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/eventhandler.h>
  85 #include <sys/mbuf.h>
  86 #include <sys/malloc.h>
  87 #include <sys/protosw.h>
  88 #include <sys/systm.h>
  89 #include <sys/jail.h>
  90 #include <sys/time.h>
  91 #include <sys/socket.h> /* for net/if.h */
  92 #include <sys/sockio.h>
  93 #include <sys/ctype.h>  /* string functions */
  94 #include <sys/kernel.h>
  95 #include <sys/random.h>
  96 #include <sys/syslog.h>
  97 #include <sys/sysctl.h>
  98 #include <vm/uma.h>
  99 #include <sys/module.h>
 100 #include <sys/priv.h>
 101 #include <sys/proc.h>
 102 #include <sys/lock.h>
 103 #include <sys/mutex.h>
 104
 105 #include <net/bpf.h>
 106 #include <net/if.h>
 107 #include <net/if_clone.h>
 108 #include <net/if_dl.h>
 109 #include <net/if_types.h>
 110 #include <net/if_var.h>
 111 #include <net/pfil.h>
 112 #include <net/vnet.h>
 113
 114 #include <netinet/in.h>
 115 #include <netinet/in_systm.h>
 116 #include <netinet/in_var.h>
 117 #include <netinet/ip.h>
 118 #include <netinet/ip_var.h>
 119 #ifdef INET6
 120 #include <netinet/ip6.h>
 121 #include <netinet6/ip6_var.h>
 122 #include <netinet6/in6_ifattach.h>
 123 #endif
 124 #if defined(INET) || defined(INET6)
 125 #include <netinet/ip_carp.h>
 126 #endif
 127 #include <machine/in_cksum.h>
 128 #include <netinet/if_ether.h>
 129 #include <net/bridgestp.h>
 130 #include <net/if_bridgevar.h>
 131 #include <net/if_llc.h>
 132 #include <net/if_vlan_var.h>
 133
 134 #include <net/route.h>
 135
 136 /*
 137  * Size of the route hash table.  Must be a power of two.
 138  */
 139 #ifndef BRIDGE_RTHASH_SIZE
 140 #define BRIDGE_RTHASH_SIZE              1024
 141 #endif
 142
 143 #define BRIDGE_RTHASH_MASK              (BRIDGE_RTHASH_SIZE - 1)
 144
 145 /*
 146  * Default maximum number of addresses to cache.
 147  */
 148 #ifndef BRIDGE_RTABLE_MAX
 149 #define BRIDGE_RTABLE_MAX               2000
 150 #endif
 151
 152 /*
 153  * Timeout (in seconds) for entries learned dynamically.
 154  */
 155 #ifndef BRIDGE_RTABLE_TIMEOUT
 156 #define BRIDGE_RTABLE_TIMEOUT           (20 * 60)       /* same as ARP */
 157 #endif
 158
 159 /*
 160  * Number of seconds between walks of the route list.
 161  */
 162 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
 163 #define BRIDGE_RTABLE_PRUNE_PERIOD      (5 * 60)
 164 #endif
 165
 166 /*
 167  * List of capabilities to possibly mask on the member interface.
 168  */
 169 #define BRIDGE_IFCAPS_MASK              (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
 170
 171 /*
 172  * List of capabilities to strip
 173  */
 174 #define BRIDGE_IFCAPS_STRIP             IFCAP_LRO
 175
 176 /*
 177  * Bridge interface list entry.
 178  */
 179 struct bridge_iflist {
 180         LIST_ENTRY(bridge_iflist) bif_next;
 181         struct ifnet            *bif_ifp;       /* member if */
 182         struct bstp_port        bif_stp;        /* STP state */
 183         uint32_t                bif_flags;      /* member if flags */
 184         int                     bif_savedcaps;  /* saved capabilities */
 185         uint32_t                bif_addrmax;    /* max # of addresses */
 186         uint32_t                bif_addrcnt;    /* cur. # of addresses */
 187         uint32_t                bif_addrexceeded;/* # of address violations */
 188 };
 189
 190 /*
 191  * Bridge route node.
 192  */
 193 struct bridge_rtnode {
 194         LIST_ENTRY(bridge_rtnode) brt_hash;     /* hash table linkage */
 195         LIST_ENTRY(bridge_rtnode) brt_list;     /* list linkage */
 196         struct bridge_iflist    *brt_dst;       /* destination if */
 197         unsigned long           brt_expire;     /* expiration time */
 198         uint8_t                 brt_flags;      /* address flags */
 199         uint8_t                 brt_addr[ETHER_ADDR_LEN];
 200         uint16_t                brt_vlan;       /* vlan id */
 201 };
 202 #define brt_ifp                 brt_dst->bif_ifp
 203
 204 /*
 205  * Software state for each bridge.
 206  */
 207 struct bridge_softc {
 208         struct ifnet            *sc_ifp;        /* make this an interface */
 209         LIST_ENTRY(bridge_softc) sc_list;
 210         struct mtx              sc_mtx;
 211         struct cv               sc_cv;
 212         uint32_t                sc_brtmax;      /* max # of addresses */
 213         uint32_t                sc_brtcnt;      /* cur. # of addresses */
 214         uint32_t                sc_brttimeout;  /* rt timeout in seconds */
 215         struct callout          sc_brcallout;   /* bridge callout */
 216         uint32_t                sc_iflist_ref;  /* refcount for sc_iflist */
 217         uint32_t                sc_iflist_xcnt; /* refcount for sc_iflist */
 218         LIST_HEAD(, bridge_iflist) sc_iflist;   /* member interface list */
 219         LIST_HEAD(, bridge_rtnode) *sc_rthash;  /* our forwarding table */
 220         LIST_HEAD(, bridge_rtnode) sc_rtlist;   /* list version of above */
 221         uint32_t                sc_rthash_key;  /* key for hash */
 222         LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
 223         struct bstp_state       sc_stp;         /* STP state */
 224         uint32_t                sc_brtexceeded; /* # of cache drops */
 225         struct ifnet            *sc_ifaddr;     /* member mac copied from */
 226         u_char                  sc_defaddr[6];  /* Default MAC address */
 227 };
 228
 229 static VNET_DEFINE(struct mtx, bridge_list_mtx);
 230 #define V_bridge_list_mtx       VNET(bridge_list_mtx)
 231 static eventhandler_tag bridge_detach_cookie;
 232
 233 int     bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
 234
 235 uma_zone_t bridge_rtnode_zone;
 236
 237 static int      bridge_clone_create(struct if_clone *, int, caddr_t);
 238 static void     bridge_clone_destroy(struct ifnet *);
 239
 240 static int      bridge_ioctl(struct ifnet *, u_long, caddr_t);
 241 static void     bridge_mutecaps(struct bridge_softc *);
 242 static void     bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
 243                     int);
 244 static void     bridge_ifdetach(void *arg __unused, struct ifnet *);
 245 static void     bridge_init(void *);
 246 static void     bridge_dummynet(struct mbuf *, struct ifnet *);
 247 static void     bridge_stop(struct ifnet *, int);
 248 static int      bridge_transmit(struct ifnet *, struct mbuf *);
 249 static void     bridge_qflush(struct ifnet *);
 250 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
 251 static int      bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
 252                     struct rtentry *);
 253 static int      bridge_enqueue(struct bridge_softc *, struct ifnet *,
 254                     struct mbuf *);
 255 static void     bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
 256
 257 static void     bridge_forward(struct bridge_softc *, struct bridge_iflist *,
 258                     struct mbuf *m);
 259
 260 static void     bridge_timer(void *);
 261
 262 static void     bridge_broadcast(struct bridge_softc *, struct ifnet *,
 263                     struct mbuf *, int);
 264 static void     bridge_span(struct bridge_softc *, struct mbuf *);
 265
 266 static int      bridge_rtupdate(struct bridge_softc *, const uint8_t *,
 267                     uint16_t, struct bridge_iflist *, int, uint8_t);
 268 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
 269                     uint16_t);
 270 static void     bridge_rttrim(struct bridge_softc *);
 271 static void     bridge_rtage(struct bridge_softc *);
 272 static void     bridge_rtflush(struct bridge_softc *, int);
 273 static int      bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
 274                     uint16_t);
 275
 276 static void     bridge_rtable_init(struct bridge_softc *);
 277 static void     bridge_rtable_fini(struct bridge_softc *);
 278
 279 static int      bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
 280 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
 281                     const uint8_t *, uint16_t);
 282 static int      bridge_rtnode_insert(struct bridge_softc *,
 283                     struct bridge_rtnode *);
 284 static void     bridge_rtnode_destroy(struct bridge_softc *,
 285                     struct bridge_rtnode *);
 286 static void     bridge_rtable_expire(struct ifnet *, int);
 287 static void     bridge_state_change(struct ifnet *, int);
 288
 289 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
 290                     const char *name);
 291 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
 292                     struct ifnet *ifp);
 293 static void     bridge_delete_member(struct bridge_softc *,
 294                     struct bridge_iflist *, int);
 295 static void     bridge_delete_span(struct bridge_softc *,
 296                     struct bridge_iflist *);
 297
 298 static int      bridge_ioctl_add(struct bridge_softc *, void *);
 299 static int      bridge_ioctl_del(struct bridge_softc *, void *);
 300 static int      bridge_ioctl_gifflags(struct bridge_softc *, void *);
 301 static int      bridge_ioctl_sifflags(struct bridge_softc *, void *);
 302 static int      bridge_ioctl_scache(struct bridge_softc *, void *);
 303 static int      bridge_ioctl_gcache(struct bridge_softc *, void *);
 304 static int      bridge_ioctl_gifs(struct bridge_softc *, void *);
 305 static int      bridge_ioctl_rts(struct bridge_softc *, void *);
 306 static int      bridge_ioctl_saddr(struct bridge_softc *, void *);
 307 static int      bridge_ioctl_sto(struct bridge_softc *, void *);
 308 static int      bridge_ioctl_gto(struct bridge_softc *, void *);
 309 static int      bridge_ioctl_daddr(struct bridge_softc *, void *);
 310 static int      bridge_ioctl_flush(struct bridge_softc *, void *);
 311 static int      bridge_ioctl_gpri(struct bridge_softc *, void *);
 312 static int      bridge_ioctl_spri(struct bridge_softc *, void *);
 313 static int      bridge_ioctl_ght(struct bridge_softc *, void *);
 314 static int      bridge_ioctl_sht(struct bridge_softc *, void *);
 315 static int      bridge_ioctl_gfd(struct bridge_softc *, void *);
 316 static int      bridge_ioctl_sfd(struct bridge_softc *, void *);
 317 static int      bridge_ioctl_gma(struct bridge_softc *, void *);
 318 static int      bridge_ioctl_sma(struct bridge_softc *, void *);
 319 static int      bridge_ioctl_sifprio(struct bridge_softc *, void *);
 320 static int      bridge_ioctl_sifcost(struct bridge_softc *, void *);
 321 static int      bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
 322 static int      bridge_ioctl_addspan(struct bridge_softc *, void *);
 323 static int      bridge_ioctl_delspan(struct bridge_softc *, void *);
 324 static int      bridge_ioctl_gbparam(struct bridge_softc *, void *);
 325 static int      bridge_ioctl_grte(struct bridge_softc *, void *);
 326 static int      bridge_ioctl_gifsstp(struct bridge_softc *, void *);
 327 static int      bridge_ioctl_sproto(struct bridge_softc *, void *);
 328 static int      bridge_ioctl_stxhc(struct bridge_softc *, void *);
 329 static int      bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
 330                     int);
 331 static int      bridge_ip_checkbasic(struct mbuf **mp);
 332 #ifdef INET6
 333 static int      bridge_ip6_checkbasic(struct mbuf **mp);
 334 #endif /* INET6 */
 335 static int      bridge_fragment(struct ifnet *, struct mbuf *,
 336                     struct ether_header *, int, struct llc *);
 337 static void     bridge_linkstate(struct ifnet *ifp);
 338 static void     bridge_linkcheck(struct bridge_softc *sc);
 339
 340 extern void (*bridge_linkstate_p)(struct ifnet *ifp);
 341
 342 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
 343 #define VLANTAGOF(_m)   \
 344     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
 345
 346 static struct bstp_cb_ops bridge_ops = {
 347         .bcb_state = bridge_state_change,
 348         .bcb_rtage = bridge_rtable_expire
 349 };
 350
 351 SYSCTL_DECL(_net_link);
 352 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
 353
 354 /* only pass IP[46] packets when pfil is enabled */
 355 static VNET_DEFINE(int, pfil_onlyip) = 1;
 356 #define V_pfil_onlyip   VNET(pfil_onlyip)
 357 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
 358     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
 359     "Only pass IP packets when pfil is enabled");
 360
 361 /* run pfil hooks on the bridge interface */
 362 static VNET_DEFINE(int, pfil_bridge) = 1;
 363 #define V_pfil_bridge   VNET(pfil_bridge)
 364 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
 365     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
 366     "Packet filter on the bridge interface");
 367
 368 /* layer2 filter with ipfw */
 369 static VNET_DEFINE(int, pfil_ipfw);
 370 #define V_pfil_ipfw     VNET(pfil_ipfw)
 371
 372 /* layer2 ARP filter with ipfw */
 373 static VNET_DEFINE(int, pfil_ipfw_arp);
 374 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
 375 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
 376     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
 377     "Filter ARP packets through IPFW layer2");
 378
 379 /* run pfil hooks on the member interface */
 380 static VNET_DEFINE(int, pfil_member) = 1;
 381 #define V_pfil_member   VNET(pfil_member)
 382 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
