1 /* $NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $ */ 2 3 /*- 4 * SPDX-License-Identifier: BSD-4-Clause 5 * 6 * Copyright 2001 Wasabi Systems, Inc. 7 * All rights reserved. 8 * 9 * Written by Jason R. Thorpe for Wasabi Systems, Inc. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. All advertising materials mentioning features or use of this software 20 * must display the following acknowledgement: 21 * This product includes software developed for the NetBSD Project by 22 * Wasabi Systems, Inc. 23 * 4. The name of Wasabi Systems, Inc. may not be used to endorse 24 * or promote products derived from this software without specific prior 25 * written permission. 26 * 27 * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND 28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 30 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC 31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 37 * POSSIBILITY OF SUCH DAMAGE. 38 */ 39 40 /* 41 * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net) 42 * All rights reserved. 43 * 44 * Redistribution and use in source and binary forms, with or without 45 * modification, are permitted provided that the following conditions 46 * are met: 47 * 1. Redistributions of source code must retain the above copyright 48 * notice, this list of conditions and the following disclaimer. 49 * 2. Redistributions in binary form must reproduce the above copyright 50 * notice, this list of conditions and the following disclaimer in the 51 * documentation and/or other materials provided with the distribution. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 54 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED 55 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 56 * DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, 57 * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 58 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR 59 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, 61 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN 62 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 63 * POSSIBILITY OF SUCH DAMAGE. 64 * 65 * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp 66 */ 67 68 /* 69 * Network interface bridge support. 70 * 71 * TODO: 72 * 73 * - Currently only supports Ethernet-like interfaces (Ethernet, 74 * 802.11, VLANs on Ethernet, etc.) Figure out a nice way 75 * to bridge other types of interfaces (maybe consider 76 * heterogeneous bridges). 77 */ 78 79 #include <sys/cdefs.h> 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/if_private.h> 112 #include <net/pfil.h> 113 #include <net/vnet.h> 114 115 #include <netinet/in.h> 116 #include <netinet/in_systm.h> 117 #include <netinet/in_var.h> 118 #include <netinet/ip.h> 119 #include <netinet/ip_var.h> 120 #ifdef INET6 121 #include <netinet/ip6.h> 122 #include <netinet6/ip6_var.h> 123 #include <netinet6/in6_ifattach.h> 124 #endif 125 #if defined(INET) || defined(INET6) 126 #include <netinet/ip_carp.h> 127 #endif 128 #include <machine/in_cksum.h> 129 #include <netinet/if_ether.h> 130 #include <net/bridgestp.h> 131 #include <net/if_bridgevar.h> 132 #include <net/if_llc.h> 133 #include <net/if_vlan_var.h> 134 135 #include <net/route.h> 136 137 /* 138 * Size of the route hash table. Must be a power of two. 139 */ 140 #ifndef BRIDGE_RTHASH_SIZE 141 #define BRIDGE_RTHASH_SIZE 1024 142 #endif 143 144 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1) 145 146 /* 147 * Default maximum number of addresses to cache. 148 */ 149 #ifndef BRIDGE_RTABLE_MAX 150 #define BRIDGE_RTABLE_MAX 2000 151 #endif 152 153 /* 154 * Timeout (in seconds) for entries learned dynamically. 155 */ 156 #ifndef BRIDGE_RTABLE_TIMEOUT 157 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */ 158 #endif 159 160 /* 161 * Number of seconds between walks of the route list. 162 */ 163 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD 164 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60) 165 #endif 166 167 /* 168 * List of capabilities to possibly mask on the member interface. 