protohdr);
2259                         mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2260                 }
2261                 if (mc2 != NULL) {
2262                         mc2->m_pkthdr.rcvif = bifp;
2263                         (*bifp->if_input)(bifp, mc2);
2264                 }
2265
2266                 /* Return the original packet for local processing. */
2267                 return (m);
2268         }
2269
2270         if ((bif->bif_flags & IFBIF_STP) &&
2271             bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2272                 BRIDGE_UNLOCK(sc);
2273                 return (m);
2274         }
2275
2276 #if (defined(INET) || defined(INET6))
2277 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2278         || ((iface)->if_carp \
2279             && (*carp_forus_p)((iface), eh->ether_dhost))
2280 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2281         || ((iface)->if_carp \
2282             && (*carp_forus_p)((iface), eh->ether_shost))
2283 #else
2284 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2285 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2286 #endif
2287
2288 #ifdef INET6
2289 #   define OR_PFIL_HOOKED_INET6 \
2290         || PFIL_HOOKED(&V_inet6_pfil_hook)
2291 #else
2292 #   define OR_PFIL_HOOKED_INET6
2293 #endif
2294
2295 #define GRAB_OUR_PACKETS(iface) \
2296         if ((iface)->if_type == IFT_GIF) \
2297                 continue; \
2298         /* It is destined for us. */ \
2299         if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2300             OR_CARP_CHECK_WE_ARE_DST((iface))                           \
2301             ) {                                                         \
2302                 if ((iface)->if_type == IFT_BRIDGE) {                   \
2303                         ETHER_BPF_MTAP(iface, m);                       \
2304                         iface->if_ipackets++;                           \
2305                         iface->if_ibytes += m->m_pkthdr.len;            \
2306                         /* Filter on the physical interface. */         \
2307                         if (pfil_local_phys &&                          \
2308                             (PFIL_HOOKED(&V_inet_pfil_hook)             \
2309                              OR_PFIL_HOOKED_INET6)) {                   \
2310                                 if (bridge_pfil(&m, NULL, ifp,          \
2311                                     PFIL_IN) != 0 || m == NULL) {       \
2312                                         BRIDGE_UNLOCK(sc);              \
2313                                         return (NULL);                  \
2314                                 }                                       \
2315                         }                                               \
2316                 }                                                       \
2317                 if (bif->bif_flags & IFBIF_LEARNING) {                  \
2318                         error = bridge_rtupdate(sc, eh->ether_shost,    \
2319                             vlan, bif, 0, IFBAF_DYNAMIC);               \
2320                         if (error && bif->bif_addrmax) {                \
2321                                 BRIDGE_UNLOCK(sc);                      \
2322                                 m_freem(m);                             \
2323                                 return (NULL);                          \
2324                         }                                               \
2325                 }                                                       \
2326                 m->m_pkthdr.rcvif = iface;                              \
2327                 BRIDGE_UNLOCK(sc);                                      \
2328                 return (m);                                             \
2329         }                                                               \
2330                                                                         \
2331         /* We just received a packet that we sent out. */               \
2332         if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2333             OR_CARP_CHECK_WE_ARE_SRC((iface))                   \
2334             ) {                                                         \
2335                 BRIDGE_UNLOCK(sc);                                      \
2336                 m_freem(m);                                             \
2337                 return (NULL);                                          \
2338         }
2339
2340         /*
2341          * Unicast.  Make sure it's not for the bridge.
2342          */
2343         do { GRAB_OUR_PACKETS(bifp) } while (0);
2344
2345         /*
2346          * Give a chance for ifp at first priority. This will help when the
2347          * packet comes through the interface like VLAN's with the same MACs
2348          * on several interfaces from the same bridge. This also will save
2349          * some CPU cycles in case the destination interface and the input
2350          * interface (eq ifp) are the same.
2351          */
2352         do { GRAB_OUR_PACKETS(ifp) } while (0);
2353
2354         /* Now check the all bridge members. */
2355         LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2356                 GRAB_OUR_PACKETS(bif2->bif_ifp)
2357         }
2358
2359 #undef OR_CARP_CHECK_WE_ARE_DST
2360 #undef OR_CARP_CHECK_WE_ARE_SRC
2361 #undef OR_PFIL_HOOKED_INET6
2362 #undef GRAB_OUR_PACKETS
2363
2364         /* Perform the bridge forwarding function. */
2365         bridge_forward(sc, bif, m);
2366
2367         return (NULL);
2368 }
2369
2370 /*
2371  * bridge_broadcast:
2372  *
2373  *      Send a frame to all interfaces that are members of
2374  *      the bridge, except for the one on which the packet
2375  *      arrived.