 383     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
 384     "Packet filter on the member interface");
 385
 386 /* run pfil hooks on the physical interface for locally destined packets */
 387 static VNET_DEFINE(int, pfil_local_phys);
 388 #define V_pfil_local_phys       VNET(pfil_local_phys)
 389 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
 390     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
 391     "Packet filter on the physical interface for locally destined packets");
 392
 393 /* log STP state changes */
 394 static VNET_DEFINE(int, log_stp);
 395 #define V_log_stp       VNET(log_stp)
 396 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
 397     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
 398     "Log STP state changes");
 399
 400 /* share MAC with first bridge member */
 401 static VNET_DEFINE(int, bridge_inherit_mac);
 402 #define V_bridge_inherit_mac    VNET(bridge_inherit_mac)
 403 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
 404     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
 405     "Inherit MAC address from the first bridge member");
 406
 407 static VNET_DEFINE(int, allow_llz_overlap) = 0;
 408 #define V_allow_llz_overlap     VNET(allow_llz_overlap)
 409 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
 410     CTLFLAG_VNET | CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
 411     "Allow overlap of link-local scope "
 412     "zones of a bridge interface and the member interfaces");
 413
 414 struct bridge_control {
 415         int     (*bc_func)(struct bridge_softc *, void *);
 416         int     bc_argsize;
 417         int     bc_flags;
 418 };
 419
 420 #define BC_F_COPYIN             0x01    /* copy arguments in */
 421 #define BC_F_COPYOUT            0x02    /* copy arguments out */
 422 #define BC_F_SUSER              0x04    /* do super-user check */
 423
 424 const struct bridge_control bridge_control_table[] = {
 425         { bridge_ioctl_add,             sizeof(struct ifbreq),
 426           BC_F_COPYIN|BC_F_SUSER },
 427         { bridge_ioctl_del,             sizeof(struct ifbreq),
 428           BC_F_COPYIN|BC_F_SUSER },
 429
 430         { bridge_ioctl_gifflags,        sizeof(struct ifbreq),
 431           BC_F_COPYIN|BC_F_COPYOUT },
 432         { bridge_ioctl_sifflags,        sizeof(struct ifbreq),
 433           BC_F_COPYIN|BC_F_SUSER },
 434
 435         { bridge_ioctl_scache,          sizeof(struct ifbrparam),
 436           BC_F_COPYIN|BC_F_SUSER },
 437         { bridge_ioctl_gcache,          sizeof(struct ifbrparam),
 438           BC_F_COPYOUT },
 439
 440         { bridge_ioctl_gifs,            sizeof(struct ifbifconf),
 441           BC_F_COPYIN|BC_F_COPYOUT },
 442         { bridge_ioctl_rts,             sizeof(struct ifbaconf),
 443           BC_F_COPYIN|BC_F_COPYOUT },
 444
 445         { bridge_ioctl_saddr,           sizeof(struct ifbareq),
 446           BC_F_COPYIN|BC_F_SUSER },
 447
 448         { bridge_ioctl_sto,             sizeof(struct ifbrparam),
 449           BC_F_COPYIN|BC_F_SUSER },
 450         { bridge_ioctl_gto,             sizeof(struct ifbrparam),
 451           BC_F_COPYOUT },
 452
 453         { bridge_ioctl_daddr,           sizeof(struct ifbareq),
 454           BC_F_COPYIN|BC_F_SUSER },
 455
 456         { bridge_ioctl_flush,           sizeof(struct ifbreq),
 457           BC_F_COPYIN|BC_F_SUSER },
 458
 459         { bridge_ioctl_gpri,            sizeof(struct ifbrparam),
 460           BC_F_COPYOUT },
 461         { bridge_ioctl_spri,            sizeof(struct ifbrparam),
 462           BC_F_COPYIN|BC_F_SUSER },
 463
 464         { bridge_ioctl_ght,             sizeof(struct ifbrparam),
 465           BC_F_COPYOUT },
 466         { bridge_ioctl_sht,             sizeof(struct ifbrparam),
 467           BC_F_COPYIN|BC_F_SUSER },
 468
 469         { bridge_ioctl_gfd,             sizeof(struct ifbrparam),
 470           BC_F_COPYOUT },
 471         { bridge_ioctl_sfd,             sizeof(struct ifbrparam),
 472           BC_F_COPYIN|BC_F_SUSER },
 473
 474         { bridge_ioctl_gma,             sizeof(struct ifbrparam),
 475           BC_F_COPYOUT },
 476         { bridge_ioctl_sma,             sizeof(struct ifbrparam),
 477           BC_F_COPYIN|BC_F_SUSER },
 478
 479         { bridge_ioctl_sifprio,         sizeof(struct ifbreq),
 480           BC_F_COPYIN|BC_F_SUSER },
 481
 482         { bridge_ioctl_sifcost,         sizeof(struct ifbreq),
 483           BC_F_COPYIN|BC_F_SUSER },
 484
 485         { bridge_ioctl_addspan,         sizeof(struct ifbreq),
 486           BC_F_COPYIN|BC_F_SUSER },
 487         { bridge_ioctl_delspan,         sizeof(struct ifbreq),
 488           BC_F_COPYIN|BC_F_SUSER },
 489
 490         { bridge_ioctl_gbparam,         sizeof(struct ifbropreq),
 491           BC_F_COPYOUT },
 492
 493         { bridge_ioctl_grte,            sizeof(struct ifbrparam),
 494           BC_F_COPYOUT },
 495
 496         { bridge_ioctl_gifsstp,         sizeof(struct ifbpstpconf),
 497           BC_F_COPYIN|BC_F_COPYOUT },
 498
 499         { bridge_ioctl_sproto,          sizeof(struct ifbrparam),
 500           BC_F_COPYIN|BC_F_SUSER },
 501
 502         { bridge_ioctl_stxhc,           sizeof(struct ifbrparam),
 503           BC_F_COPYIN|BC_F_SUSER },
 504
 505         { bridge_ioctl_sifmaxaddr,      sizeof(struct ifbreq),
 506           BC_F_COPYIN|BC_F_SUSER },
 507
 508 };
 509 const int bridge_control_table_size = nitems(bridge_control_table);
 510
 511 static VNET_DEFINE(LIST_HEAD(, bridge_softc), bridge_list);
 512 #define V_bridge_list   VNET(bridge_list)
 513 #define BRIDGE_LIST_LOCK_INIT(x)        mtx_init(&V_bridge_list_mtx,    \
 514                                             "if_bridge list", NULL, MTX_DEF)
 515 #define BRIDGE_LIST_LOCK_DESTROY(x)     mtx_destroy(&V_bridge_list_mtx)
 516 #define BRIDGE_LIST_LOCK(x)             mtx_lock(&V_bridge_list_mtx)
 517 #define BRIDGE_LIST_UNLOCK(x)           mtx_unlock(&V_bridge_list_mtx)
 518
 519 static VNET_DEFINE(struct if_clone *, bridge_cloner);
 520 #define V_bridge_cloner VNET(bridge_cloner)
 521
 522 static const char bridge_name[] = "bridge";
 523
 524 static void
 525 vnet_bridge_init(const void *unused __unused)
 526 {
 527
 528         BRIDGE_LIST_LOCK_INIT();
 529         LIST_INIT(&V_bridge_list);
 530         V_bridge_cloner = if_clone_simple(bridge_name,
 531             bridge_clone_create, bridge_clone_destroy, 0);
 532 }
 533 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
 534     vnet_bridge_init, NULL);
 535
 536 static void
 537 vnet_bridge_uninit(const void *unused __unused)
 538 {
 539
 540         if_clone_detach(V_bridge_cloner);
 541         V_bridge_cloner = NULL;
 542         BRIDGE_LIST_LOCK_DESTROY();
 543 }
 544 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
 545     vnet_bridge_uninit, NULL);
 546
 547 static int
 548 bridge_modevent(module_t mod, int type, void *data)
 549 {
 550
 551         switch (type) {
 552         case MOD_LOAD:
 553                 bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
 554                     sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
 555                     UMA_ALIGN_PTR, 0);
 556                 bridge_input_p = bridge_input;
 557                 bridge_output_p = bridge_output;
 558                 bridge_dn_p = bridge_dummynet;
 559                 bridge_linkstate_p = bridge_linkstate;
 560                 bridge_detach_cookie = EVENTHANDLER_REGISTER(
 561                     ifnet_departure_event, bridge_ifdetach, NULL,
 562                     EVENTHANDLER_PRI_ANY);
 563                 break;
 564         case MOD_UNLOAD:
 565                 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
 566                     bridge_detach_cookie);
 567                 uma_zdestroy(bridge_rtnode_zone);
 568                 bridge_input_p = NULL;
 569                 bridge_output_p = NULL;
 570                 bridge_dn_p = NULL;
 571                 bridge_linkstate_p = NULL;
 572                 break;
 573         default:
 574                 return (EOPNOTSUPP);
 575         }
 576         return (0);
 577 }
 578
 579 static moduledata_t bridge_mod = {
 580         "if_bridge",
 581         bridge_modevent,
 582         0
 583 };
 584
 585 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
 586 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
 587
 588 /*
 589  * handler for net.link.bridge.ipfw
 590  */
 591 static int
 592 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
 593 {
 594         int enable = V_pfil_ipfw;
 595         int error;
 596
 597         error = sysctl_handle_int(oidp, &enable, 0, req);
 598         enable &= 1;
 599
 600         if (enable != V_pfil_ipfw) {
 601                 V_pfil_ipfw = enable;
 602
 603                 /*
 604                  * Disable pfil so that ipfw doesnt run twice, if the user
 605                  * really wants both then they can re-enable pfil_bridge and/or
 606                  * pfil_member. Also allow non-ip packets as ipfw can filter by
 607                  * layer2 type.
 608                  */
 609                 if (V_pfil_ipfw) {
 610                         V_pfil_onlyip = 0;
 611                         V_pfil_bridge = 0;
 612                         V_pfil_member = 0;
 613                 }
 614         }
 615
 616         return (error);
 617 }
 618 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
 619     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET,
 620     &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
 621     "Layer2 filter with IPFW");
 622
 623 /*
 624  * bridge_clone_create:
 625  *
 626  *      Create a new bridge instance.
 627  */
 628 static int
 629 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
 630 {
 631         struct bridge_softc *sc, *sc2;
 632         struct ifnet *bifp, *ifp;
 633         int fb, retry;
 634         unsigned long hostid;
 635
 636         sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
 637         ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
 638         if (ifp == NULL) {
 639                 free(sc, M_DEVBUF);
 640                 return (ENOSPC);
 641         }
 642
 643         BRIDGE_LOCK_INIT(sc);
 644         sc->sc_brtmax = BRIDGE_RTABLE_MAX;
 645         sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
 646
 647         /* Initialize our routing table. */
 648         bridge_rtable_init(sc);
 649
 650         callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
 651
 652         LIST_INIT(&sc->sc_iflist);
 653         LIST_INIT(&sc->sc_spanlist);
 654
 655         ifp->if_softc = sc;
 656         if_initname(ifp, bridge_name, unit);
 657         ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
 658         ifp->if_ioctl = bridge_ioctl;
 659         ifp->if_transmit = bridge_transmit;
 660         ifp->if_qflush = bridge_qflush;
 661         ifp->if_init = bridge_init;
 662         ifp->if_type = IFT_BRIDGE;
 663
 664         /*
 665          * Generate an ethernet address with a locally administered address.
 666          *
 667          * Since we are using random ethernet addresses for the bridge, it is
 668          * possible that we might have address collisions, so make sure that
 669          * this hardware address isn't already in use on another bridge.
 670          * The first try uses the hostid and falls back to arc4rand().
 671          */
 672         fb = 0;
 673         getcredhostid(curthread->td_ucred, &hostid);
 674         do {
 675                 if (fb || hostid == 0) {
 676                         arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
 677                         sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
 678                         sc->sc_defaddr[0] |= 2; /* set the LAA bit */
 679                 } else {
 680                         sc->sc_defaddr[0] = 0x2;
 681                         sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
 682                         sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
 683                         sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
 684                         sc->sc_defaddr[4] =  hostid        & 0xff;
 685                         sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
 686                 }
 687
 688                 fb = 1;
 689                 retry = 0;
 690                 BRIDGE_LIST_LOCK();
 691                 LIST_FOREACH(sc2, &V_bridge_list, sc_list) {
 692                         bifp = sc2->sc_ifp;
 693                         if (memcmp(sc->sc_defaddr,
 694                             IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
 695                                 retry = 1;
 696                                 break;
 697                         }
 698                 }
 699                 BRIDGE_LIST_UNLOCK();
 700         } while (retry == 1);
 701
 702         bstp_attach(&sc->sc_stp, &bridge_ops);
 703         ether_ifattach(ifp, sc->sc_defaddr);
 704         /* Now undo some of the damage... */
 705         ifp->if_baudrate = 0;
 706         ifp->if_type = IFT_BRIDGE;
 707
 708         BRIDGE_LIST_LOCK();
 709         LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
 710         BRIDGE_LIST_UNLOCK();
 711
 712         return (0);
 713 }
 714
 715 /*
 716  * bridge_clone_destroy:
 717  *
 718  *      Destroy a bridge instance.