169 */ 170 #define BRIDGE_IFCAPS_MASK (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\ 171 IFCAP_TXCSUM_IPV6) 172 173 /* 174 * List of capabilities to strip 175 */ 176 #define BRIDGE_IFCAPS_STRIP IFCAP_LRO 177 178 /* 179 * Bridge locking 180 * 181 * The bridge relies heavily on the epoch(9) system to protect its data 182 * structures. This means we can safely use CK_LISTs while in NET_EPOCH, but we 183 * must ensure there is only one writer at a time. 184 * 185 * That is: for read accesses we only need to be in NET_EPOCH, but for write 186 * accesses we must hold: 187 * 188 * - BRIDGE_RT_LOCK, for any change to bridge_rtnodes 189 * - BRIDGE_LOCK, for any other change 190 * 191 * The BRIDGE_LOCK is a sleepable lock, because it is held across ioctl() 192 * calls to bridge member interfaces and these ioctl()s can sleep. 193 * The BRIDGE_RT_LOCK is a non-sleepable mutex, because it is sometimes 194 * required while we're in NET_EPOCH and then we're not allowed to sleep. 195 */ 196 #define BRIDGE_LOCK_INIT(_sc) do { \ 197 sx_init(&(_sc)->sc_sx, "if_bridge"); \ 198 mtx_init(&(_sc)->sc_rt_mtx, "if_bridge rt", NULL, MTX_DEF); \ 199 } while (0) 200 #define BRIDGE_LOCK_DESTROY(_sc) do { \ 201 sx_destroy(&(_sc)->sc_sx); \ 202 mtx_destroy(&(_sc)->sc_rt_mtx); \ 203 } while (0) 204 #define BRIDGE_LOCK(_sc) sx_xlock(&(_sc)->sc_sx) 205 #define BRIDGE_UNLOCK(_sc) sx_xunlock(&(_sc)->sc_sx) 206 #define BRIDGE_LOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_XLOCKED) 207 #define BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(_sc) \ 208 MPASS(in_epoch(net_epoch_preempt) || sx_xlocked(&(_sc)->sc_sx)) 209 #define BRIDGE_UNLOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_UNLOCKED) 210 #define BRIDGE_RT_LOCK(_sc) mtx_lock(&(_sc)->sc_rt_mtx) 211 #define BRIDGE_RT_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_rt_mtx) 212 #define BRIDGE_RT_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_rt_mtx, MA_OWNED) 213 #define BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(_sc) \ 214 MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(_sc)->sc_rt_mtx)) 215 216 /* 217 * Bridge interface list entry. 218 */ 219 struct bridge_iflist { 220 CK_LIST_ENTRY(bridge_iflist) bif_next; 221 struct ifnet *bif_ifp; /* member if */ 222 struct bstp_port bif_stp; /* STP state */ 223 uint32_t bif_flags; /* member if flags */ 224 int bif_savedcaps; /* saved capabilities */ 225 uint32_t bif_addrmax; /* max # of addresses */ 226 uint32_t bif_addrcnt; /* cur. # of addresses */ 227 uint32_t bif_addrexceeded;/* # of address violations */ 228 struct epoch_context bif_epoch_ctx; 229 }; 230 231 /* 232 * Bridge route node. 233 */ 234 struct bridge_rtnode { 235 CK_LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */ 236 CK_LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */ 237 struct bridge_iflist *brt_dst; /* destination if */ 238 unsigned long brt_expire; /* expiration time */ 239 uint8_t brt_flags; /* address flags */ 240 uint8_t brt_addr[ETHER_ADDR_LEN]; 241 uint16_t brt_vlan; /* vlan id */ 242 struct vnet *brt_vnet; 243 struct epoch_context brt_epoch_ctx; 244 }; 245 #define brt_ifp brt_dst->bif_ifp 246 247 /* 248 * Software state for each bridge. 249 */ 250 struct bridge_softc { 251 struct ifnet *sc_ifp; /* make this an interface */ 252 LIST_ENTRY(bridge_softc) sc_list; 253 struct sx sc_sx; 254 struct mtx sc_rt_mtx; 255 uint32_t sc_brtmax; /* max # of addresses */ 256 uint32_t sc_brtcnt; /* cur. # of addresses */ 257 uint32_t sc_brttimeout; /* rt timeout in seconds */ 258 struct callout sc_brcallout; /* bridge callout */ 259 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */ 260 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */ 261 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */ 262 uint32_t sc_rthash_key; /* key for hash */ 263 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */ 264 struct bstp_state sc_stp; /* STP state */ 265 uint32_t sc_brtexceeded; /* # of cache drops */ 266 struct ifnet *sc_ifaddr; /* member mac