2376  *
2377  *      NOTE: Releases the lock on return.
2378  */
2379 static void
2380 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2381     struct mbuf *m, int runfilt)
2382 {
2383         struct bridge_iflist *dbif, *sbif;
2384         struct mbuf *mc;
2385         struct ifnet *dst_if;
2386         int error = 0, used = 0, i;
2387
2388         sbif = bridge_lookup_member_if(sc, src_if);
2389
2390         BRIDGE_LOCK2REF(sc, error);
2391         if (error) {
2392                 m_freem(m);
2393                 return;
2394         }
2395
2396         /* Filter on the bridge interface before broadcasting */
2397         if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2398 #ifdef INET6
2399             || PFIL_HOOKED(&V_inet6_pfil_hook)
2400 #endif
2401             )) {
2402                 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2403                         goto out;
2404                 if (m == NULL)
2405                         goto out;
2406         }
2407
2408         LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2409                 dst_if = dbif->bif_ifp;
2410                 if (dst_if == src_if)
2411                         continue;
2412
2413                 /* Private segments can not talk to each other */
2414                 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2415                         continue;
2416
2417                 if ((dbif->bif_flags & IFBIF_STP) &&
2418                     dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2419                         continue;
2420
2421                 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2422                     (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2423                         continue;
2424
2425                 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2426                         continue;
2427
2428                 if (LIST_NEXT(dbif, bif_next) == NULL) {
2429                         mc = m;
2430                         used = 1;
2431                 } else {
2432                         mc = m_dup(m, M_DONTWAIT);
2433                         if (mc == NULL) {
2434                                 sc->sc_ifp->if_oerrors++;
2435                                 continue;
2436                         }
2437                 }
2438
2439                 /*
2440                  * Filter on the output interface. Pass a NULL bridge interface
2441                  * pointer so we do not redundantly filter on the bridge for
2442                  * each interface we broadcast on.
2443                  */
2444                 if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2445 #ifdef INET6
2446                     || PFIL_HOOKED(&V_inet6_pfil_hook)
2447 #endif
2448                     )) {
2449                         if (used == 0) {
2450                                 /* Keep the layer3 header aligned */
2451                                 i = min(mc->m_pkthdr.len, max_protohdr);
2452                                 mc = m_copyup(mc, i, ETHER_ALIGN);
2453                                 if (mc == NULL) {
2454                                         sc->sc_ifp->if_oerrors++;
2455                                         continue;
2456                                 }
2457                         }
2458                         if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2459                                 continue;
2460                         if (mc == NULL)
2461                                 continue;
2462                 }
2463
2464                 bridge_enqueue(sc, dst_if, mc);
2465         }
2466         if (used == 0)
2467                 m_freem(m);
2468
2469 out:
2470         BRIDGE_UNREF(sc);
2471 }
2472
2473 /*
2474  * bridge_span:
2475  *
2476  *      Duplicate a packet out one or more interfaces that are in span mode,
2477  *      the original mbuf is unmodified.
2478  */
2479 static void
2480 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2481 {
2482         struct bridge_iflist *bif;
2483         struct ifnet *dst_if;
2484         struct mbuf *mc;
2485
2486         if (LIST_EMPTY(&sc->sc_spanlist))
2487                 return;
2488
2489         LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2490                 dst_if = bif->bif_ifp;
2491
2492                 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2493                         continue;
2494
2495                 mc = m_copypacket(m, M_DONTWAIT);
2496                 if (mc == NULL) {
2497                         sc->sc_ifp->if_oerrors++;
2498                         continue;
2499                 }
2500
2501                 bridge_enqueue(sc, dst_if, mc);
2502         }
2503 }
2504
2505 /*
2506  * bridge_rtupdate:
2507  *
2508  *      Add a bridge routing entry.
2509  */
2510 static int
2511 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2512     struct bridge_iflist *bif, int setflags, uint8_t flags)
2513 {
2514         struct bridge_rtnode *brt;
2515         int error;
2516
2517         BRIDGE_LOCK_ASSERT(sc);
2518
2519         /* Check the source address is valid and not multicast. */
2520         if (ETHER_IS_MULTICAST(dst) ||
2521             (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2522              dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2523                 return (EINVAL);
2524
2525         /* 802.1p frames map to vlan 1 */
2526         if (vlan == 0)
2527                 vlan = 1;
2528
2529         /*
2530          * A route for this destination might already exist.  If so,
2531          * update it, otherwise create a new one.