 719  */
 720 static void
 721 bridge_clone_destroy(struct ifnet *ifp)
 722 {
 723         struct bridge_softc *sc = ifp->if_softc;
 724         struct bridge_iflist *bif;
 725
 726         BRIDGE_LOCK(sc);
 727
 728         bridge_stop(ifp, 1);
 729         ifp->if_flags &= ~IFF_UP;
 730
 731         while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
 732                 bridge_delete_member(sc, bif, 0);
 733
 734         while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
 735                 bridge_delete_span(sc, bif);
 736         }
 737
 738         BRIDGE_UNLOCK(sc);
 739
 740         callout_drain(&sc->sc_brcallout);
 741
 742         BRIDGE_LIST_LOCK();
 743         LIST_REMOVE(sc, sc_list);
 744         BRIDGE_LIST_UNLOCK();
 745
 746         bstp_detach(&sc->sc_stp);
 747         ether_ifdetach(ifp);
 748         if_free(ifp);
 749
 750         /* Tear down the routing table. */
 751         bridge_rtable_fini(sc);
 752
 753         BRIDGE_LOCK_DESTROY(sc);
 754         free(sc, M_DEVBUF);
 755 }
 756
 757 /*
 758  * bridge_ioctl:
 759  *
 760  *      Handle a control request from the operator.
 761  */
 762 static int
 763 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
 764 {
 765         struct bridge_softc *sc = ifp->if_softc;
 766         struct ifreq *ifr = (struct ifreq *)data;
 767         struct bridge_iflist *bif;
 768         struct thread *td = curthread;
 769         union {
 770                 struct ifbreq ifbreq;
 771                 struct ifbifconf ifbifconf;
 772                 struct ifbareq ifbareq;
 773                 struct ifbaconf ifbaconf;
 774                 struct ifbrparam ifbrparam;
 775                 struct ifbropreq ifbropreq;
 776         } args;
 777         struct ifdrv *ifd = (struct ifdrv *) data;
 778         const struct bridge_control *bc;
 779         int error = 0;
 780
 781         switch (cmd) {
 782
 783         case SIOCADDMULTI:
 784         case SIOCDELMULTI:
 785                 break;
 786
 787         case SIOCGDRVSPEC:
 788         case SIOCSDRVSPEC:
 789                 if (ifd->ifd_cmd >= bridge_control_table_size) {
 790                         error = EINVAL;
 791                         break;
 792                 }
 793                 bc = &bridge_control_table[ifd->ifd_cmd];
 794
 795                 if (cmd == SIOCGDRVSPEC &&
 796                     (bc->bc_flags & BC_F_COPYOUT) == 0) {
 797                         error = EINVAL;
 798                         break;
 799                 }
 800                 else if (cmd == SIOCSDRVSPEC &&
 801                     (bc->bc_flags & BC_F_COPYOUT) != 0) {
 802                         error = EINVAL;
 803                         break;
 804                 }
 805
 806                 if (bc->bc_flags & BC_F_SUSER) {
 807                         error = priv_check(td, PRIV_NET_BRIDGE);
 808                         if (error)
 809                                 break;
 810                 }
 811
 812                 if (ifd->ifd_len != bc->bc_argsize ||
 813                     ifd->ifd_len > sizeof(args)) {
 814                         error = EINVAL;
 815                         break;
 816                 }
 817
 818                 bzero(&args, sizeof(args));
 819                 if (bc->bc_flags & BC_F_COPYIN) {
 820                         error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
 821                         if (error)
 822                                 break;
 823                 }
 824
 825                 BRIDGE_LOCK(sc);
 826                 error = (*bc->bc_func)(sc, &args);
 827                 BRIDGE_UNLOCK(sc);
 828                 if (error)
 829                         break;
 830
 831                 if (bc->bc_flags & BC_F_COPYOUT)
 832                         error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
 833
 834                 break;
 835
 836         case SIOCSIFFLAGS:
 837                 if (!(ifp->if_flags & IFF_UP) &&
 838                     (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
 839                         /*
 840                          * If interface is marked down and it is running,
 841                          * then stop and disable it.
 842                          */
 843                         BRIDGE_LOCK(sc);
 844                         bridge_stop(ifp, 1);
 845                         BRIDGE_UNLOCK(sc);
 846                 } else if ((ifp->if_flags & IFF_UP) &&
 847                     !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
 848                         /*
 849                          * If interface is marked up and it is stopped, then
 850                          * start it.
 851                          */
 852                         (*ifp->if_init)(sc);
 853                 }
 854                 break;
 855
 856         case SIOCSIFMTU:
 857                 if (ifr->ifr_mtu < 576) {
 858                         error = EINVAL;
 859                         break;
 860                 }
 861                 if (LIST_EMPTY(&sc->sc_iflist)) {
 862                         sc->sc_ifp->if_mtu = ifr->ifr_mtu;
 863                         break;
 864                 }
 865                 BRIDGE_LOCK(sc);
 866                 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
 867                         if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
 868                                 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
 869                                     " != %d\n", sc->sc_ifp->if_xname,
 870                                     bif->bif_ifp->if_mtu,
 871                                     bif->bif_ifp->if_xname, ifr->ifr_mtu);
 872                                 error = EINVAL;
 873                                 break;
 874                         }
 875                 }
 876                 if (!error)
 877                         sc->sc_ifp->if_mtu = ifr->ifr_mtu;
 878                 BRIDGE_UNLOCK(sc);
 879                 break;
 880         default:
 881                 /*
 882                  * drop the lock as ether_ioctl() will call bridge_start() and
 883                  * cause the lock to be recursed.
 884                  */
 885                 error = ether_ioctl(ifp, cmd, data);
 886                 break;
 887         }
 888
 889         return (error);
 890 }
 891
 892 /*
 893  * bridge_mutecaps:
 894  *
 895  *      Clear or restore unwanted capabilities on the member interface
 896  */
 897 static void
 898 bridge_mutecaps(struct bridge_softc *sc)
 899 {
 900         struct bridge_iflist *bif;
 901         int enabled, mask;
 902
 903         /* Initial bitmask of capabilities to test */
 904         mask = BRIDGE_IFCAPS_MASK;
 905
 906         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
 907                 /* Every member must support it or its disabled */
 908                 mask &= bif->bif_savedcaps;
 909         }
 910
 911         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
 912                 enabled = bif->bif_ifp->if_capenable;
 913                 enabled &= ~BRIDGE_IFCAPS_STRIP;
 914                 /* strip off mask bits and enable them again if allowed */
 915                 enabled &= ~BRIDGE_IFCAPS_MASK;
 916                 enabled |= mask;
 917                 bridge_set_ifcap(sc, bif, enabled);
 918         }
 919
 920 }
 921
 922 static void
 923 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
 924 {
 925         struct ifnet *ifp = bif->bif_ifp;
 926         struct ifreq ifr;
 927         int error;
 928
 929         bzero(&ifr, sizeof(ifr));
 930         ifr.ifr_reqcap = set;
 931
 932         if (ifp->if_capenable != set) {
 933                 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
 934                 if (error)
 935                         if_printf(sc->sc_ifp,
 936                             "error setting interface capabilities on %s\n",
 937                             ifp->if_xname);
 938         }
 939 }
 940
 941 /*
 942  * bridge_lookup_member:
 943  *
 944  *      Lookup a bridge member interface.
 945  */
 946 static struct bridge_iflist *
 947 bridge_lookup_member(struct bridge_softc *sc, const char *name)
 948 {
 949         struct bridge_iflist *bif;
 950         struct ifnet *ifp;
 951
 952         BRIDGE_LOCK_ASSERT(sc);
 953
 954         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
 955                 ifp = bif->bif_ifp;
 956                 if (strcmp(ifp->if_xname, name) == 0)
 957                         return (bif);
 958         }
 959
 960         return (NULL);
 961 }
 962
 963 /*
 964  * bridge_lookup_member_if:
 965  *
 966  *      Lookup a bridge member interface by ifnet*.
 967  */
 968 static struct bridge_iflist *
 969 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
 970 {
 971         struct bridge_iflist *bif;
 972
 973         BRIDGE_LOCK_ASSERT(sc);
 974
 975         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
 976                 if (bif->bif_ifp == member_ifp)
 977                         return (bif);
 978         }
 979
 980         return (NULL);
 981 }
 982
 983 /*
 984  * bridge_delete_member:
 985  *
 986  *      Delete the specified member interface.
 987  */
 988 static void
 989 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
 990     int gone)
 991 {
 992         struct ifnet *ifs = bif->bif_ifp;
 993         struct ifnet *fif = NULL;
 994
 995         BRIDGE_LOCK_ASSERT(sc);
 996
 997         if (bif->bif_flags & IFBIF_STP)
 998                 bstp_disable(&bif->bif_stp);
 999
1000         ifs->if_bridge = NULL;
1001         BRIDGE_XLOCK(sc);
1002         LIST_REMOVE(bif, bif_next);
1003         BRIDGE_XDROP(sc);
1004
1005         /*
1006          * If removing the interface that gave the bridge its mac address, set
1007          * the mac address of the bridge to the address of the next member, or
1008          * to its default address if no members are left.
1009          */
1010         if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1011                 if (LIST_EMPTY(&sc->sc_iflist)) {
1012                         bcopy(sc->sc_defaddr,
1013                             IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1014                         sc->sc_ifaddr = NULL;
1015                 } else {
1016                         fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
1017                         bcopy(IF_LLADDR(fif),
1018                             IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1019                         sc->sc_ifaddr = fif;
1020                 }
1021                 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1022         }
1023
1024         bridge_linkcheck(sc);
1025         bridge_mutecaps(sc);    /* recalcuate now this interface is removed */
1026         bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1027         KASSERT(bif->bif_addrcnt == 0,
1028             ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1029
1030         BRIDGE_UNLOCK(sc);
1031         if (!gone) {
1032                 switch (ifs->if_type) {
1033                 case IFT_ETHER:
1034                 case IFT_L2VLAN:
1035                         /*
1036                          * Take the interface out of promiscuous mode.
1037                          */
1038                         (void) ifpromisc(ifs, 0);
1039                         break;
1040
1041                 case IFT_GIF:
1042                         break;
1043
1044                 default:
1045 #ifdef DIAGNOSTIC
1046                         panic("bridge_delete_member: impossible");
1047 #endif
1048                         break;
1049                 }
1050                 /* reneable any interface capabilities */
1051                 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1052         }
1053         bstp_destroy(&bif->bif_stp);    /* prepare to free */
1054         BRIDGE_LOCK(sc);
1055         free(bif, M_DEVBUF);
1056 }
1057
1058 /*
1059  * bridge_delete_span:
1060  *
1061  *      Delete the specified span interface.
1062  */
1063 static void
1064 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1065 {
1066         BRIDGE_LOCK_ASSERT(sc);
1067
1068         KASSERT(bif->bif_ifp->if_bridge == NULL,
1069             ("%s: not a span interface", __func__));
1070
1071         LIST_REMOVE(bif, bif_next);
1072         free(bif, M_DEVBUF);
1073 }
1074
1075 static int
1076 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1077 {
1078         struct ifbreq *req = arg;
1079         struct bridge_iflist *bif = NULL;
1080         struct ifnet *ifs;
1081         int error = 0;
1082
1083         ifs = ifunit(req->ifbr_ifsname);
1084         if (ifs == NULL)
1085                 return (ENOENT);
1086         if (ifs->if_ioctl == NULL)      /* must be supported */
1087                 return (EINVAL);
1088
1089         /* If it's in the span list, it can't be a member. */
1090         LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1091                 if (ifs == bif->bif_ifp)
1092                         return (EBUSY);
1093
1094         if (ifs->if_bridge == sc)
1095                 return (EEXIST);
1096
1097         if (ifs->if_bridge != NULL)
1098                 return (EBUSY);
1099
1100         switch (ifs->if_type) {
1101         case IFT_ETHER:
1102         case IFT_L2VLAN:
1103         case IFT_GIF:
1104                 /* permitted interface types */
1105                 break;
1106         default:
1107                 return (EINVAL);
1108         }
1109
1110 #ifdef INET6
1111         /*
1112          * Two valid inet6 addresses with link-local scope must not be
1113          * on the parent interface and the member interfaces at the
1114          * same time.  This restriction is needed to prevent violation
1115          * of link-local scope zone.  Attempts to add a member
1116          * interface which has inet6 addresses when the parent has
1117          * inet6 triggers removal of all inet6 addresses on the member
1118          * interface.
1119          */
1120
1121         /* Check if the parent interface has a link-local scope addr. */
1122         if (V_allow_llz_overlap == 0 &&
1123             in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1124                 /*
1125                  * If any, remove all inet6 addresses from the member
1126                  * interfaces.
1127                  */
1128                 BRIDGE_XLOCK(sc);
1129                 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1130                         if (in6ifa_llaonifp(bif->bif_ifp)) {
1131                                 BRIDGE_UNLOCK(sc);
1132                                 in6_ifdetach(bif->bif_ifp);
1133                                 BRIDGE_LOCK(sc);
1134                                 if_printf(sc->sc_ifp,
1135                                     "IPv6 addresses on %s have been removed "
1136                                     "before adding it as a member to prevent "
1137                                     "IPv6 address scope violation.\n",
1138                                     bif->bif_ifp->if_xname);
1139                         }
1140                 }
1141                 BRIDGE_XDROP(sc);
1142                 if (in6ifa_llaonifp(ifs)) {
1143                         BRIDGE_UNLOCK(sc);
1144                         in6_ifdetach(ifs);
1145                         BRIDGE_LOCK(sc);
1146                         if_printf(sc->sc_ifp,
1147                             "IPv6 addresses on %s have been removed "
1148                             "before adding it as a member to prevent "
1149                             "IPv6 address scope violation.\n",
1150                             ifs->if_xname);
1151                 }
1152         }
1153 #endif
1154         /* Allow the first Ethernet member to define the MTU */
1155         if (LIST_EMPTY(&sc->sc_iflist))
1156                 sc->sc_ifp->if_mtu = ifs->if_mtu;
1157         else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1158                 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1159                     ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1160                 return (EINVAL);
1161         }
1162
1163         bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1164         if (bif == NULL)
1165                 return (ENOMEM);
1166
1167         bif->bif_ifp = ifs;
1168         bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1169         bif->bif_savedcaps = ifs->if_capenable;
1170
1171         /*
1172          * Assign the interface's MAC address to the bridge if it's the first
1173          * member and the MAC address of the bridge has not been changed from
1174          * the default randomly generated one.