copied from */ 267 struct ether_addr sc_defaddr; /* Default MAC address */ 268 if_input_fn_t sc_if_input; /* Saved copy of if_input */ 269 struct epoch_context sc_epoch_ctx; 270 }; 271 272 VNET_DEFINE_STATIC(struct sx, bridge_list_sx); 273 #define V_bridge_list_sx VNET(bridge_list_sx) 274 static eventhandler_tag bridge_detach_cookie; 275 276 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD; 277 278 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone); 279 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone) 280 281 static int bridge_clone_create(struct if_clone *, char *, size_t, 282 struct ifc_data *, struct ifnet **); 283 static int bridge_clone_destroy(struct if_clone *, struct ifnet *, uint32_t); 284 285 static int bridge_ioctl(struct ifnet *, u_long, caddr_t); 286 static void bridge_mutecaps(struct bridge_softc *); 287 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *, 288 int); 289 static void bridge_ifdetach(void *arg __unused, struct ifnet *); 290 static void bridge_init(void *); 291 static void bridge_dummynet(struct mbuf *, struct ifnet *); 292 static void bridge_stop(struct ifnet *, int); 293 static int bridge_transmit(struct ifnet *, struct mbuf *); 294 #ifdef ALTQ 295 static void bridge_altq_start(if_t); 296 static int bridge_altq_transmit(if_t, struct mbuf *); 297 #endif 298 static void bridge_qflush(struct ifnet *); 299 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *); 300 static void bridge_inject(struct ifnet *, struct mbuf *); 301 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *, 302 struct rtentry *); 303 static int bridge_enqueue(struct bridge_softc *, struct ifnet *, 304 struct mbuf *); 305 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int); 306 307 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *, 308 struct mbuf *m); 309 310 static void bridge_timer(void *); 311 312 static void bridge_broadcast(struct bridge_softc *, struct ifnet *, 313 struct mbuf *, int); 314 static void bridge_span(struct bridge_softc *, struct mbuf *); 315 316 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *, 317 uint16_t, struct bridge_iflist *, int, uint8_t); 318 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *, 319 uint16_t); 320 static void bridge_rttrim(struct bridge_softc *); 321 static void bridge_rtage(struct bridge_softc *); 322 static void bridge_rtflush(struct bridge_softc *, int); 323 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *, 324 uint16_t); 325 326 static void bridge_rtable_init(struct bridge_softc *); 327 static void bridge_rtable_fini(struct bridge_softc *); 328 329 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *); 330 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *, 331 const uint8_t *, uint16_t); 332 static int bridge_rtnode_insert(struct bridge_softc *, 333 struct bridge_rtnode *); 334 static void bridge_rtnode_destroy(struct bridge_softc *, 335 struct bridge_rtnode *); 336 static void bridge_rtable_expire(struct ifnet *, int); 337 static void bridge_state_change(struct ifnet *, int); 338 339 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *, 340 const char *name); 341 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *, 342 struct ifnet *ifp); 343 static void bridge_delete_member(struct bridge_softc *, 344 struct bridge_iflist *, int); 345 static void bridge_delete_span(struct bridge_softc *, 346 struct bridge_iflist *); 347 348 static int bridge_ioctl_add(struct bridge_softc *, void *); 349 static int bridge_ioctl_del(struct bridge_softc *, void *); 350 static int bridge_ioctl_gifflags(struct bridge_softc *, void *); 351 static int bridge_ioctl_sifflags(struct bridge_softc *, void *); 352 static int bridge_ioctl_scache(struct bridge_softc *, void *); 353 static int bridge_ioctl_gcache(struct bridge_softc *, void *); 354 static int bridge_ioctl_gifs(struct bridge_softc *, void *); 355 static int bridge_ioctl_rts(struct bridge_softc *, void *); 356 static int bridge_ioctl_saddr(struct bridge_softc *, void *); 357 static int bridge_ioctl_sto(struct bridge_softc *, void *); 358 static int bridge_ioctl_gto(struct bridge_softc *, void *); 359 static int bridge_ioctl_daddr(struct bridge_softc *, void *); 360 static int bridge_ioctl_flush(struct bridge_softc *, void *); 361 static int bridge_ioctl_gpri(struct bridge_softc *, void *); 362 static int bridge_ioctl_spri(struct bridge_softc *, void *); 363 static int bridge_ioctl_ght(struct bridge_softc *, void *); 364 static int bridge_ioctl_sht(struct bridge_softc *, void *); 365 static int bridge_ioctl_gfd(struct bridge_softc *, void *); 366 static int bridge_ioctl_sfd(struct bridge_softc *, void *); 367 static int bridge_ioctl_gma(struct bridge_softc *, void *); 368 static int bridge_ioctl_sma(struct bridge_softc *, void *); 369 static int bridge_ioctl_sifprio(struct bridge_softc *, void *); 370 static int bridge_ioctl_sifcost(struct bridge_softc *, void *); 371 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *); 372 static int bridge_ioctl_addspan(struct bridge_softc *, void *); 373 static int bridge_ioctl_delspan(struct bridge_softc *, void *); 374 static int bridge_ioctl_gbparam(struct bridge_softc *, void *); 375 static int bridge_ioctl_grte(struct bridge_softc *, void *); 376 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *); 377 static int bridge_ioctl_sproto(struct bridge_softc *, void *); 378 static int bridge_ioctl_stxhc(struct bridge_softc *, void *); 379 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *, 380 int); 381 static int bridge_ip_checkbasic(struct mbuf **mp); 382 #ifdef INET6 383 static int bridge_ip6_checkbasic(struct mbuf **mp); 384 #endif /* INET6 */ 385 static int bridge_fragment(struct ifnet *, struct mbuf **mp, 386 struct ether_header *, int, struct llc *); 387 static void bridge_linkstate(struct ifnet *ifp); 388 static void bridge_linkcheck(struct bridge_softc *sc); 389 390 /* 391 * Use the "null" value from IEEE 802.1Q-2014 Table 9-2 392 * to indicate untagged frames. 393 */ 394 #define VLANTAGOF(_m) \ 395 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : DOT1Q_VID_NULL 396 397 static struct bstp_cb_ops bridge_ops = { 398 .bcb_state = bridge_state_change, 399 .bcb_rtage = bridge_rtable_expire 400 }; 401 402 SYSCTL_DECL(_net_link); 403 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, 404 "Bridge"); 405 406 /* only pass IP[46] packets when pfil is enabled */ 407 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1; 408 #define V_pfil_onlyip VNET(pfil_onlyip) 409 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, 410 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0, 411 "Only pass IP packets when pfil is enabled"); 412 413 /* run pfil hooks on the bridge interface */ 414 VNET_DEFINE_STATIC(int, pfil_bridge) = 0; 415 #define V_pfil_bridge VNET(pfil_bridge) 416 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, 417 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0, 418 "Packet filter on the bridge interface"); 419 420 /* layer2 filter with ipfw */ 421 VNET_DEFINE_STATIC(int, pfil_ipfw); 422 #define V_pfil_ipfw VNET(pfil_ipfw) 423 424 /* layer2 ARP filter with ipfw */ 425 VNET_DEFINE_STATIC(int, pfil_ipfw_arp); 426 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp) 427 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, 428 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0, 429 "Filter ARP packets through IPFW layer2"); 430 431 /* run pfil hooks on the member interface */ 432 VNET_DEFINE_STATIC(int, pfil_member) = 0; 433 #define V_pfil_member VNET(pfil_member) 434 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, 435 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0, 436 "Packet filter on the member interface"); 437 438 /* run pfil hooks on the physical interface for locally destined packets */ 439 VNET_DEFINE_STATIC(int, pfil_local_phys); 440 #define V_pfil_local_phys VNET(pfil_local_phys) 441 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, 442 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0, 443 "Packet filter on the physical interface for locally destined packets"); 444 445 /* log STP state changes */ 446 VNET_DEFINE_STATIC(int, log_stp); 447 #define V_log_stp VNET(log_stp) 448 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, 449 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0, 450 "Log STP state changes"); 451 452 /* share MAC with first bridge member */ 453 VNET_DEFINE_STATIC(int, bridge_inherit_mac); 454 #define V_bridge_inherit_mac VNET(bridge_inherit_mac) 455 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, 456 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0, 457 "Inherit MAC address from the first bridge member"); 458 459 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0; 460 #define V_allow_llz_overlap VNET(allow_llz_overlap) 461 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap, 462 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0, 463 "Allow overlap of link-local scope " 464 "zones of a bridge interface and the member interfaces"); 465 466 /* log MAC address port flapping */ 467 VNET_DEFINE_STATIC(bool, log_mac_flap) = true; 468 #define V_log_mac_flap VNET(log_mac_flap) 469 SYSCTL_BOOL(_net_link_bridge, OID_AUTO, log_mac_flap, 470 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(log_mac_flap), true, 471 "Log MAC address port flapping"); 472 473 VNET_DEFINE_STATIC(int, log_interval) = 5; 474 VNET_DEFINE_STATIC(int, log_count) = 0; 475 VNET_DEFINE_STATIC(struct timeval, log_last) = { 0 }; 476 477 #define V_log_interval VNET(log_interval) 478 #define V_log_count VNET(log_count) 479 #define V_log_last VNET(log_last) 480 481 struct bridge_control { 482 int (*bc_func)(struct bridge_softc *, void *); 483 int bc_argsize; 484 int bc_flags; 485 }; 486 487 #define BC_F_COPYIN 0x01 /* copy arguments in */ 488 #define BC_F_COPYOUT 0x02 /* copy arguments out */ 489 #define BC_F_SUSER 0x04 /* do super-user check */ 490 491 static const struct bridge_control bridge_control_table[] = { 492 { bridge_ioctl_add, sizeof(struct ifbreq), 493 BC_F_COPYIN|BC_F_SUSER }, 494 { bridge_ioctl_del, sizeof(struct ifbreq), 495 BC_F_COPYIN|BC_F_SUSER }, 496 497 { bridge_ioctl_gifflags, sizeof(struct ifbreq), 498 BC_F_COPYIN|BC_F_COPYOUT }, 499 { bridge_ioctl_sifflags, sizeof(struct ifbreq), 500 BC_F_COPYIN|BC_F_SUSER }, 501 502 { bridge_ioctl_scache, sizeof(struct ifbrparam), 503 BC_F_COPYIN|BC_F_SUSER }, 504 { bridge_ioctl_gcache, sizeof(struct ifbrparam), 505 BC_F_COPYOUT }, 506 507 { bridge_ioctl_gifs, sizeof(struct ifbifconf), 508 BC_F_COPYIN|BC_F_COPYOUT }, 509 { bridge_ioctl_rts, sizeof(struct ifbaconf), 510 BC_F_COPYIN|BC_F_COPYOUT }, 511 512 { bridge_ioctl_saddr, sizeof(struct ifbareq), 513 BC_F_COPYIN|BC_F_SUSER }, 514 515 { bridge_ioctl_sto, sizeof(struct ifbrparam), 516 BC_F_COPYIN|BC_F_SUSER }, 517 { bridge_ioctl_gto, sizeof(struct ifbrparam), 518 BC_F_COPYOUT }, 519 520 { bridge_ioctl_daddr, sizeof(struct ifbareq), 521 BC_F_COPYIN|BC_F_SUSER }, 522 523 { bridge_ioctl_flush, sizeof(struct ifbreq), 524 BC_F_COPYIN|BC_F_SUSER }, 525 526 { bridge_ioctl_gpri, sizeof(struct ifbrparam), 527 BC_F_COPYOUT }, 528 { bridge_ioctl_spri, sizeof(struct ifbrparam), 529 BC_F_COPYIN|BC_F_SUSER }, 530 531 { bridge_ioctl_ght, sizeof(struct ifbrparam), 532 BC_F_COPYOUT }, 533 { bridge_ioctl_sht, sizeof(struct ifbrparam), 534 BC_F_COPYIN|BC_F_SUSER }, 535 536 { bridge_ioctl_gfd, sizeof(struct ifbrparam), 537 BC_F_COPYOUT }, 538 { bridge_ioctl_sfd, sizeof(struct ifbrparam), 539 BC_F_COPYIN|BC_F_SUSER }, 540 541 { bridge_ioctl_gma, sizeof(struct ifbrparam), 542 BC_F_COPYOUT }, 543 { bridge_ioctl_sma, sizeof(struct ifbrparam), 544 BC_F_COPYIN|BC_F_SUSER }, 545 546 { bridge_ioctl_sifprio, sizeof(struct ifbreq), 547 BC_F_COPYIN|BC_F_SUSER }, 548 549 { bridge_ioctl_sifcost, sizeof(struct ifbreq), 550 BC_F_COPYIN|BC_F_SUSER }, 551 552 { bridge_ioctl_addspan, sizeof(struct ifbreq), 553 BC_F_COPYIN|BC_F_SUSER }, 554 { bridge_ioctl_delspan, sizeof(struct ifbreq), 555 BC_F_COPYIN|BC_F_SUSER }, 556 557 { bridge_ioctl_gbparam, sizeof(struct ifbropreq), 558 BC_F_COPYOUT }, 559 560 { bridge_ioctl_grte, sizeof(struct ifbrparam), 561 