2532          */
2533         if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2534                 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2535                         sc->sc_brtexceeded++;
2536                         return (ENOSPC);
2537                 }
2538                 /* Check per interface address limits (if enabled) */
2539                 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2540                         bif->bif_addrexceeded++;
2541                         return (ENOSPC);
2542                 }
2543
2544                 /*
2545                  * Allocate a new bridge forwarding node, and
2546                  * initialize the expiration time and Ethernet
2547                  * address.
2548                  */
2549                 brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2550                 if (brt == NULL)
2551                         return (ENOMEM);
2552
2553                 if (bif->bif_flags & IFBIF_STICKY)
2554                         brt->brt_flags = IFBAF_STICKY;
2555                 else
2556                         brt->brt_flags = IFBAF_DYNAMIC;
2557
2558                 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2559                 brt->brt_vlan = vlan;
2560
2561                 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2562                         uma_zfree(bridge_rtnode_zone, brt);
2563                         return (error);
2564                 }
2565                 brt->brt_dst = bif;
2566                 bif->bif_addrcnt++;
2567         }
2568
2569         if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2570             brt->brt_dst != bif) {
2571                 brt->brt_dst->bif_addrcnt--;
2572                 brt->brt_dst = bif;
2573                 brt->brt_dst->bif_addrcnt++;
2574         }
2575
2576         if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2577                 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2578         if (setflags)
2579                 brt->brt_flags = flags;
2580
2581         return (0);
2582 }
2583
2584 /*
2585  * bridge_rtlookup:
2586  *
2587  *      Lookup the destination interface for an address.
2588  */
2589 static struct ifnet *
2590 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2591 {
2592         struct bridge_rtnode *brt;
2593
2594         BRIDGE_LOCK_ASSERT(sc);
2595
2596         if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2597                 return (NULL);
2598
2599         return (brt->brt_ifp);
2600 }
2601
2602 /*
2603  * bridge_rttrim:
2604  *
2605  *      Trim the routine table so that we have a number
2606  *      of routing entries less than or equal to the
2607  *      maximum number.
2608  */
2609 static void
2610 bridge_rttrim(struct bridge_softc *sc)
2611 {
2612         struct bridge_rtnode *brt, *nbrt;
2613
2614         BRIDGE_LOCK_ASSERT(sc);
2615
2616         /* Make sure we actually need to do this. */
2617         if (sc->sc_brtcnt <= sc->sc_brtmax)
2618                 return;
2619
2620         /* Force an aging cycle; this might trim enough addresses. */
2621         bridge_rtage(sc);
2622         if (sc->sc_brtcnt <= sc->sc_brtmax)
2623                 return;
2624
2625         LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2626                 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2627                         bridge_rtnode_destroy(sc, brt);
2628                         if (sc->sc_brtcnt <= sc->sc_brtmax)
2629                                 return;
2630                 }
2631         }
2632 }
2633
2634 /*
2635  * bridge_timer:
2636  *
2637  *      Aging timer for the bridge.
2638  */
2639 static void
2640 bridge_timer(void *arg)
2641 {
2642         struct bridge_softc *sc = arg;
2643
2644         BRIDGE_LOCK_ASSERT(sc);
2645
2646         bridge_rtage(sc);
2647
2648         if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2649                 callout_reset(&sc->sc_brcallout,
2650                     bridge_rtable_prune_period * hz, bridge_timer, sc);
2651 }
2652
2653 /*
2654  * bridge_rtage:
2655  *
2656  *      Perform an aging cycle.
2657  */
2658 static void
2659 bridge_rtage(struct bridge_softc *sc)
2660 {
2661         struct bridge_rtnode *brt, *nbrt;
2662
2663         BRIDGE_LOCK_ASSERT(sc);
2664
2665         LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2666                 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2667                         if (time_uptime >= brt->brt_expire)
2668                                 bridge_rtnode_destroy(sc, brt);
2669                 }
2670         }
2671 }
2672
2673 /*
2674  * bridge_rtflush:
2675  *
2676  *      Remove all dynamic addresses from the bridge.
2677  */
2678 static void
2679 bridge_rtflush(struct bridge_softc *sc, int full)
2680 {
2681         struct bridge_rtnode *brt, *nbrt;
2682
2683         BRIDGE_LOCK_ASSERT(sc);
2684
2685         LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2686                 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2687                         bridge_rtnode_destroy(sc, brt);
2688         }
2689 }
2690
2691 /*
2692  * bridge_rtdaddr:
2693  *
2694  *      Remove an address from the table.