1175          */
1176         if (V_bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1177             !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1178                 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1179                 sc->sc_ifaddr = ifs;
1180                 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1181         }
1182
1183         ifs->if_bridge = sc;
1184         bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1185         /*
1186          * XXX: XLOCK HERE!?!
1187          *
1188          * NOTE: insert_***HEAD*** should be safe for the traversals.
1189          */
1190         LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1191
1192         /* Set interface capabilities to the intersection set of all members */
1193         bridge_mutecaps(sc);
1194         bridge_linkcheck(sc);
1195
1196         /* Place the interface into promiscuous mode */
1197         switch (ifs->if_type) {
1198                 case IFT_ETHER:
1199                 case IFT_L2VLAN:
1200                         BRIDGE_UNLOCK(sc);
1201                         error = ifpromisc(ifs, 1);
1202                         BRIDGE_LOCK(sc);
1203                         break;
1204         }
1205
1206         if (error) {
1207                 bridge_delete_member(sc, bif, 0);
1208                 free(bif, M_DEVBUF);
1209         }
1210         return (error);
1211 }
1212
1213 static int
1214 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1215 {
1216         struct ifbreq *req = arg;
1217         struct bridge_iflist *bif;
1218
1219         bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1220         if (bif == NULL)
1221                 return (ENOENT);
1222
1223         bridge_delete_member(sc, bif, 0);
1224
1225         return (0);
1226 }
1227
1228 static int
1229 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1230 {
1231         struct ifbreq *req = arg;
1232         struct bridge_iflist *bif;
1233         struct bstp_port *bp;
1234
1235         bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1236         if (bif == NULL)
1237                 return (ENOENT);
1238
1239         bp = &bif->bif_stp;
1240         req->ifbr_ifsflags = bif->bif_flags;
1241         req->ifbr_state = bp->bp_state;
1242         req->ifbr_priority = bp->bp_priority;
1243         req->ifbr_path_cost = bp->bp_path_cost;
1244         req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1245         req->ifbr_proto = bp->bp_protover;
1246         req->ifbr_role = bp->bp_role;
1247         req->ifbr_stpflags = bp->bp_flags;
1248         req->ifbr_addrcnt = bif->bif_addrcnt;
1249         req->ifbr_addrmax = bif->bif_addrmax;
1250         req->ifbr_addrexceeded = bif->bif_addrexceeded;
1251
1252         /* Copy STP state options as flags */
1253         if (bp->bp_operedge)
1254                 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1255         if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1256                 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1257         if (bp->bp_ptp_link)
1258                 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1259         if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1260                 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1261         if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1262                 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1263         if (bp->bp_flags & BSTP_PORT_ADMCOST)
1264                 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1265         return (0);
1266 }
1267
1268 static int
1269 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1270 {
1271         struct ifbreq *req = arg;
1272         struct bridge_iflist *bif;
1273         struct bstp_port *bp;
1274         int error;
1275
1276         bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1277         if (bif == NULL)
1278                 return (ENOENT);
1279         bp = &bif->bif_stp;
1280
1281         if (req->ifbr_ifsflags & IFBIF_SPAN)
1282                 /* SPAN is readonly */
1283                 return (EINVAL);
1284
1285         if (req->ifbr_ifsflags & IFBIF_STP) {
1286                 if ((bif->bif_flags & IFBIF_STP) == 0) {
1287                         error = bstp_enable(&bif->bif_stp);
1288                         if (error)
1289                                 return (error);
1290                 }
1291         } else {
1292                 if ((bif->bif_flags & IFBIF_STP) != 0)
1293                         bstp_disable(&bif->bif_stp);
1294         }
1295
1296         /* Pass on STP flags */
1297         bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1298         bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1299         bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1300         bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1301
1302         /* Save the bits relating to the bridge */
1303         bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1304
1305         return (0);
1306 }
1307
1308 static int
1309 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1310 {
1311         struct ifbrparam *param = arg;
1312
1313         sc->sc_brtmax = param->ifbrp_csize;
1314         bridge_rttrim(sc);
1315
1316         return (0);
1317 }
1318
1319 static int
1320 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1321 {
1322         struct ifbrparam *param = arg;
1323
1324         param->ifbrp_csize = sc->sc_brtmax;
1325
1326         return (0);
1327 }
1328
1329 static int
1330 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1331 {
1332         struct ifbifconf *bifc = arg;
1333         struct bridge_iflist *bif;
1334         struct ifbreq breq;
1335         char *buf, *outbuf;
1336         int count, buflen, len, error = 0;
1337
1338         count = 0;
1339         LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1340                 count++;
1341         LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1342                 count++;
1343
1344         buflen = sizeof(breq) * count;
1345         if (bifc->ifbic_len == 0) {
1346                 bifc->ifbic_len = buflen;
1347                 return (0);
1348         }
1349         BRIDGE_UNLOCK(sc);
1350         outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1351         BRIDGE_LOCK(sc);
1352
1353         count = 0;
1354         buf = outbuf;
1355         len = min(bifc->ifbic_len, buflen);
1356         bzero(&breq, sizeof(breq));
1357         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1358                 if (len < sizeof(breq))
1359                         break;
1360
1361                 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1362                     sizeof(breq.ifbr_ifsname));
1363                 /* Fill in the ifbreq structure */
1364                 error = bridge_ioctl_gifflags(sc, &breq);
1365                 if (error)
1366                         break;
1367                 memcpy(buf, &breq, sizeof(breq));
1368                 count++;
1369                 buf += sizeof(breq);
1370                 len -= sizeof(breq);
1371         }
1372         LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1373                 if (len < sizeof(breq))
1374                         break;
1375
1376                 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1377                     sizeof(breq.ifbr_ifsname));
1378                 breq.ifbr_ifsflags = bif->bif_flags;
1379                 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1380                 memcpy(buf, &breq, sizeof(breq));
1381                 count++;
1382                 buf += sizeof(breq);
1383                 len -= sizeof(breq);
1384         }
1385
1386         BRIDGE_UNLOCK(sc);
1387         bifc->ifbic_len = sizeof(breq) * count;
1388         error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1389         BRIDGE_LOCK(sc);
1390         free(outbuf, M_TEMP);
1391         return (error);
1392 }
1393
1394 static int
1395 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1396 {
1397         struct ifbaconf *bac = arg;
1398         struct bridge_rtnode *brt;
1399         struct ifbareq bareq;
1400         char *buf, *outbuf;
1401         int count, buflen, len, error = 0;
1402
1403         if (bac->ifbac_len == 0)
1404                 return (0);
1405
1406         count = 0;
1407         LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1408                 count++;
1409         buflen = sizeof(bareq) * count;
1410
1411         BRIDGE_UNLOCK(sc);
1412         outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1413         BRIDGE_LOCK(sc);
1414
1415         count = 0;
1416         buf = outbuf;
1417         len = min(bac->ifbac_len, buflen);
1418         bzero(&bareq, sizeof(bareq));
1419         LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1420                 if (len < sizeof(bareq))
1421                         goto out;
1422                 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1423                     sizeof(bareq.ifba_ifsname));
1424                 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1425                 bareq.ifba_vlan = brt->brt_vlan;
1426                 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1427                                 time_uptime < brt->brt_expire)
1428                         bareq.ifba_expire = brt->brt_expire - time_uptime;
1429                 else
1430                         bareq.ifba_expire = 0;
1431                 bareq.ifba_flags = brt->brt_flags;
1432
1433                 memcpy(buf, &bareq, sizeof(bareq));
1434                 count++;
1435                 buf += sizeof(bareq);
1436                 len -= sizeof(bareq);
1437         }
1438 out:
1439         BRIDGE_UNLOCK(sc);
1440         bac->ifbac_len = sizeof(bareq) * count;
1441         error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1442         BRIDGE_LOCK(sc);
1443         free(outbuf, M_TEMP);
1444         return (error);
1445 }
1446
1447 static int
1448 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1449 {
1450         struct ifbareq *req = arg;
1451         struct bridge_iflist *bif;
1452         int error;
1453
1454         bif = bridge_lookup_member(sc, req->ifba_ifsname);
1455         if (bif == NULL)
1456                 return (ENOENT);
1457
1458         error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1459             req->ifba_flags);
1460
1461         return (error);
1462 }
1463
1464 static int
1465 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1466 {
1467         struct ifbrparam *param = arg;
1468
1469         sc->sc_brttimeout = param->ifbrp_ctime;
1470         return (0);
1471 }
1472
1473 static int
1474 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1475 {
1476         struct ifbrparam *param = arg;
1477
1478         param->ifbrp_ctime = sc->sc_brttimeout;
1479         return (0);
1480 }
1481
1482 static int
1483 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1484 {
1485         struct ifbareq *req = arg;
1486
1487         return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1488 }
1489
1490 static int
1491 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1492 {
1493         struct ifbreq *req = arg;
1494
1495         bridge_rtflush(sc, req->ifbr_ifsflags);
1496         return (0);
1497 }
1498
1499 static int
1500 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1501 {
1502         struct ifbrparam *param = arg;
1503         struct bstp_state *bs = &sc->sc_stp;
1504
1505         param->ifbrp_prio = bs->bs_bridge_priority;
1506         return (0);
1507 }
1508
1509 static int
1510 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1511 {
1512         struct ifbrparam *param = arg;
1513
1514         return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1515 }
1516
1517 static int
1518 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1519 {
1520         struct ifbrparam *param = arg;
1521         struct bstp_state *bs = &sc->sc_stp;
1522
1523         param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1524         return (0);
1525 }
1526
1527 static int
1528 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1529 {
1530         struct ifbrparam *param = arg;
1531
1532         return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1533 }
1534
1535 static int
1536 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1537 {
1538         struct ifbrparam *param = arg;
1539         struct bstp_state *bs = &sc->sc_stp;
1540
1541         param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1542         return (0);
1543 }
1544
1545 static int
1546 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1547 {
1548         struct ifbrparam *param = arg;
1549
1550         return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1551 }
1552
1553 static int
1554 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1555 {
1556         struct ifbrparam *param = arg;
1557         struct bstp_state *bs = &sc->sc_stp;
1558
1559         param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1560         return (0);
1561 }
1562
1563 static int
1564 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1565 {
1566         struct ifbrparam *param = arg;
1567
1568         return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1569 }
1570
1571 static int
1572 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1573 {
1574         struct ifbreq *req = arg;
1575         struct bridge_iflist *bif;
1576
1577         bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1578         if (bif == NULL)
1579                 return (ENOENT);
1580
1581         return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1582 }
1583
1584 static int
1585 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1586 {
1587         struct ifbreq *req = arg;
1588         struct bridge_iflist *bif;
1589
1590         bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1591         if (bif == NULL)
1592                 return (ENOENT);
1593
1594         return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1595 }
1596
1597 static int
1598 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1599 {
1600         struct ifbreq *req = arg;
1601         struct bridge_iflist *bif;
1602
1603         bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1604         if (bif == NULL)
1605                 return (ENOENT);
1606
1607         bif->bif_addrmax = req->ifbr_addrmax;
1608         return (0);
1609 }
1610
1611 static int
1612 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1613 {
1614         struct ifbreq *req = arg;
1615         struct bridge_iflist *bif = NULL;
1616         struct ifnet *ifs;
1617
1618         ifs = ifunit(req->ifbr_ifsname);
1619         if (ifs == NULL)
1620                 return (ENOENT);
1621
1622         LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1623                 if (ifs == bif->bif_ifp)
1624                         return (EBUSY);
1625
1626         if (ifs->if_bridge != NULL)
1627                 return (EBUSY);
1628
1629         switch (ifs->if_type) {
1630                 case IFT_ETHER:
1631                 case IFT_GIF:
1632                 case IFT_L2VLAN:
1633                         break;
1634                 default:
1635                         return (EINVAL);
1636         }
1637
1638         bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1639         if (bif == NULL)
1640                 return (ENOMEM);
1641
1642         bif->bif_ifp = ifs;
1643         bif->bif_flags = IFBIF_SPAN;
1644
1645         LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1646
1647         return (0);
1648 }
1649
1650 static int
1651 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1652 {
1653         struct ifbreq *req = arg;
1654         struct bridge_iflist *bif;
1655         struct ifnet *ifs;
1656
1657         ifs = ifunit(req->ifbr_ifsname);
1658         if (ifs == NULL)
1659                 return (ENOENT);
1660
1661         LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1662                 if (ifs == bif->bif_ifp)
1663                         break;
1664
1665         if (bif == NULL)
1666                 return (ENOENT);
1667
1668         bridge_delete_span(sc, bif);
1669
1670         return (0);
1671 }
1672
1673 static int
1674 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1675 {
1676         struct ifbropreq *req = arg;
1677         struct bstp_state *bs = &sc->sc_stp;
1678         struct bstp_port *root_port;
1679
1680         req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1681         req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1682         req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1683
1684         root_port = bs->bs_root_port;
1685         if (root_port == NULL)
1686                 req->ifbop_root_port = 0;
1687         else
1688                 req->ifbop_root_port = root_port->bp_ifp->if_index;
1689
1690         req->ifbop_holdcount = bs->bs_txholdcount;
1691         req->ifbop_priority = bs->bs_bridge_priority;
1692         req->ifbop_protocol = bs->bs_protover;
1693         req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1694         req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1695         req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1696         req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1697         req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1698         req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1699
1700         return (0);
1701 }
1702
1703 static int
1704 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1705 {
1706         struct ifbrparam *param = arg;
1707
1708         param->ifbrp_cexceeded = sc->sc_brtexceeded;
1709         return (0);
1710 }
1711
1712 static int
1713 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1714 {
1715         struct ifbpstpconf *bifstp = arg;
1716         struct bridge_iflist *bif;
1717         struct bstp_port *bp;
1718         struct ifbpstpreq bpreq;
1719         char *buf, *outbuf;
1720         int count, buflen, len, error = 0;
1721
1722         count = 0;
1723         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1724                 if ((bif->bif_flags & IFBIF_STP) != 0)
1725                         count++;
1726         }
1727
1728         buflen = sizeof(bpreq) * count;
1729         if (bifstp->ifbpstp_len == 0) {
1730                 bifstp->ifbpstp_len = buflen;
1731                 return (0);
1732         }
1733
1734         BRIDGE_UNLOCK(sc);
1735         outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1736         BRIDGE_LOCK(sc);
1737
1738         count = 0;
1739         buf = outbuf;
1740         len = min(bifstp->ifbpstp_len, buflen);
1741         bzero(&bpreq, sizeof(bpreq));
1742         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1743                 if (len < sizeof(bpreq))
1744                         break;
1745
1746                 if ((bif->bif_flags & IFBIF_STP) == 0)
1747                         continue;
1748
1749                 bp = &bif->bif_stp;
1750                 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1751                 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1752                 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1753                 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1754                 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1755                 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1756
1757                 memcpy(buf, &bpreq, sizeof(bpreq));
1758                 count++;
1759                 buf += sizeof(bpreq);
1760                 len -= sizeof(bpreq);
1761         }
1762
1763         BRIDGE_UNLOCK(sc);
1764         bifstp->ifbpstp_len = sizeof(bpreq) * count;
1765         error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1766         BRIDGE_LOCK(sc);
1767         free(outbuf, M_TEMP);
1768         return (error);
1769 }
1770
1771 static int
1772 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1773 {
1774         struct ifbrparam *param = arg;
1775
1776         return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1777 }
1778
1779 static int
1780 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1781 {
1782         struct ifbrparam *param = arg;
1783
1784         return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1785 }
1786
1787 /*
1788  * bridge_ifdetach:
1789  *
1790  *      Detach an interface from a bridge.  Called when a member
1791  *      interface is detaching.