BC_F_COPYOUT }, 562 563 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf), 564 BC_F_COPYIN|BC_F_COPYOUT }, 565 566 { bridge_ioctl_sproto, sizeof(struct ifbrparam), 567 BC_F_COPYIN|BC_F_SUSER }, 568 569 { bridge_ioctl_stxhc, sizeof(struct ifbrparam), 570 BC_F_COPYIN|BC_F_SUSER }, 571 572 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq), 573 BC_F_COPYIN|BC_F_SUSER }, 574 575 }; 576 static const int bridge_control_table_size = nitems(bridge_control_table); 577 578 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list); 579 #define V_bridge_list VNET(bridge_list) 580 #define BRIDGE_LIST_LOCK_INIT(x) sx_init(&V_bridge_list_sx, \ 581 "if_bridge list") 582 #define BRIDGE_LIST_LOCK_DESTROY(x) sx_destroy(&V_bridge_list_sx) 583 #define BRIDGE_LIST_LOCK(x) sx_xlock(&V_bridge_list_sx) 584 #define BRIDGE_LIST_UNLOCK(x) sx_xunlock(&V_bridge_list_sx) 585 586 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner); 587 #define V_bridge_cloner VNET(bridge_cloner) 588 589 static const char bridge_name[] = "bridge"; 590 591 static void 592 vnet_bridge_init(const void *unused __unused) 593 { 594 595 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode", 596 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL, 597 UMA_ALIGN_PTR, 0); 598 BRIDGE_LIST_LOCK_INIT(); 599 LIST_INIT(&V_bridge_list); 600 601 struct if_clone_addreq req = { 602 .create_f = bridge_clone_create, 603 .destroy_f = bridge_clone_destroy, 604 .flags = IFC_F_AUTOUNIT, 605 }; 606 V_bridge_cloner = ifc_attach_cloner(bridge_name, &req); 607 } 608 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY, 609 vnet_bridge_init, NULL); 610 611 static void 612 vnet_bridge_uninit(const void *unused __unused) 613 { 614 615 ifc_detach_cloner(V_bridge_cloner); 616 V_bridge_cloner = NULL; 617 BRIDGE_LIST_LOCK_DESTROY(); 618 619 /* Callbacks may use the UMA zone. */ 620 NET_EPOCH_DRAIN_CALLBACKS(); 621 622 uma_zdestroy(V_bridge_rtnode_zone); 623 } 624 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY, 625 vnet_bridge_uninit, NULL); 626 627 static int 628 bridge_modevent(module_t mod, int type, void *data) 629 { 630 631 switch (type) { 632 case MOD_LOAD: 633 bridge_dn_p = bridge_dummynet; 634 bridge_detach_cookie = EVENTHANDLER_REGISTER( 635 ifnet_departure_event, bridge_ifdetach, NULL, 636 EVENTHANDLER_PRI_ANY); 637 break; 638 case MOD_UNLOAD: 639 EVENTHANDLER_DEREGISTER(ifnet_departure_event, 640 bridge_detach_cookie); 641 bridge_dn_p = NULL; 642 break; 643 default: 644 return (EOPNOTSUPP); 645 } 646 return (0); 647 } 648 649 static moduledata_t bridge_mod = { 650 "if_bridge", 651 bridge_modevent, 652 0 653 }; 654 655 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY); 656 MODULE_VERSION(if_bridge, 1); 657 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1); 658 659 /* 660 * handler for net.link.bridge.ipfw 661 */ 662 static int 663 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS) 664 { 665 int enable = V_pfil_ipfw; 666 int error; 667 668 error = sysctl_handle_int(oidp, &enable, 0, req); 669 enable &= 1; 670 671 if (enable != V_pfil_ipfw) { 672 V_pfil_ipfw = enable; 673 674 /* 675 * Disable pfil so that ipfw doesnt run twice, if the user 676 * really wants both then they can re-enable pfil_bridge and/or 677 * pfil_member. Also allow non-ip packets as ipfw can filter by 678 * layer2 type. 679 */ 680 if (V_pfil_ipfw) { 681 V_pfil_onlyip = 0; 682 V_pfil_bridge = 0; 683 V_pfil_member = 0; 684 } 685 } 686 687 return (error); 688 } 689 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, 690 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT, 691 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I", 692 "Layer2 filter with IPFW"); 693 694 #ifdef VIMAGE 695 static void 696 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg) 697 { 698 struct bridge_softc *sc = ifp->if_softc; 699 struct bridge_iflist *bif; 700 701 BRIDGE_LOCK(sc); 702 703 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL) 704 bridge_delete_member(sc, bif, 0); 705 706 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) { 707 bridge_delete_span(sc, bif); 708 } 709 710 BRIDGE_UNLOCK(sc); 711 712 ether_reassign(ifp, newvnet, arg); 713 } 714 #endif 715 716 /* 717 * bridge_clone_create: 718 * 719 * Create a new bridge instance. 