2695  */
2696 static int
2697 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2698 {
2699         struct bridge_rtnode *brt;
2700         int found = 0;
2701
2702         BRIDGE_LOCK_ASSERT(sc);
2703
2704         /*
2705          * If vlan is zero then we want to delete for all vlans so the lookup
2706          * may return more than one.
2707          */
2708         while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2709                 bridge_rtnode_destroy(sc, brt);
2710                 found = 1;
2711         }
2712
2713         return (found ? 0 : ENOENT);
2714 }
2715
2716 /*
2717  * bridge_rtdelete:
2718  *
2719  *      Delete routes to a speicifc member interface.
2720  */
2721 static void
2722 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2723 {
2724         struct bridge_rtnode *brt, *nbrt;
2725
2726         BRIDGE_LOCK_ASSERT(sc);
2727
2728         LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2729                 if (brt->brt_ifp == ifp && (full ||
2730                             (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2731                         bridge_rtnode_destroy(sc, brt);
2732         }
2733 }
2734
2735 /*
2736  * bridge_rtable_init:
2737  *
2738  *      Initialize the route table for this bridge.
2739  */
2740 static void
2741 bridge_rtable_init(struct bridge_softc *sc)
2742 {
2743         int i;
2744
2745         sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2746             M_DEVBUF, M_WAITOK);
2747
2748         for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2749                 LIST_INIT(&sc->sc_rthash[i]);
2750
2751         sc->sc_rthash_key = arc4random();
2752         LIST_INIT(&sc->sc_rtlist);
2753 }
2754
2755 /*
2756  * bridge_rtable_fini:
2757  *
2758  *      Deconstruct the route table for this bridge.
2759  */
2760 static void
2761 bridge_rtable_fini(struct bridge_softc *sc)
2762 {
2763
2764         KASSERT(sc->sc_brtcnt == 0,
2765             ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2766         free(sc->sc_rthash, M_DEVBUF);
2767 }
2768
2769 /*
2770  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2771  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2772  */
2773 #define mix(a, b, c)                                                    \
2774 do {                                                                    \
2775         a -= b; a -= c; a ^= (c >> 13);                                 \
2776         b -= c; b -= a; b ^= (a << 8);                                  \
2777         c -= a; c -= b; c ^= (b >> 13);                                 \
2778         a -= b; a -= c; a ^= (c >> 12);                                 \
2779         b -= c; b -= a; b ^= (a << 16);                                 \
2780         c -= a; c -= b; c ^= (b >> 5);                                  \
2781         a -= b; a -= c; a ^= (c >> 3);                                  \
2782         b -= c; b -= a; b ^= (a << 10);                                 \
2783         c -= a; c -= b; c ^= (b >> 15);                                 \
2784 } while (/*CONSTCOND*/0)
2785
2786 static __inline uint32_t
2787 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2788 {
2789         uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2790
2791         b += addr[5] << 8;
2792         b += addr[4];
2793         a += addr[3] << 24;
2794         a += addr[2] << 16;
2795         a += addr[1] << 8;
2796         a += addr[0];
2797
2798         mix(a, b, c);
2799
2800         return (c & BRIDGE_RTHASH_MASK);
2801 }
2802
2803 #undef mix
2804
2805 static int
2806 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2807 {
2808         int i, d;
2809
2810         for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2811                 d = ((int)a[i]) - ((int)b[i]);
2812         }
2813
2814         return (d);
2815 }
2816
2817 /*
2818  * bridge_rtnode_lookup:
2819  *
2820  *      Look up a bridge route node for the specified destination. Compare the
2821  *      vlan id or if zero then just return the first match.
2822  */
2823 static struct bridge_rtnode *
2824 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2825 {
2826         struct bridge_rtnode *brt;
2827         uint32_t hash;
2828         int dir;
2829
2830         BRIDGE_LOCK_ASSERT(sc);
2831
2832         hash = bridge_rthash(sc, addr);
2833         LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2834                 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2835                 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2836                         return (brt);
2837                 if (dir > 0)
2838                         return (NULL);
2839         }
2840
2841         return (NULL);
2842 }
2843
2844 /*
2845  * bridge_rtnode_insert:
2846  *
2847  *      Insert the specified bridge node into the route table.  We
2848  *      assume the entry is not already in the table.