1792  */
1793 static void
1794 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1795 {
1796         struct bridge_softc *sc = ifp->if_bridge;
1797         struct bridge_iflist *bif;
1798
1799         if (ifp->if_flags & IFF_RENAMING)
1800                 return;
1801         if (V_bridge_cloner == NULL) {
1802                 /*
1803                  * This detach handler can be called after
1804                  * vnet_bridge_uninit().  Just return in that case.
1805                  */
1806                 return;
1807         }
1808         /* Check if the interface is a bridge member */
1809         if (sc != NULL) {
1810                 BRIDGE_LOCK(sc);
1811
1812                 bif = bridge_lookup_member_if(sc, ifp);
1813                 if (bif != NULL)
1814                         bridge_delete_member(sc, bif, 1);
1815
1816                 BRIDGE_UNLOCK(sc);
1817                 return;
1818         }
1819
1820         /* Check if the interface is a span port */
1821         BRIDGE_LIST_LOCK();
1822         LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1823                 BRIDGE_LOCK(sc);
1824                 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1825                         if (ifp == bif->bif_ifp) {
1826                                 bridge_delete_span(sc, bif);
1827                                 break;
1828                         }
1829
1830                 BRIDGE_UNLOCK(sc);
1831         }
1832         BRIDGE_LIST_UNLOCK();
1833 }
1834
1835 /*
1836  * bridge_init:
1837  *
1838  *      Initialize a bridge interface.
1839  */
1840 static void
1841 bridge_init(void *xsc)
1842 {
1843         struct bridge_softc *sc = (struct bridge_softc *)xsc;
1844         struct ifnet *ifp = sc->sc_ifp;
1845
1846         if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1847                 return;
1848
1849         BRIDGE_LOCK(sc);
1850         callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1851             bridge_timer, sc);
1852
1853         ifp->if_drv_flags |= IFF_DRV_RUNNING;
1854         bstp_init(&sc->sc_stp);         /* Initialize Spanning Tree */
1855
1856         BRIDGE_UNLOCK(sc);
1857 }
1858
1859 /*
1860  * bridge_stop:
1861  *
1862  *      Stop the bridge interface.
1863  */
1864 static void
1865 bridge_stop(struct ifnet *ifp, int disable)
1866 {
1867         struct bridge_softc *sc = ifp->if_softc;
1868
1869         BRIDGE_LOCK_ASSERT(sc);
1870
1871         if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1872                 return;
1873
1874         callout_stop(&sc->sc_brcallout);
1875         bstp_stop(&sc->sc_stp);
1876
1877         bridge_rtflush(sc, IFBF_FLUSHDYN);
1878
1879         ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1880 }
1881
1882 /*
1883  * bridge_enqueue:
1884  *
1885  *      Enqueue a packet on a bridge member interface.
1886  *
1887  */
1888 static int
1889 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1890 {
1891         int len, err = 0;
1892         short mflags;
1893         struct mbuf *m0;
1894
1895         /* We may be sending a fragment so traverse the mbuf */
1896         for (; m; m = m0) {
1897                 m0 = m->m_nextpkt;
1898                 m->m_nextpkt = NULL;
1899                 len = m->m_pkthdr.len;
1900                 mflags = m->m_flags;
1901
1902                 /*
1903                  * If underlying interface can not do VLAN tag insertion itself
1904                  * then attach a packet tag that holds it.
1905                  */
1906                 if ((m->m_flags & M_VLANTAG) &&
1907                     (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1908                         m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1909                         if (m == NULL) {
1910                                 if_printf(dst_ifp,
1911                                     "unable to prepend VLAN header\n");
1912                                 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
1913                                 continue;
1914                         }
1915                         m->m_flags &= ~M_VLANTAG;
1916                 }
1917
1918                 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1919                         m_freem(m0);
1920                         if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
1921                         break;
1922                 }
1923
1924                 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
1925                 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
1926                 if (mflags & M_MCAST)
1927                         if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
1928         }
1929
1930         return (err);
1931 }
1932
1933 /*
1934  * bridge_dummynet:
1935  *
1936  *      Receive a queued packet from dummynet and pass it on to the output
1937  *      interface.
1938  *
1939  *      The mbuf has the Ethernet header already attached.
1940  */
1941 static void
1942 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1943 {
1944         struct bridge_softc *sc;
1945
1946         sc = ifp->if_bridge;
1947
1948         /*
1949          * The packet didnt originate from a member interface. This should only
1950          * ever happen if a member interface is removed while packets are
1951          * queued for it.
1952          */
1953         if (sc == NULL) {
1954                 m_freem(m);
1955                 return;
1956         }
1957
1958         if (PFIL_HOOKED(&V_inet_pfil_hook)
1959 #ifdef INET6
1960             || PFIL_HOOKED(&V_inet6_pfil_hook)
1961 #endif
1962             ) {
1963                 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1964                         return;
1965                 if (m == NULL)
1966                         return;
1967         }
1968
1969         bridge_enqueue(sc, ifp, m);
1970 }
1971
1972 /*
1973  * bridge_output:
1974  *
1975  *      Send output from a bridge member interface.  This
1976  *      performs the bridging function for locally originated
1977  *      packets.
1978  *
1979  *      The mbuf has the Ethernet header already attached.  We must
1980  *      enqueue or free the mbuf before returning.
1981  */
1982 static int
1983 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1984     struct rtentry *rt)
1985 {
1986         struct ether_header *eh;
1987         struct ifnet *dst_if;
1988         struct bridge_softc *sc;
1989         uint16_t vlan;
1990
1991         if (m->m_len < ETHER_HDR_LEN) {
1992                 m = m_pullup(m, ETHER_HDR_LEN);
1993                 if (m == NULL)
1994                         return (0);
1995         }
1996
1997         eh = mtod(m, struct ether_header *);
1998         sc = ifp->if_bridge;
1999         vlan = VLANTAGOF(m);
2000
2001         BRIDGE_LOCK(sc);
2002
2003         /*
2004          * If bridge is down, but the original output interface is up,
2005          * go ahead and send out that interface.  Otherwise, the packet
2006          * is dropped below.
2007          */
2008         if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2009                 dst_if = ifp;
2010                 goto sendunicast;
2011         }
2012
2013         /*
2014          * If the packet is a multicast, or we don't know a better way to
2015          * get there, send to all interfaces.
2016          */
2017         if (ETHER_IS_MULTICAST(eh->ether_dhost))
2018                 dst_if = NULL;
2019         else
2020                 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2021         if (dst_if == NULL) {
2022                 struct bridge_iflist *bif;
2023                 struct mbuf *mc;
2024                 int error = 0, used = 0;
2025
2026                 bridge_span(sc, m);
2027
2028                 BRIDGE_LOCK2REF(sc, error);
2029                 if (error) {
2030                         m_freem(m);
2031                         return (0);
2032                 }
2033
2034                 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2035                         dst_if = bif->bif_ifp;
2036
2037                         if (dst_if->if_type == IFT_GIF)
2038                                 continue;
2039                         if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2040                                 continue;
2041
2042                         /*
2043                          * If this is not the original output interface,
2044                          * and the interface is participating in spanning
2045                          * tree, make sure the port is in a state that
2046                          * allows forwarding.
2047                          */
2048                         if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2049                             bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2050                                 continue;
2051
2052                         if (LIST_NEXT(bif, bif_next) == NULL) {
2053                                 used = 1;
2054                                 mc = m;
2055                         } else {
2056                                 mc = m_copypacket(m, M_NOWAIT);
2057                                 if (mc == NULL) {
2058                                         if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2059                                         continue;
2060                                 }
2061                         }
2062
2063                         bridge_enqueue(sc, dst_if, mc);
2064                 }
2065                 if (used == 0)
2066                         m_freem(m);
2067                 BRIDGE_UNREF(sc);
2068                 return (0);
2069         }
2070
2071 sendunicast:
2072         /*
2073          * XXX Spanning tree consideration here?
2074          */
2075
2076         bridge_span(sc, m);
2077         if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2078                 m_freem(m);
2079                 BRIDGE_UNLOCK(sc);
2080                 return (0);
2081         }
2082
2083         BRIDGE_UNLOCK(sc);
2084         bridge_enqueue(sc, dst_if, m);
2085         return (0);
2086 }
2087
2088 /*
2089  * bridge_transmit:
2090  *
2091  *      Do output on a bridge.
2092  *
2093  */
2094 static int
2095 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2096 {
2097         struct bridge_softc *sc;
2098         struct ether_header *eh;
2099         struct ifnet *dst_if;
2100         int error = 0;
2101
2102         sc = ifp->if_softc;
2103
2104         ETHER_BPF_MTAP(ifp, m);
2105
2106         eh = mtod(m, struct ether_header *);
2107
2108         BRIDGE_LOCK(sc);
2109         if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2110             (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2111                 BRIDGE_UNLOCK(sc);
2112                 error = bridge_enqueue(sc, dst_if, m);
2113         } else
2114                 bridge_broadcast(sc, ifp, m, 0);
2115
2116         return (error);
2117 }
2118
2119 /*
2120  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2121  */
2122 static void
2123 bridge_qflush(struct ifnet *ifp __unused)
2124 {
2125 }
2126
2127 /*
2128  * bridge_forward:
2129  *
2130  *      The forwarding function of the bridge.
2131  *
2132  *      NOTE: Releases the lock on return.
2133  */
2134 static void
2135 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2136     struct mbuf *m)
2137 {
2138         struct bridge_iflist *dbif;
2139         struct ifnet *src_if, *dst_if, *ifp;
2140         struct ether_header *eh;
2141         uint16_t vlan;
2142         uint8_t *dst;
2143         int error;
2144
2145         src_if = m->m_pkthdr.rcvif;
2146         ifp = sc->sc_ifp;
2147
2148         if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2149         if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2150         vlan = VLANTAGOF(m);
2151
2152         if ((sbif->bif_flags & IFBIF_STP) &&
2153             sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2154                 goto drop;
2155
2156         eh = mtod(m, struct ether_header *);
2157         dst = eh->ether_dhost;
2158
2159         /* If the interface is learning, record the address. */
2160         if (sbif->bif_flags & IFBIF_LEARNING) {
2161                 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2162                     sbif, 0, IFBAF_DYNAMIC);
2163                 /*
2164                  * If the interface has addresses limits then deny any source
2165                  * that is not in the cache.
2166                  */
2167                 if (error && sbif->bif_addrmax)
2168                         goto drop;
2169         }
2170
2171         if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2172             sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2173                 goto drop;
2174
2175         /*
2176          * At this point, the port either doesn't participate
2177          * in spanning tree or it is in the forwarding state.
2178          */
2179
2180         /*
2181          * If the packet is unicast, destined for someone on
2182          * "this" side of the bridge, drop it.
2183          */
2184         if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2185                 dst_if = bridge_rtlookup(sc, dst, vlan);
2186                 if (src_if == dst_if)
2187                         goto drop;
2188         } else {
2189                 /*
2190                  * Check if its a reserved multicast address, any address
2191                  * listed in 802.1D section 7.12.6 may not be forwarded by the
2192                  * bridge.
2193                  * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2194                  */
2195                 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2196                     dst[2] == 0xc2 && dst[3] == 0x00 &&
2197                     dst[4] == 0x00 && dst[5] <= 0x0f)
2198                         goto drop;
2199
2200                 /* ...forward it to all interfaces. */
2201                 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2202                 dst_if = NULL;
2203         }
2204
2205         /*
2206          * If we have a destination interface which is a member of our bridge,
2207          * OR this is a unicast packet, push it through the bpf(4) machinery.