720 */ 721 static int 722 bridge_clone_create(struct if_clone *ifc, char *name, size_t len, 723 struct ifc_data *ifd, struct ifnet **ifpp) 724 { 725 struct bridge_softc *sc; 726 struct ifnet *ifp; 727 728 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO); 729 ifp = sc->sc_ifp = if_alloc(IFT_ETHER); 730 if (ifp == NULL) { 731 free(sc, M_DEVBUF); 732 return (ENOSPC); 733 } 734 735 BRIDGE_LOCK_INIT(sc); 736 sc->sc_brtmax = BRIDGE_RTABLE_MAX; 737 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT; 738 739 /* Initialize our routing table. */ 740 bridge_rtable_init(sc); 741 742 callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0); 743 744 CK_LIST_INIT(&sc->sc_iflist); 745 CK_LIST_INIT(&sc->sc_spanlist); 746 747 ifp->if_softc = sc; 748 if_initname(ifp, bridge_name, ifd->unit); 749 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; 750 ifp->if_ioctl = bridge_ioctl; 751 #ifdef ALTQ 752 ifp->if_start = bridge_altq_start; 753 ifp->if_transmit = bridge_altq_transmit; 754 IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen); 755 ifp->if_snd.ifq_drv_maxlen = 0; 756 IFQ_SET_READY(&ifp->if_snd); 757 #else 758 ifp->if_transmit = bridge_transmit; 759 #endif 760 ifp->if_qflush = bridge_qflush; 761 ifp->if_init = bridge_init; 762 ifp->if_type = IFT_BRIDGE; 763 764 ether_gen_addr(ifp, &sc->sc_defaddr); 765 766 bstp_attach(&sc->sc_stp, &bridge_ops); 767 ether_ifattach(ifp, sc->sc_defaddr.octet); 768 /* Now undo some of the damage... */ 769 ifp->if_baudrate = 0; 770 ifp->if_type = IFT_BRIDGE; 771 #ifdef VIMAGE 772 ifp->if_reassign = bridge_reassign; 773 #endif 774 sc->sc_if_input = ifp->if_input; /* ether_input */ 775 ifp->if_input = bridge_inject; 776 777 /* 778 * Allow BRIDGE_INPUT() to pass in packets originating from the bridge 779 * itself via bridge_inject(). This is required for netmap but 780 * otherwise has no effect. 781 */ 782 ifp->if_bridge_input = bridge_input; 783 784 BRIDGE_LIST_LOCK(); 785 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list); 786 BRIDGE_LIST_UNLOCK(); 787 *ifpp = ifp; 788 789 return (0); 790 } 791 792 static void 793 bridge_clone_destroy_cb(struct epoch_context *ctx) 794 { 795 struct bridge_softc *sc; 796 797 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx); 798 799 BRIDGE_LOCK_DESTROY(sc); 800 free(sc, M_DEVBUF); 801 } 802 803 /* 804 * bridge_clone_destroy: 805 * 806 * Destroy a bridge instance. 807 */ 808 static int 809 bridge_clone_destroy(struct if_clone *ifc, struct ifnet *ifp, uint32_t flags) 810 { 811 struct bridge_softc *sc = ifp->if_softc; 812 struct bridge_iflist *bif; 813 struct epoch_tracker et; 814 815 BRIDGE_LOCK(sc); 816 817 bridge_stop(ifp, 1); 818 ifp->if_flags &= ~IFF_UP; 819 820 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL) 821 bridge_delete_member(sc, bif, 0); 822 823 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) { 824 bridge_delete_span(sc, bif); 825 } 826 827 /* Tear down the routing table. */ 828 bridge_rtable_fini(sc); 829 830 BRIDGE_UNLOCK(sc); 831 832 NET_EPOCH_ENTER(et); 833 834 callout_drain(&sc->sc_brcallout); 835 836 BRIDGE_LIST_LOCK(); 837 LIST_REMOVE(sc, sc_list); 838 BRIDGE_LIST_UNLOCK(); 839 840 bstp_detach(&sc->sc_stp); 841 #ifdef ALTQ 842 IFQ_PURGE(&ifp->if_snd); 843 #endif 844 NET_EPOCH_EXIT(et); 845 846 ether_ifdetach(ifp); 847 if_free(ifp); 848 849 NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx); 850 851 return (0); 852 } 853 854 /* 855 * bridge_ioctl: 856 * 857 * Handle a control request from the operator. 