2849  */
2850 static int
2851 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2852 {
2853         struct bridge_rtnode *lbrt;
2854         uint32_t hash;
2855         int dir;
2856
2857         BRIDGE_LOCK_ASSERT(sc);
2858
2859         hash = bridge_rthash(sc, brt->brt_addr);
2860
2861         lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2862         if (lbrt == NULL) {
2863                 LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2864                 goto out;
2865         }
2866
2867         do {
2868                 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2869                 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2870                         return (EEXIST);
2871                 if (dir > 0) {
2872                         LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2873                         goto out;
2874                 }
2875                 if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2876                         LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2877                         goto out;
2878                 }
2879                 lbrt = LIST_NEXT(lbrt, brt_hash);
2880         } while (lbrt != NULL);
2881
2882 #ifdef DIAGNOSTIC
2883         panic("bridge_rtnode_insert: impossible");
2884 #endif
2885
2886 out:
2887         LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2888         sc->sc_brtcnt++;
2889
2890         return (0);
2891 }
2892
2893 /*
2894  * bridge_rtnode_destroy:
2895  *
2896  *      Destroy a bridge rtnode.
2897  */
2898 static void
2899 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2900 {
2901         BRIDGE_LOCK_ASSERT(sc);
2902
2903         LIST_REMOVE(brt, brt_hash);
2904
2905         LIST_REMOVE(brt, brt_list);
2906         sc->sc_brtcnt--;
2907         brt->brt_dst->bif_addrcnt--;
2908         uma_zfree(bridge_rtnode_zone, brt);
2909 }
2910
2911 /*
2912  * bridge_rtable_expire:
2913  *
2914  *      Set the expiry time for all routes on an interface.
2915  */
2916 static void
2917 bridge_rtable_expire(struct ifnet *ifp, int age)
2918 {
2919         struct bridge_softc *sc = ifp->if_bridge;
2920         struct bridge_rtnode *brt;
2921
2922         BRIDGE_LOCK(sc);
2923
2924         /*
2925          * If the age is zero then flush, otherwise set all the expiry times to
2926          * age for the interface
2927          */
2928         if (age == 0)
2929                 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2930         else {
2931                 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2932                         /* Cap the expiry time to 'age' */
2933                         if (brt->brt_ifp == ifp &&
2934                             brt->brt_expire > time_uptime + age &&
2935                             (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2936                                 brt->brt_expire = time_uptime + age;
2937                 }
2938         }
2939         BRIDGE_UNLOCK(sc);
2940 }
2941
2942 /*
2943  * bridge_state_change:
2944  *
2945  *      Callback from the bridgestp code when a port changes states.
2946  */
2947 static void
2948 bridge_state_change(struct ifnet *ifp, int state)
2949 {
2950         struct bridge_softc *sc = ifp->if_bridge;
2951         static const char *stpstates[] = {
2952                 "disabled",
2953                 "listening",
2954                 "learning",
2955                 "forwarding",
2956                 "blocking",
2957                 "discarding"
2958         };
2959
2960         if (log_stp)
2961                 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
2962                     sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
2963 }
2964
2965 /*
2966  * Send bridge packets through pfil if they are one of the types pfil can deal
2967  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
2968  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
2969  * that interface.
2970  */
2971 static int
2972 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
2973 {
2974         int snap, error, i, hlen;
2975         struct ether_header *eh1, eh2;
2976         struct ip *ip;
2977         struct llc llc1;
2978         u_int16_t ether_type;
2979
2980         snap = 0;
2981         error = -1;     /* Default error if not error == 0 */
2982
2983 #if 0
2984         /* we may return with the IP fields swapped, ensure its not shared */
2985         KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
2986 #endif
2987
2988         if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
2989                 return (0); /* filtering is disabled */
2990
2991         i = min((*mp)->m_pkthdr.len, max_protohdr);
2992         if ((*mp)->m_len < i) {
2993             *mp = m_pullup(*mp, i);
2994             if (*mp == NULL) {
2995                 printf("%s: m_pullup failed\n", __func__);
2996                 return (-1);
2997             }
2998         }
2999
3000         eh1 = mtod(*mp, struct ether_header *);
3001         ether_type = ntohs(eh1->ether_type);
3002
3003         /*
3004          * Check for SNAP/LLC.
3005          */
3006         if (ether_type < ETHERMTU) {
3007                 struct llc *llc2 = (struct llc *)(eh1 + 1);
3008
3009                 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3010                     llc2->llc_dsap == LLC_SNAP_LSAP &&
3011                     llc2->llc_ssap == LLC_SNAP_LSAP &&
3012                     llc2->llc_control == LLC_UI) {
3013                         ether_type = htons(llc2->llc_un.type_snap.ether_type);
3014                         snap = 1;
3015                 }
3016         }
3017
3018         /*
3019          * If we're trying to filter bridge traffic, don't look at anything
3020          * other than IP and ARP traffic.  If the filter doesn't understand
3021          * IPv6, don't allow IPv6 through the bridge either.  This is lame
3022          * since if we really wanted, say, an AppleTalk filter, we are hosed,
3023          * but of course we don't have an AppleTalk filter to begin with.