2208          * For broadcast or multicast packets, don't bother because it will
2209          * be reinjected into ether_input. We do this before we pass the packets
2210          * through the pfil(9) framework, as it is possible that pfil(9) will
2211          * drop the packet, or possibly modify it, making it difficult to debug
2212          * firewall issues on the bridge.
2213          */
2214         if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2215                 ETHER_BPF_MTAP(ifp, m);
2216
2217         /* run the packet filter */
2218         if (PFIL_HOOKED(&V_inet_pfil_hook)
2219 #ifdef INET6
2220             || PFIL_HOOKED(&V_inet6_pfil_hook)
2221 #endif
2222             ) {
2223                 BRIDGE_UNLOCK(sc);
2224                 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2225                         return;
2226                 if (m == NULL)
2227                         return;
2228                 BRIDGE_LOCK(sc);
2229         }
2230
2231         if (dst_if == NULL) {
2232                 bridge_broadcast(sc, src_if, m, 1);
2233                 return;
2234         }
2235
2236         /*
2237          * At this point, we're dealing with a unicast frame
2238          * going to a different interface.
2239          */
2240         if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2241                 goto drop;
2242
2243         dbif = bridge_lookup_member_if(sc, dst_if);
2244         if (dbif == NULL)
2245                 /* Not a member of the bridge (anymore?) */
2246                 goto drop;
2247
2248         /* Private segments can not talk to each other */
2249         if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2250                 goto drop;
2251
2252         if ((dbif->bif_flags & IFBIF_STP) &&
2253             dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2254                 goto drop;
2255
2256         BRIDGE_UNLOCK(sc);
2257
2258         if (PFIL_HOOKED(&V_inet_pfil_hook)
2259 #ifdef INET6
2260             || PFIL_HOOKED(&V_inet6_pfil_hook)
2261 #endif
2262             ) {
2263                 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2264                         return;
2265                 if (m == NULL)
2266                         return;
2267         }
2268
2269         bridge_enqueue(sc, dst_if, m);
2270         return;
2271
2272 drop:
2273         BRIDGE_UNLOCK(sc);
2274         m_freem(m);
2275 }
2276
2277 /*
2278  * bridge_input:
2279  *
2280  *      Receive input from a member interface.  Queue the packet for
2281  *      bridging if it is not for us.
2282  */
2283 static struct mbuf *
2284 bridge_input(struct ifnet *ifp, struct mbuf *m)
2285 {
2286         struct bridge_softc *sc = ifp->if_bridge;
2287         struct bridge_iflist *bif, *bif2;
2288         struct ifnet *bifp;
2289         struct ether_header *eh;
2290         struct mbuf *mc, *mc2;
2291         uint16_t vlan;
2292         int error;
2293
2294         if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2295                 return (m);
2296
2297         bifp = sc->sc_ifp;
2298         vlan = VLANTAGOF(m);
2299
2300         /*
2301          * Implement support for bridge monitoring. If this flag has been
2302          * set on this interface, discard the packet once we push it through
2303          * the bpf(4) machinery, but before we do, increment the byte and
2304          * packet counters associated with this interface.
2305          */
2306         if ((bifp->if_flags & IFF_MONITOR) != 0) {
2307                 m->m_pkthdr.rcvif  = bifp;
2308                 ETHER_BPF_MTAP(bifp, m);
2309                 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2310                 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2311                 m_freem(m);
2312                 return (NULL);
2313         }
2314         BRIDGE_LOCK(sc);
2315         bif = bridge_lookup_member_if(sc, ifp);
2316         if (bif == NULL) {
2317                 BRIDGE_UNLOCK(sc);
2318                 return (m);
2319         }
2320
2321         eh = mtod(m, struct ether_header *);
2322
2323         bridge_span(sc, m);
2324
2325         if (m->m_flags & (M_BCAST|M_MCAST)) {
2326                 /* Tap off 802.1D packets; they do not get forwarded. */
2327                 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2328                     ETHER_ADDR_LEN) == 0) {
2329                         bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2330                         BRIDGE_UNLOCK(sc);
2331                         return (NULL);
2332                 }
2333
2334                 if ((bif->bif_flags & IFBIF_STP) &&
2335                     bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2336                         BRIDGE_UNLOCK(sc);
2337                         return (m);
2338                 }
2339
2340                 /*
2341                  * Make a deep copy of the packet and enqueue the copy
2342                  * for bridge processing; return the original packet for
2343                  * local processing.
2344                  */
2345                 mc = m_dup(m, M_NOWAIT);
2346                 if (mc == NULL) {
2347                         BRIDGE_UNLOCK(sc);
2348                         return (m);
2349                 }
2350
2351                 /* Perform the bridge forwarding function with the copy. */
2352                 bridge_forward(sc, bif, mc);
2353
2354                 /*
2355                  * Reinject the mbuf as arriving on the bridge so we have a
2356                  * chance at claiming multicast packets. We can not loop back
2357                  * here from ether_input as a bridge is never a member of a
2358                  * bridge.
2359                  */
2360                 KASSERT(bifp->if_bridge == NULL,
2361                     ("loop created in bridge_input"));
2362                 mc2 = m_dup(m, M_NOWAIT);
2363                 if (mc2 != NULL) {
2364                         /* Keep the layer3 header aligned */
2365                         int i = min(mc2->m_pkthdr.len, max_protohdr);
2366                         mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2367                 }
2368                 if (mc2 != NULL) {
2369                         mc2->m_pkthdr.rcvif = bifp;
2370                         (*bifp->if_input)(bifp, mc2);
2371                 }
2372
2373                 /* Return the original packet for local processing. */
2374                 return (m);
2375         }
2376
2377         if ((bif->bif_flags & IFBIF_STP) &&
2378             bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2379                 BRIDGE_UNLOCK(sc);
2380                 return (m);
2381         }
2382
2383 #if (defined(INET) || defined(INET6))
2384 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2385         || ((iface)->if_carp \
2386             && (*carp_forus_p)((iface), eh->ether_dhost))
2387 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2388         || ((iface)->if_carp \
2389             && (*carp_forus_p)((iface), eh->ether_shost))
2390 #else
2391 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2392 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2393 #endif
2394
2395 #ifdef INET6
2396 #   define OR_PFIL_HOOKED_INET6 \
2397         || PFIL_HOOKED(&V_inet6_pfil_hook)
2398 #else
2399 #   define OR_PFIL_HOOKED_INET6
2400 #endif
2401
2402 #define GRAB_OUR_PACKETS(iface) \
2403         if ((iface)->if_type == IFT_GIF) \
2404                 continue; \
2405         /* It is destined for us. */ \
2406         if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2407             OR_CARP_CHECK_WE_ARE_DST((iface))                           \
2408             ) {                                                         \
2409                 if ((iface)->if_type == IFT_BRIDGE) {                   \
2410                         ETHER_BPF_MTAP(iface, m);                       \
2411                         if_inc_counter(iface, IFCOUNTER_IPACKETS, 1);                           \
2412                         if_inc_counter(iface, IFCOUNTER_IBYTES, m->m_pkthdr.len);               \
2413                         /* Filter on the physical interface. */         \
2414                         if (V_pfil_local_phys &&                        \
2415                             (PFIL_HOOKED(&V_inet_pfil_hook)             \
2416                              OR_PFIL_HOOKED_INET6)) {                   \
2417                                 if (bridge_pfil(&m, NULL, ifp,          \
2418                                     PFIL_IN) != 0 || m == NULL) {       \
2419                                         BRIDGE_UNLOCK(sc);              \
2420                                         return (NULL);                  \
2421                                 }                                       \
2422                                 eh = mtod(m, struct ether_header *);    \
2423                         }                                               \
2424                 }                                                       \
2425                 if (bif->bif_flags & IFBIF_LEARNING) {                  \
2426                         error = bridge_rtupdate(sc, eh->ether_shost,    \
2427                             vlan, bif, 0, IFBAF_DYNAMIC);               \
2428                         if (error && bif->bif_addrmax) {                \
2429                                 BRIDGE_UNLOCK(sc);                      \
2430                                 m_freem(m);                             \
2431                                 return (NULL);                          \
2432                         }                                               \
2433                 }                                                       \
2434                 m->m_pkthdr.rcvif = iface;                              \
2435                 BRIDGE_UNLOCK(sc);                                      \
2436                 return (m);                                             \
2437         }                                                               \
2438                                                                         \
2439         /* We just received a packet that we sent out. */               \
2440         if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2441             OR_CARP_CHECK_WE_ARE_SRC((iface))                   \
2442             ) {                                                         \
2443                 BRIDGE_UNLOCK(sc);                                      \
2444                 m_freem(m);                                             \
2445                 return (NULL);                                          \
2446         }
2447
2448         /*
2449          * Unicast.  Make sure it's not for the bridge.
2450          */
2451         do { GRAB_OUR_PACKETS(bifp) } while (0);
2452
2453         /*
2454          * Give a chance for ifp at first priority. This will help when the
2455          * packet comes through the interface like VLAN's with the same MACs
2456          * on several interfaces from the same bridge. This also will save
2457          * some CPU cycles in case the destination interface and the input
2458          * interface (eq ifp) are the same.
2459          */
2460         do { GRAB_OUR_PACKETS(ifp) } while (0);
2461
2462         /* Now check the all bridge members. */
2463         LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2464                 GRAB_OUR_PACKETS(bif2->bif_ifp)
2465         }
2466
2467 #undef OR_CARP_CHECK_WE_ARE_DST
2468 #undef OR_CARP_CHECK_WE_ARE_SRC
2469 #undef OR_PFIL_HOOKED_INET6
2470 #undef GRAB_OUR_PACKETS
2471
2472         /* Perform the bridge forwarding function. */
2473         bridge_forward(sc, bif, m);
2474
2475         return (NULL);
2476 }
2477
2478 /*
2479  * bridge_broadcast:
2480  *
2481  *      Send a frame to all interfaces that are members of
2482  *      the bridge, except for the one on which the packet
2483  *      arrived.
2484  *
2485  *      NOTE: Releases the lock on return.
2486  */
2487 static void
2488 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2489     struct mbuf *m, int runfilt)
2490 {
2491         struct bridge_iflist *dbif, *sbif;
2492         struct mbuf *mc;
2493         struct ifnet *dst_if;
2494         int error = 0, used = 0, i;
2495
2496         sbif = bridge_lookup_member_if(sc, src_if);
2497
2498         BRIDGE_LOCK2REF(sc, error);
2499         if (error) {
2500                 m_freem(m);
2501                 return;
2502         }
2503
2504         /* Filter on the bridge interface before broadcasting */
2505         if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2506 #ifdef INET6
2507             || PFIL_HOOKED(&V_inet6_pfil_hook)
2508 #endif
2509             )) {
2510                 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2511                         goto out;
2512                 if (m == NULL)
2513                         goto out;
2514         }
2515
2516         LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2517                 dst_if = dbif->bif_ifp;
2518                 if (dst_if == src_if)
2519                         continue;
2520
2521                 /* Private segments can not talk to each other */
2522                 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2523                         continue;
2524
2525                 if ((dbif->bif_flags & IFBIF_STP) &&
2526                     dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2527                         continue;
2528
2529                 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2530                     (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2531                         continue;
2532
2533                 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2534                         continue;
2535
2536                 if (LIST_NEXT(dbif, bif_next) == NULL) {
2537                         mc = m;
2538                         used = 1;
2539                 } else {
2540                         mc = m_dup(m, M_NOWAIT);
2541                         if (mc == NULL) {
2542                                 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2543                                 continue;
2544                         }
2545                 }
2546
2547                 /*
2548                  * Filter on the output interface. Pass a NULL bridge interface
2549                  * pointer so we do not redundantly filter on the bridge for
2550                  * each interface we broadcast on.
2551                  */
2552                 if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2553 #ifdef INET6
2554                     || PFIL_HOOKED(&V_inet6_pfil_hook)
2555 #endif
2556                     )) {
2557                         if (used == 0) {
2558                                 /* Keep the layer3 header aligned */
2559                                 i = min(mc->m_pkthdr.len, max_protohdr);
2560                                 mc = m_copyup(mc, i, ETHER_ALIGN);
2561                                 if (mc == NULL) {
2562                                         if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2563                                         continue;
2564                                 }
2565                         }
2566                         if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2567                                 continue;
2568                         if (mc == NULL)
2569                                 continue;
2570                 }
2571
2572                 bridge_enqueue(sc, dst_if, mc);
2573         }
2574         if (used == 0)
2575                 m_freem(m);
2576
2577 out:
2578         BRIDGE_UNREF(sc);
2579 }
2580
2581 /*
2582  * bridge_span:
2583  *
2584  *      Duplicate a packet out one or more interfaces that are in span mode,
2585  *      the original mbuf is unmodified.
2586  */
2587 static void
2588 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2589 {
2590         struct bridge_iflist *bif;
2591         struct ifnet *dst_if;
2592         struct mbuf *mc;
2593
2594         if (LIST_EMPTY(&sc->sc_spanlist))
2595                 return;
2596
2597         LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2598                 dst_if = bif->bif_ifp;
2599
2600                 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2601                         continue;
2602
2603                 mc = m_copypacket(m, M_NOWAIT);
2604                 if (mc == NULL) {
2605                         if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2606                         continue;
2607                 }
2608
2609                 bridge_enqueue(sc, dst_if, mc);
2610         }
2611 }
2612
2613 /*
2614  * bridge_rtupdate:
2615  *
2616  *      Add a bridge routing entry.
2617  */
2618 static int
2619 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2620     struct bridge_iflist *bif, int setflags, uint8_t flags)
2621 {
2622         struct bridge_rtnode *brt;
2623         int error;
2624
2625         BRIDGE_LOCK_ASSERT(sc);
2626
2627         /* Check the source address is valid and not multicast. */
2628         if (ETHER_IS_MULTICAST(dst) ||
2629             (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2630              dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2631                 return (EINVAL);
2632
2633         /* 802.1p frames map to vlan 1 */
2634         if (vlan == 0)
2635                 vlan = 1;
2636
2637         /*
2638          * A route for this destination might already exist.  If so,
2639          * update it, otherwise create a new one.
2640          */
2641         if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2642                 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2643                         sc->sc_brtexceeded++;
2644                         return (ENOSPC);
2645                 }
2646                 /* Check per interface address limits (if enabled) */
2647                 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2648                         bif->bif_addrexceeded++;
2649                         return (ENOSPC);
2650                 }
2651
2652                 /*
2653                  * Allocate a new bridge forwarding node, and
2654                  * initialize the expiration time and Ethernet
2655                  * address.
2656                  */
2657                 brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2658                 if (brt == NULL)
2659                         return (ENOMEM);
2660
2661                 if (bif->bif_flags & IFBIF_STICKY)
2662                         brt->brt_flags = IFBAF_STICKY;
2663                 else
2664                         brt->brt_flags = IFBAF_DYNAMIC;
2665
2666                 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2667                 brt->brt_vlan = vlan;
2668
2669                 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2670                         uma_zfree(bridge_rtnode_zone, brt);
2671                         return (error);
2672                 }
2673                 brt->brt_dst = bif;
2674                 bif->bif_addrcnt++;
2675         }
2676
2677         if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2678             brt->brt_dst != bif) {
2679                 brt->brt_dst->bif_addrcnt--;
2680                 brt->brt_dst = bif;
2681                 brt->brt_dst->bif_addrcnt++;
2682         }
2683
2684         if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2685                 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2686         if (setflags)
2687                 brt->brt_flags = flags;
2688
2689         return (0);
2690 }
2691
2692 /*
2693  * bridge_rtlookup:
2694  *
2695  *      Lookup the destination interface for an address.
2696  */
2697 static struct ifnet *
2698 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2699 {
2700         struct bridge_rtnode *brt;
2701
2702         BRIDGE_LOCK_ASSERT(sc);
2703
2704         if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2705                 return (NULL);
2706
2707         return (brt->brt_ifp);
2708 }
2709
2710 /*
2711  * bridge_rttrim:
2712  *
2713  *      Trim the routine table so that we have a number
2714  *      of routing entries less than or equal to the
2715  *      maximum number.
2716  */
2717 static void
2718 bridge_rttrim(struct bridge_softc *sc)
2719 {
2720         struct bridge_rtnode *brt, *nbrt;
2721
2722         BRIDGE_LOCK_ASSERT(sc);
2723
2724         /* Make sure we actually need to do this. */
2725         if (sc->sc_brtcnt <= sc->sc_brtmax)
2726                 return;
2727
2728         /* Force an aging cycle; this might trim enough addresses. */
2729         bridge_rtage(sc);
2730         if (sc->sc_brtcnt <= sc->sc_brtmax)
2731                 return;
2732
2733         LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2734                 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2735                         bridge_rtnode_destroy(sc, brt);
2736                         if (sc->sc_brtcnt <= sc->sc_brtmax)
2737                                 return;
2738                 }
2739         }
2740 }
2741
2742 /*
2743  * bridge_timer:
2744  *
2745  *      Aging timer for the bridge.
2746  */
2747 static void
2748 bridge_timer(void *arg)
2749 {
2750         struct bridge_softc *sc = arg;
2751
2752         BRIDGE_LOCK_ASSERT(sc);
2753
2754         bridge_rtage(sc);
2755
2756         if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2757                 callout_reset(&sc->sc_brcallout,
2758                     bridge_rtable_prune_period * hz, bridge_timer, sc);
2759 }
2760
2761 /*
2762  * bridge_rtage:
2763  *
2764  *      Perform an aging cycle.
2765  */
2766 static void
2767 bridge_rtage(struct bridge_softc *sc)
2768 {
2769         struct bridge_rtnode *brt, *nbrt;
2770
2771         BRIDGE_LOCK_ASSERT(sc);
2772
2773         LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2774                 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2775                         if (time_uptime >= brt->brt_expire)
2776                                 bridge_rtnode_destroy(sc, brt);
2777                 }
2778         }
2779 }
2780
2781 /*
2782  * bridge_rtflush:
2783  *
2784  *      Remove all dynamic addresses from the bridge.
2785  */
2786 static void
2787 bridge_rtflush(struct bridge_softc *sc, int full)
2788 {
2789         struct bridge_rtnode *brt, *nbrt;
2790
2791         BRIDGE_LOCK_ASSERT(sc);
2792
2793         LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2794                 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2795                         bridge_rtnode_destroy(sc, brt);
2796         }
2797 }
2798
2799 /*
2800  * bridge_rtdaddr:
2801  *
2802  *      Remove an address from the table.
2803  */
2804 static int
2805 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2806 {
2807         struct bridge_rtnode *brt;
2808         int found = 0;
2809
2810         BRIDGE_LOCK_ASSERT(sc);
2811
2812         /*
2813          * If vlan is zero then we want to delete for all vlans so the lookup
2814          * may return more than one.
2815          */
2816         while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2817                 bridge_rtnode_destroy(sc, brt);
2818                 found = 1;
2819         }
2820
2821         return (found ? 0 : ENOENT);
2822 }
2823
2824 /*
2825  * bridge_rtdelete:
2826  *
2827  *      Delete routes to a speicifc member interface.
2828  */
2829 static void
2830 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2831 {
2832         struct bridge_rtnode *brt, *nbrt;
2833
2834         BRIDGE_LOCK_ASSERT(sc);
2835
2836         LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2837                 if (brt->brt_ifp == ifp && (full ||
2838                             (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2839                         bridge_rtnode_destroy(sc, brt);
2840         }
2841 }
2842
2843 /*
2844  * bridge_rtable_init:
2845  *
2846  *      Initialize the route table for this bridge.
2847  */
2848 static void
2849 bridge_rtable_init(struct bridge_softc *sc)
2850 {
2851         int i;
2852
2853         sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2854             M_DEVBUF, M_WAITOK);
2855
2856         for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2857                 LIST_INIT(&sc->sc_rthash[i]);
2858
2859         sc->sc_rthash_key = arc4random();
2860         LIST_INIT(&sc->sc_rtlist);
2861 }
2862
2863 /*
2864  * bridge_rtable_fini:
2865  *
2866  *      Deconstruct the route table for this bridge.
2867  */
2868 static void
2869 bridge_rtable_fini(struct bridge_softc *sc)
2870 {
2871
2872         KASSERT(sc->sc_brtcnt == 0,
2873             ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2874         free(sc->sc_rthash, M_DEVBUF);
2875 }
2876
2877 /*
2878  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2879  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2880  */
2881 #define mix(a, b, c)                                                    \
2882 do {                                                                    \
2883         a -= b; a -= c; a ^= (c >> 13);                                 \
2884         b -= c; b -= a; b ^= (a << 8);                                  \
2885         c -= a; c -= b; c ^= (b >> 13);                                 \
2886         a -= b; a -= c; a ^= (c >> 12);                                 \
2887         b -= c; b -= a; b ^= (a << 16);                                 \
2888         c -= a; c -= b; c ^= (b >> 5);                                  \
2889         a -= b; a -= c; a ^= (c >> 3);                                  \
2890         b -= c; b -= a; b ^= (a << 10);                                 \
2891         c -= a; c -= b; c ^= (b >> 15);                                 \
2892 } while (/*CONSTCOND*/0)
2893
2894 static __inline uint32_t
2895 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2896 {
2897         uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2898
2899         b += addr[5] << 8;
2900         b += addr[4];
2901         a += addr[3] << 24;
2902         a += addr[2] << 16;
2903         a += addr[1] << 8;
2904         a += addr[0];
2905
2906         mix(a, b, c);
2907
2908         return (c & BRIDGE_RTHASH_MASK);
2909 }
2910
2911 #undef mix
2912
2913 static int
2914 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2915 {
2916         int i, d;
2917
2918         for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2919                 d = ((int)a[i]) - ((int)b[i]);
2920         }
2921
2922         return (d);
2923 }
2924
2925 /*
2926  * bridge_rtnode_lookup:
2927  *
2928  *      Look up a bridge route node for the specified destination. Compare the
2929  *      vlan id or if zero then just return the first match.
2930  */
2931 static struct bridge_rtnode *
2932 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2933 {
2934         struct bridge_rtnode *brt;
2935         uint32_t hash;
2936         int dir;
2937
2938         BRIDGE_LOCK_ASSERT(sc);
2939
2940         hash = bridge_rthash(sc, addr);
2941         LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2942                 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2943                 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2944                         return (brt);
2945                 if (dir > 0)
2946                         return (NULL);
2947         }
2948
2949         return (NULL);
2950 }
2951
2952 /*
2953  * bridge_rtnode_insert:
2954  *
2955  *      Insert the specified bridge node into the route table.  We
2956  *      assume the entry is not already in the table.
2957  */
2958 static int
2959 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2960 {
2961         struct bridge_rtnode *lbrt;
2962         uint32_t hash;
2963         int dir;
2964
2965         BRIDGE_LOCK_ASSERT(sc);
2966
2967         hash = bridge_rthash(sc, brt->brt_addr);
2968
2969         lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2970         if (lbrt == NULL) {
2971                 LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2972                 goto out;
2973         }
2974
2975         do {
2976                 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2977                 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2978                         return (EEXIST);
2979                 if (dir > 0) {
2980                         LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2981                         goto out;
2982                 }
2983                 if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2984                         LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2985                         goto out;
2986                 }
2987                 lbrt = LIST_NEXT(lbrt, brt_hash);
2988         } while (lbrt != NULL);
2989
2990 #ifdef DIAGNOSTIC
2991         panic("bridge_rtnode_insert: impossible");
2992 #endif
2993
2994 out:
2995         LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2996         sc->sc_brtcnt++;
2997
2998         return (0);
2999 }
3000
3001 /*
3002  * bridge_rtnode_destroy:
3003  *
3004  *      Destroy a bridge rtnode.
3005  */
3006 static void
3007 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3008 {
3009         BRIDGE_LOCK_ASSERT(sc);
3010
3011         LIST_REMOVE(brt, brt_hash);
3012
3013         LIST_REMOVE(brt, brt_list);
3014         sc->sc_brtcnt--;
3015         brt->brt_dst->bif_addrcnt--;
3016         uma_zfree(bridge_rtnode_zone, brt);
3017 }
3018
3019 /*
3020  * bridge_rtable_expire:
3021  *
3022  *      Set the expiry time for all routes on an interface.
3023  */
3024 static void
3025 bridge_rtable_expire(struct ifnet *ifp, int age)
3026 {
3027         struct bridge_softc *sc = ifp->if_bridge;
3028         struct bridge_rtnode *brt;
3029
3030         BRIDGE_LOCK(sc);
3031
3032         /*
3033          * If the age is zero then flush, otherwise set all the expiry times to
3034          * age for the interface
3035          */
3036         if (age == 0)
3037                 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3038         else {
3039                 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3040                         /* Cap the expiry time to 'age' */
3041                         if (brt->brt_ifp == ifp &&
3042                             brt->brt_expire > time_uptime + age &&
3043                             (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3044                                 brt->brt_expire = time_uptime + age;
3045                 }
3046         }
3047         BRIDGE_UNLOCK(sc);
3048 }
3049
3050 /*
3051  * bridge_state_change:
3052  *
3053  *      Callback from the bridgestp code when a port changes states.
3054  */
3055 static void
3056 bridge_state_change(struct ifnet *ifp, int state)
3057 {
3058         struct bridge_softc *sc = ifp->if_bridge;
3059         static const char *stpstates[] = {
3060                 "disabled",
3061                 "listening",
3062                 "learning",
3063                 "forwarding",
3064                 "blocking",
3065                 "discarding"
3066         };
3067
3068         if (V_log_stp)
3069                 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3070                     sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3071 }
3072
3073 /*
3074  * Send bridge packets through pfil if they are one of the types pfil can deal
3075  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3076  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3077  * that interface.
3078  */
3079 static int
3080 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3081 {
3082         int snap, error, i, hlen;
3083         struct ether_header *eh1, eh2;
3084         struct ip *ip;
3085         struct llc llc1;
3086         u_int16_t ether_type;
3087
3088         snap = 0;
3089         error = -1;     /* Default error if not error == 0 */
3090
3091 #if 0
3092         /* we may return with the IP fields swapped, ensure its not shared */
3093         KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3094 #endif
3095
3096         if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3097                 return (0); /* filtering is disabled */
3098
3099         i = min((*mp)->m_pkthdr.len, max_protohdr);
3100         if ((*mp)->m_len < i) {
3101             *mp = m_pullup(*mp, i);
3102             if (*mp == NULL) {
3103                 printf("%s: m_pullup failed\n", __func__);
3104                 return (-1);
3105             }
3106         }
3107
3108         eh1 = mtod(*mp, struct ether_header *);
3109         ether_type = ntohs(eh1->ether_type);
3110
3111         /*
3112          * Check for SNAP/LLC.