3024          * (Note that since pfil doesn't understand ARP it will pass *ALL*
3025          * ARP traffic.)
3026          */
3027         switch (ether_type) {
3028                 case ETHERTYPE_ARP:
3029                 case ETHERTYPE_REVARP:
3030                         if (pfil_ipfw_arp == 0)
3031                                 return (0); /* Automatically pass */
3032                         break;
3033
3034                 case ETHERTYPE_IP:
3035 #ifdef INET6
3036                 case ETHERTYPE_IPV6:
3037 #endif /* INET6 */
3038                         break;
3039                 default:
3040                         /*
3041                          * Check to see if the user wants to pass non-ip
3042                          * packets, these will not be checked by pfil(9) and
3043                          * passed unconditionally so the default is to drop.
3044                          */
3045                         if (pfil_onlyip)
3046                                 goto bad;
3047         }
3048
3049         /* Run the packet through pfil before stripping link headers */
3050         if (PFIL_HOOKED(&V_link_pfil_hook) && pfil_ipfw != 0 &&
3051                         dir == PFIL_OUT && ifp != NULL) {
3052
3053                 error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, NULL);
3054
3055                 if (*mp == NULL || error != 0) /* packet consumed by filter */
3056                         return (error);
3057         }
3058
3059         /* Strip off the Ethernet header and keep a copy. */
3060         m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3061         m_adj(*mp, ETHER_HDR_LEN);
3062
3063         /* Strip off snap header, if present */
3064         if (snap) {
3065                 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3066                 m_adj(*mp, sizeof(struct llc));
3067         }
3068
3069         /*
3070          * Check the IP header for alignment and errors
3071          */
3072         if (dir == PFIL_IN) {
3073                 switch (ether_type) {
3074                         case ETHERTYPE_IP:
3075                                 error = bridge_ip_checkbasic(mp);
3076                                 break;
3077 #ifdef INET6
3078                         case ETHERTYPE_IPV6:
3079                                 error = bridge_ip6_checkbasic(mp);
3080                                 break;
3081 #endif /* INET6 */
3082                         default:
3083                                 error = 0;
3084                 }
3085                 if (error)
3086                         goto bad;
3087         }
3088
3089         error = 0;
3090
3091         /*
3092          * Run the packet through pfil
3093          */
3094         switch (ether_type) {
3095         case ETHERTYPE_IP:
3096                 /*
3097                  * Run pfil on the member interface and the bridge, both can
3098                  * be skipped by clearing pfil_member or pfil_bridge.
3099                  *
3100                  * Keep the order:
3101                  *   in_if -> bridge_if -> out_if
3102                  */
3103                 if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3104                         error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3105                                         dir, NULL);
3106
3107                 if (*mp == NULL || error != 0) /* filter may consume */
3108                         break;
3109
3110                 if (pfil_member && ifp != NULL)
3111                         error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3112                                         dir, NULL);
3113
3114                 if (*mp == NULL || error != 0) /* filter may consume */
3115                         break;
3116
3117                 if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3118                         error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3119                                         dir, NULL);
3120
3121                 if (*mp == NULL || error != 0) /* filter may consume */
3122                         break;
3123
3124                 /* check if we need to fragment the packet */
3125                 if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3126                         i = (*mp)->m_pkthdr.len;
3127                         if (i > ifp->if_mtu) {
3128                                 error = bridge_fragment(ifp, *mp, &eh2, snap,
3129                                             &llc1);
3130                                 return (error);
3131                         }
3132                 }
3133
3134                 /* Recalculate the ip checksum. */
3135                 ip = mtod(*mp, struct ip *);
3136                 hlen = ip->ip_hl << 2;
3137                 if (hlen < sizeof(struct ip))
3138                         goto bad;
3139                 if (hlen > (*mp)->m_len) {
3140                         if ((*mp = m_pullup(*mp, hlen)) == 0)
3141                                 goto bad;
3142                         ip = mtod(*mp, struct ip *);
3143                         if (ip == NULL)
3144                                 goto bad;
3145                 }
3146                 ip->ip_sum = 0;
3147                 if (hlen == sizeof(struct ip))
3148                         ip->ip_sum = in_cksum_hdr(ip);
3149                 else
3150                         ip->ip_sum = in_cksum(*mp, hlen);
3151
3152                 break;
3153 #ifdef INET6
3154         case ETHERTYPE_IPV6:
3155                 if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3156                         error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3157                                         dir, NULL);
3158
3159                 if (*mp == NULL || error != 0) /* filter may consume */
3160                         break;
3161
3162                 if (pfil_member && ifp != NULL)
3163                         error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3164                                         dir, NULL);
3165
3166                 if (*mp == NULL || error != 0) /* filter may consume */
3167                         break;
3168
3169                 if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3170                         error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3171                                         dir, NULL);
3172                 break;
3173 #endif
3174         default:
3175                 error = 0;
3176                 break;
3177         }
3178
3179         if (*mp == NULL)
3180                 return (error);
3181         if (error != 0)
3182                 goto bad;
3183
3184         error = -1;
3185
3186         /*
3187          * Finally, put everything back the way it was and return
3188          */
3189         if (snap) {
3190                 M_PREPEND(*mp, sizeof(struct llc), M_DONTWAIT);
3191                 if (*mp == NULL)
3192                         return (error);
3193                 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3194         }
3195
3196         M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3197         if (*mp == NULL)
3198                 return (error);
3199         bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3200
3201         return (0);
3202
3203 bad:
3204         m_freem(*mp);
3205         *mp = NULL;
3206         return (error);
3207 }
3208
3209 /*
3210  * Perform basic checks on header size since
3211  * pfil assumes ip_input has already processed
3212  * it for it.  Cut-and-pasted from ip_input.c.