3113          */
3114         if (ether_type < ETHERMTU) {
3115                 struct llc *llc2 = (struct llc *)(eh1 + 1);
3116
3117                 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3118                     llc2->llc_dsap == LLC_SNAP_LSAP &&
3119                     llc2->llc_ssap == LLC_SNAP_LSAP &&
3120                     llc2->llc_control == LLC_UI) {
3121                         ether_type = htons(llc2->llc_un.type_snap.ether_type);
3122                         snap = 1;
3123                 }
3124         }
3125
3126         /*
3127          * If we're trying to filter bridge traffic, don't look at anything
3128          * other than IP and ARP traffic.  If the filter doesn't understand
3129          * IPv6, don't allow IPv6 through the bridge either.  This is lame
3130          * since if we really wanted, say, an AppleTalk filter, we are hosed,
3131          * but of course we don't have an AppleTalk filter to begin with.
3132          * (Note that since pfil doesn't understand ARP it will pass *ALL*
3133          * ARP traffic.)
3134          */
3135         switch (ether_type) {
3136                 case ETHERTYPE_ARP:
3137                 case ETHERTYPE_REVARP:
3138                         if (V_pfil_ipfw_arp == 0)
3139                                 return (0); /* Automatically pass */
3140                         break;
3141
3142                 case ETHERTYPE_IP:
3143 #ifdef INET6
3144                 case ETHERTYPE_IPV6:
3145 #endif /* INET6 */
3146                         break;
3147                 default:
3148                         /*
3149                          * Check to see if the user wants to pass non-ip
3150                          * packets, these will not be checked by pfil(9) and
3151                          * passed unconditionally so the default is to drop.
3152                          */
3153                         if (V_pfil_onlyip)
3154                                 goto bad;
3155         }
3156
3157         /* Run the packet through pfil before stripping link headers */
3158         if (PFIL_HOOKED(&V_link_pfil_hook) && V_pfil_ipfw != 0 &&
3159                         dir == PFIL_OUT && ifp != NULL) {
3160
3161                 error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, NULL);
3162
3163                 if (*mp == NULL || error != 0) /* packet consumed by filter */
3164                         return (error);
3165         }
3166
3167         /* Strip off the Ethernet header and keep a copy. */
3168         m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3169         m_adj(*mp, ETHER_HDR_LEN);
3170
3171         /* Strip off snap header, if present */
3172         if (snap) {
3173                 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3174                 m_adj(*mp, sizeof(struct llc));
3175         }
3176
3177         /*
3178          * Check the IP header for alignment and errors
3179          */
3180         if (dir == PFIL_IN) {
3181                 switch (ether_type) {
3182                         case ETHERTYPE_IP:
3183                                 error = bridge_ip_checkbasic(mp);
3184                                 break;
3185 #ifdef INET6
3186                         case ETHERTYPE_IPV6:
3187                                 error = bridge_ip6_checkbasic(mp);
3188                                 break;
3189 #endif /* INET6 */
3190                         default:
3191                                 error = 0;
3192                 }
3193                 if (error)
3194                         goto bad;
3195         }
3196
3197         error = 0;
3198
3199         /*
3200          * Run the packet through pfil
3201          */
3202         switch (ether_type) {
3203         case ETHERTYPE_IP:
3204                 /*
3205                  * Run pfil on the member interface and the bridge, both can
3206                  * be skipped by clearing pfil_member or pfil_bridge.
3207                  *
3208                  * Keep the order:
3209                  *   in_if -> bridge_if -> out_if
3210                  */
3211                 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3212                         error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3213                                         dir, NULL);
3214
3215                 if (*mp == NULL || error != 0) /* filter may consume */
3216                         break;
3217
3218                 if (V_pfil_member && ifp != NULL)
3219                         error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3220                                         dir, NULL);
3221
3222                 if (*mp == NULL || error != 0) /* filter may consume */
3223                         break;
3224
3225                 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3226                         error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3227                                         dir, NULL);
3228
3229                 if (*mp == NULL || error != 0) /* filter may consume */
3230                         break;
3231
3232                 /* check if we need to fragment the packet */
3233                 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3234                         i = (*mp)->m_pkthdr.len;
3235                         if (i > ifp->if_mtu) {
3236                                 error = bridge_fragment(ifp, *mp, &eh2, snap,
3237                                             &llc1);
3238                                 return (error);
3239                         }
3240                 }
3241
3242                 /* Recalculate the ip checksum. */
3243                 ip = mtod(*mp, struct ip *);
3244                 hlen = ip->ip_hl << 2;
3245                 if (hlen < sizeof(struct ip))
3246                         goto bad;
3247                 if (hlen > (*mp)->m_len) {
3248                         if ((*mp = m_pullup(*mp, hlen)) == 0)
3249                                 goto bad;
3250                         ip = mtod(*mp, struct ip *);
3251                         if (ip == NULL)
3252                                 goto bad;
3253                 }
3254                 ip->ip_sum = 0;
3255                 if (hlen == sizeof(struct ip))
3256                         ip->ip_sum = in_cksum_hdr(ip);
3257                 else
3258                         ip->ip_sum = in_cksum(*mp, hlen);
3259
3260                 break;
3261 #ifdef INET6
3262         case ETHERTYPE_IPV6:
3263                 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3264                         error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3265                                         dir, NULL);
3266
3267                 if (*mp == NULL || error != 0) /* filter may consume */
3268                         break;
3269
3270                 if (V_pfil_member && ifp != NULL)
3271                         error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3272                                         dir, NULL);
3273
3274                 if (*mp == NULL || error != 0) /* filter may consume */
3275                         break;
3276
3277                 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3278                         error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3279                                         dir, NULL);
3280                 break;
3281 #endif
3282         default:
3283                 error = 0;
3284                 break;
3285         }
3286
3287         if (*mp == NULL)
3288                 return (error);
3289         if (error != 0)
3290                 goto bad;
3291
3292         error = -1;
3293
3294         /*
3295          * Finally, put everything back the way it was and return
3296          */
3297         if (snap) {
3298                 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3299                 if (*mp == NULL)
3300                         return (error);
3301                 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3302         }
3303
3304         M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3305         if (*mp == NULL)
3306                 return (error);
3307         bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3308
3309         return (0);
3310
3311 bad:
3312         m_freem(*mp);
3313         *mp = NULL;
3314         return (error);
3315 }
3316
3317 /*
3318  * Perform basic checks on header size since
3319  * pfil assumes ip_input has already processed
3320  * it for it.  Cut-and-pasted from ip_input.c.
3321  * Given how simple the IPv6 version is,
3322  * does the IPv4 version really need to be
3323  * this complicated?
3324  *
3325  * XXX Should we update ipstat here, or not?
3326  * XXX Right now we update ipstat but not
3327  * XXX csum_counter.
3328  */
3329 static int
3330 bridge_ip_checkbasic(struct mbuf **mp)
3331 {
3332         struct mbuf *m = *mp;
3333         struct ip *ip;
3334         int len, hlen;
3335         u_short sum;
3336
3337         if (*mp == NULL)
3338                 return (-1);
3339
3340         if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3341                 if ((m = m_copyup(m, sizeof(struct ip),
3342                         (max_linkhdr + 3) & ~3)) == NULL) {
3343                         /* XXXJRT new stat, please */
3344                         KMOD_IPSTAT_INC(ips_toosmall);
3345                         goto bad;
3346                 }
3347         } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3348                 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3349                         KMOD_IPSTAT_INC(ips_toosmall);
3350                         goto bad;
3351                 }
3352         }
3353         ip = mtod(m, struct ip *);
3354         if (ip == NULL) goto bad;
3355
3356         if (ip->ip_v != IPVERSION) {
3357                 KMOD_IPSTAT_INC(ips_badvers);
3358                 goto bad;
3359         }
3360         hlen = ip->ip_hl << 2;
3361         if (hlen < sizeof(struct ip)) { /* minimum header length */
3362                 KMOD_IPSTAT_INC(ips_badhlen);
3363                 goto bad;
3364         }
3365         if (hlen > m->m_len) {
3366                 if ((m = m_pullup(m, hlen)) == 0) {
3367                         KMOD_IPSTAT_INC(ips_badhlen);
3368                         goto bad;
3369                 }
3370                 ip = mtod(m, struct ip *);
3371                 if (ip == NULL) goto bad;
3372         }
3373
3374         if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3375                 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3376         } else {
3377                 if (hlen == sizeof(struct ip)) {
3378                         sum = in_cksum_hdr(ip);
3379                 } else {
3380                         sum = in_cksum(m, hlen);
3381                 }
3382         }
3383         if (sum) {
3384                 KMOD_IPSTAT_INC(ips_badsum);
3385                 goto bad;
3386         }
3387
3388         /* Retrieve the packet length. */
3389         len = ntohs(ip->ip_len);
3390
3391         /*
3392          * Check for additional length bogosity
3393          */
3394         if (len < hlen) {
3395                 KMOD_IPSTAT_INC(ips_badlen);
3396                 goto bad;
3397         }
3398
3399         /*
3400          * Check that the amount of data in the buffers
3401          * is as at least much as the IP header would have us expect.
3402          * Drop packet if shorter than we expect.
3403          */
3404         if (m->m_pkthdr.len < len) {
3405                 KMOD_IPSTAT_INC(ips_tooshort);
3406                 goto bad;
3407         }
3408
3409         /* Checks out, proceed */
3410         *mp = m;
3411         return (0);
3412
3413 bad:
3414         *mp = m;
3415         return (-1);
3416 }
3417
3418 #ifdef INET6
3419 /*
3420  * Same as above, but for IPv6.
3421  * Cut-and-pasted from ip6_input.c.
3422  * XXX Should we update ip6stat, or not?
3423  */
3424 static int
3425 bridge_ip6_checkbasic(struct mbuf **mp)
3426 {
3427         struct mbuf *m = *mp;
3428         struct ip6_hdr *ip6;
3429
3430         /*
3431          * If the IPv6 header is not aligned, slurp it up into a new
3432          * mbuf with space for link headers, in the event we forward
3433          * it.  Otherwise, if it is aligned, make sure the entire base
3434          * IPv6 header is in the first mbuf of the chain.
3435          */
3436         if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3437                 struct ifnet *inifp = m->m_pkthdr.rcvif;
3438                 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3439                             (max_linkhdr + 3) & ~3)) == NULL) {
3440                         /* XXXJRT new stat, please */
3441                         IP6STAT_INC(ip6s_toosmall);
3442                         in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3443                         goto bad;
3444                 }
3445         } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3446                 struct ifnet *inifp = m->m_pkthdr.rcvif;
3447                 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3448                         IP6STAT_INC(ip6s_toosmall);
3449                         in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3450                         goto bad;
3451                 }
3452         }
3453
3454         ip6 = mtod(m, struct ip6_hdr *);
3455
3456         if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3457                 IP6STAT_INC(ip6s_badvers);
3458                 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3459                 goto bad;
3460         }
3461
3462         /* Checks out, proceed */
3463         *mp = m;
3464         return (0);
3465
3466 bad:
3467         *mp = m;
3468         return (-1);
3469 }
3470 #endif /* INET6 */
3471
3472 /*
3473  * bridge_fragment:
3474  *
3475  *      Return a fragmented mbuf chain.
3476  */
3477 static int
3478 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3479     int snap, struct llc *llc)
3480 {
3481         struct mbuf *m0;
3482         struct ip *ip;
3483         int error = -1;
3484
3485         if (m->m_len < sizeof(struct ip) &&
3486             (m = m_pullup(m, sizeof(struct ip))) == NULL)
3487                 goto out;
3488         ip = mtod(m, struct ip *);
3489
3490         m->m_pkthdr.csum_flags |= CSUM_IP;
3491         error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3492         if (error)
3493                 goto out;
3494
3495         /* walk the chain and re-add the Ethernet header */
3496         for (m0 = m; m0; m0 = m0->m_nextpkt) {
3497                 if (error == 0) {
3498                         if (snap) {
3499                                 M_PREPEND(m0, sizeof(struct llc), M_NOWAIT);
3500                                 if (m0 == NULL) {
3501                                         error = ENOBUFS;
3502                                         continue;
3503                                 }
3504                                 bcopy(llc, mtod(m0, caddr_t),
3505                                     sizeof(struct llc));
3506                         }
3507                         M_PREPEND(m0, ETHER_HDR_LEN, M_NOWAIT);
3508                         if (m0 == NULL) {
3509                                 error = ENOBUFS;
3510                                 continue;
3511                         }
3512                         bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3513                 } else
3514                         m_freem(m);
3515         }
3516
3517         if (error == 0)
3518                 KMOD_IPSTAT_INC(ips_fragmented);
3519
3520         return (error);
3521
3522 out:
3523         if (m != NULL)
3524                 m_freem(m);
3525         return (error);
3526 }
3527
3528 static void
3529 bridge_linkstate(struct ifnet *ifp)
3530 {
3531         struct bridge_softc *sc = ifp->if_bridge;
3532         struct bridge_iflist *bif;
3533
3534         BRIDGE_LOCK(sc);
3535         bif = bridge_lookup_member_if(sc, ifp);
3536         if (bif == NULL) {
3537                 BRIDGE_UNLOCK(sc);
3538                 return;
3539         }
3540         bridge_linkcheck(sc);
3541         BRIDGE_UNLOCK(sc);
3542
3543         bstp_linkstate(&bif->bif_stp);
3544 }
3545
3546 static void
3547 bridge_linkcheck(struct bridge_softc *sc)
3548 {
3549         struct bridge_iflist *bif;
3550         int new_link, hasls;
3551
3552         BRIDGE_LOCK_ASSERT(sc);
3553         new_link = LINK_STATE_DOWN;
3554         hasls = 0;
3555         /* Our link is considered up if at least one of our ports is active */
3556         LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3557                 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3558                         hasls++;
3559                 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3560                         new_link = LINK_STATE_UP;
3561                         break;
3562                 }
3563         }
3564         if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3565                 /* If no interfaces support link-state then we default to up */
3566                 new_link = LINK_STATE_UP;
3567         }
3568         if_link_state_change(sc->sc_ifp, new_link);
3569 }