3213  * Given how simple the IPv6 version is,
3214  * does the IPv4 version really need to be
3215  * this complicated?
3216  *
3217  * XXX Should we update ipstat here, or not?
3218  * XXX Right now we update ipstat but not
3219  * XXX csum_counter.
3220  */
3221 static int
3222 bridge_ip_checkbasic(struct mbuf **mp)
3223 {
3224         struct mbuf *m = *mp;
3225         struct ip *ip;
3226         int len, hlen;
3227         u_short sum;
3228
3229         if (*mp == NULL)
3230                 return (-1);
3231
3232         if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3233                 if ((m = m_copyup(m, sizeof(struct ip),
3234                         (max_linkhdr + 3) & ~3)) == NULL) {
3235                         /* XXXJRT new stat, please */
3236                         KMOD_IPSTAT_INC(ips_toosmall);
3237                         goto bad;
3238                 }
3239         } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3240                 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3241                         KMOD_IPSTAT_INC(ips_toosmall);
3242                         goto bad;
3243                 }
3244         }
3245         ip = mtod(m, struct ip *);
3246         if (ip == NULL) goto bad;
3247
3248         if (ip->ip_v != IPVERSION) {
3249                 KMOD_IPSTAT_INC(ips_badvers);
3250                 goto bad;
3251         }
3252         hlen = ip->ip_hl << 2;
3253         if (hlen < sizeof(struct ip)) { /* minimum header length */
3254                 KMOD_IPSTAT_INC(ips_badhlen);
3255                 goto bad;
3256         }
3257         if (hlen > m->m_len) {
3258                 if ((m = m_pullup(m, hlen)) == 0) {
3259                         KMOD_IPSTAT_INC(ips_badhlen);
3260                         goto bad;
3261                 }
3262                 ip = mtod(m, struct ip *);
3263                 if (ip == NULL) goto bad;
3264         }
3265
3266         if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3267                 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3268         } else {
3269                 if (hlen == sizeof(struct ip)) {
3270                         sum = in_cksum_hdr(ip);
3271                 } else {
3272                         sum = in_cksum(m, hlen);
3273                 }
3274         }
3275         if (sum) {
3276                 KMOD_IPSTAT_INC(ips_badsum);
3277                 goto bad;
3278         }
3279
3280         /* Retrieve the packet length. */
3281         len = ntohs(ip->ip_len);
3282
3283         /*
3284          * Check for additional length bogosity
3285          */
3286         if (len < hlen) {
3287                 KMOD_IPSTAT_INC(ips_badlen);
3288                 goto bad;
3289         }
3290
3291         /*
3292          * Check that the amount of data in the buffers
3293          * is as at least much as the IP header would have us expect.
3294          * Drop packet if shorter than we expect.
3295          */
3296         if (m->m_pkthdr.len < len) {
3297                 KMOD_IPSTAT_INC(ips_tooshort);
3298                 goto bad;
3299         }
3300
3301         /* Checks out, proceed */
3302         *mp = m;
3303         return (0);
3304
3305 bad:
3306         *mp = m;
3307         return (-1);
3308 }
3309
3310 #ifdef INET6
3311 /*
3312  * Same as above, but for IPv6.
3313  * Cut-and-pasted from ip6_input.c.
3314  * XXX Should we update ip6stat, or not?
3315  */
3316 static int
3317 bridge_ip6_checkbasic(struct mbuf **mp)
3318 {
3319         struct mbuf *m = *mp;
3320         struct ip6_hdr *ip6;
3321
3322         /*
3323          * If the IPv6 header is not aligned, slurp it up into a new
3324          * mbuf with space for link headers, in the event we forward
3325          * it.  Otherwise, if it is aligned, make sure the entire base
3326          * IPv6 header is in the first mbuf of the chain.
3327          */
3328         if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3329                 struct ifnet *inifp = m->m_pkthdr.rcvif;
3330                 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3331                             (max_linkhdr + 3) & ~3)) == NULL) {
3332                         /* XXXJRT new stat, please */
3333                         V_ip6stat.ip6s_toosmall++;
3334                         in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3335                         goto bad;
3336                 }
3337         } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3338                 struct ifnet *inifp = m->m_pkthdr.rcvif;
3339                 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3340                         V_ip6stat.ip6s_toosmall++;
3341                         in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3342                         goto bad;
3343                 }
3344         }
3345
3346         ip6 = mtod(m, struct ip6_hdr *);
3347
3348         if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3349                 V_ip6stat.ip6s_badvers++;
3350                 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3351                 goto bad;
3352         }
3353
3354         /* Checks out, proceed */
3355         *mp = m;
3356         return (0);
3357
3358 bad:
3359         *mp = m;
3360         return (-1);
3361 }
3362 #endif /* INET6 */
3363
3364 /*
3365  * bridge_fragment:
3366  *
3367  *      Return a fragmented mbuf chain.
3368  */
3369 static int
3370 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3371     int snap, struct llc *llc)
3372 {
3373         struct mbuf *m0;
3374         struct ip *ip;
3375         int error = -1;
3376
3377         if (m->m_len < sizeof(struct ip) &&
3378             (m = m_pullup(m, sizeof(struct ip))) == NULL)
3379                 goto out;
3380         ip = mtod(m, struct ip *);
3381
3382         m->m_pkthdr.csum_flags |= CSUM_IP;
3383         error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3384         if (error)
3385                 goto out;
3386
3387         /* walk the chain and re-add the Ethernet header */
3388         for (m0 = m; m0; m0 = m0->m_nextpkt) {
3389                 if (error == 0) {
3390                         if (snap) {
3391                                 M_PREPEND(m0, sizeof(struct llc), M_DONTWAIT);
3392                                 if (m0 == NULL) {
3393                                         error = ENOBUFS;
3394                                         continue;
3395                                 }
3396                                 bcopy(llc, mtod(m0, caddr_t),
3397                                     sizeof(struct llc));
3398                         }
3399                         M_PREPEND(m0, ETHER_HDR_LEN, M_DONTWAIT);
3400                         if (m0 == NULL) {
3401                                 error = ENOBUFS;
3402                                 continue;
3403                         }
3404                         bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3405                 } else
3406                         m_freem(m);
3407         }
3408
3409         if (error == 0)
3410                 KMOD_IPSTAT_INC(ips_fragmented);
3411
3412         return (error);
3413
3414 out:
3415         if (m != NULL)
3416                 m_freem(m);
3417         return (error);
3418 }
3419
3420 static void
3421 bridge_linkstate(struct ifnet *ifp)
3422 {
3423         struct bridge_softc *sc = ifp->if_bridge;
3424         struct bridge_iflist *bif;
3425
3426         BRIDGE_LOCK(sc);
3427         bif = bridge_lookup_member_if(sc, ifp);
3428         if (bif == NULL) {
3429                 BRIDGE_UNLOCK(sc);
3430                 return;
3431         }
3432         bridge_linkcheck(sc);
3433         BRIDGE_UNLOCK(sc);
3434
3435         bstp_linkstate(&bif->bif_stp);
3436 }
3437
3438 static void
3439 bridge_linkcheck(struct bridge_softc *sc)
3440 {
3441         struct bridge_iflist *bif;
3442         int new_link, hasls;
3443
3444         BRIDGE_LOCK_ASSERT(sc);
3445         new_link = LINK_STATE_DOWN;
3446         hasls = 0;
3447         /* Our link is considered up if at least one of our ports is active */
3448         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3449                 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3450                         hasls++;
3451                 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3452                         new_link = LINK_STATE_UP;
3453                         break;
3454                 }
3455         }
3456         if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3457                 /* If no interfaces support link-state then we default to up */
3458                 new_link = LINK_STATE_UP;
3459         }
3460         if_link_state_change(sc->sc_ifp, new_link);
3461 }