858 */ 859 static int 860 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data) 861 { 862 struct bridge_softc *sc = ifp->if_softc; 863 struct ifreq *ifr = (struct ifreq *)data; 864 struct bridge_iflist *bif; 865 struct thread *td = curthread; 866 union { 867 struct ifbreq ifbreq; 868 struct ifbifconf ifbifconf; 869 struct ifbareq ifbareq; 870 struct ifbaconf ifbaconf; 871 struct ifbrparam ifbrparam; 872 struct ifbropreq ifbropreq; 873 } args; 874 struct ifdrv *ifd = (struct ifdrv *) data; 875 const struct bridge_control *bc; 876 int error = 0, oldmtu; 877 878 BRIDGE_LOCK(sc); 879 880 switch (cmd) { 881 case SIOCADDMULTI: 882 case SIOCDELMULTI: 883 break; 884 885 case SIOCGDRVSPEC: 886 case SIOCSDRVSPEC: 887 if (ifd->ifd_cmd >= bridge_control_table_size) { 888 error = EINVAL; 889 break; 890 } 891 bc = &bridge_control_table[ifd->ifd_cmd]; 892 893 if (cmd == SIOCGDRVSPEC && 894 (bc->bc_flags & BC_F_COPYOUT) == 0) { 895 error = EINVAL; 896 break; 897 } 898 else if (cmd == SIOCSDRVSPEC && 899 (bc->bc_flags & BC_F_COPYOUT) != 0) { 900 error = EINVAL; 901 break; 902 } 903 904 if (bc->bc_flags & BC_F_SUSER) { 905 error = priv_check(td, PRIV_NET_BRIDGE); 906 if (error) 907 break; 908 } 909 910 if (ifd->ifd_len != bc->bc_argsize || 911 ifd->ifd_len > sizeof(args)) { 912 error = EINVAL; 913 break; 914 } 915 916 bzero(&args, sizeof(args)); 917 if (bc->bc_flags & BC_F_COPYIN) { 918 error = copyin(ifd->ifd_data, &args, ifd->ifd_len); 919 if (error) 920 break; 921 } 922 923 oldmtu = ifp->if_mtu; 924 error = (*bc->bc_func)(sc, &args); 925 if (error) 926 break; 927 928 /* 929 * Bridge MTU may change during addition of the first port. 930 * If it did, do network layer specific procedure. 931 */ 932 if (ifp->if_mtu != oldmtu) 933 if_notifymtu(ifp); 934 935 if (bc->bc_flags & BC_F_COPYOUT) 936 error = copyout(&args, ifd->ifd_data, ifd->ifd_len); 937 938 break; 939 940 case SIOCSIFFLAGS: 941 if (!(ifp->if_flags & IFF_UP) && 942 (ifp->if_drv_flags & IFF_DRV_RUNNING)) { 943 /* 944 * If interface is marked down and it is running, 945 * then stop and disable it. 946 */ 947 bridge_stop(ifp, 1); 948 } else if ((ifp->if_flags & IFF_UP) && 949 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) { 950 /* 951 * If interface is marked up and it is stopped, then 952 * start it. 953 */ 954 BRIDGE_UNLOCK(sc); 955 (*ifp->if_init)(sc); 956 BRIDGE_LOCK(sc); 957 } 958 break; 959 960 case SIOCSIFMTU: 961 oldmtu = sc->sc_ifp->if_mtu; 962 963 if (ifr->ifr_mtu < 576) { 964 error = EINVAL; 965 break; 966 } 967 if (CK_LIST_EMPTY(&sc->sc_iflist)) { 968 sc->sc_ifp->if_mtu = ifr->ifr_mtu; 969 break; 970 } 971 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 972 error = (*bif->bif_ifp->if_ioctl)(bif->bif_ifp, 973 SIOCSIFMTU, (caddr_t)ifr); 974 if (error != 0) { 975 log(LOG_NOTICE, "%s: invalid MTU: %u for" 976 " member %s\n", sc->sc_ifp->if_xname, 977 ifr->ifr_mtu, 978 bif->bif_ifp->if_xname); 979 error = EINVAL; 980 break; 981 } 982 } 983 if (error) { 984 /* Restore the previous MTU on all member interfaces. */ 985 ifr->ifr_mtu = oldmtu; 986 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 987 (*bif->bif_ifp->if_ioctl)(bif->bif_ifp, 988 SIOCSIFMTU, (caddr_t)ifr); 989 } 990 } else { 991 sc->sc_ifp->if_mtu = ifr->ifr_mtu; 992 } 993 break; 994 default: 995 /* 996 * drop the lock as ether_ioctl() will call bridge_start() and 997 * cause the lock to be recursed. 998 */ 999 BRIDGE_UNLOCK(sc); 1000 error = ether_ioctl(ifp, cmd, data); 1001 BRIDGE_LOCK(sc); 1002 break; 1003 } 1004 1005 BRIDGE_UNLOCK(sc); 1006 1007 return (error); 1008 } 1009 1010 /* 1011 * bridge_mutecaps: 1012 * 1013 * Clear or restore unwanted capabilities on the member interface 1014 */ 1015 static void 1016 bridge_mutecaps(struct bridge_softc *sc) 1017 { 1018 struct bridge_iflist *bif; 1019 int enabled, mask; 1020 1021 BRIDGE_LOCK_ASSERT(sc); 1022 1023 /* Initial bitmask of capabilities to test */ 1024 mask = BRIDGE_IFCAPS_MASK; 1025 1026 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1027 /* Every member must support it or its disabled */ 1028 mask &= bif->bif_savedcaps; 1029 } 1030 1031 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1032 enabled = bif->bif_ifp->if_capenable; 1033 enabled &= ~BRIDGE_IFCAPS_STRIP; 1034 /* strip off mask bits and enable them again if allowed */ 1035 enabled &= ~BRIDGE_IFCAPS_MASK; 1036 enabled |= mask; 1037 bridge_set_ifcap(sc, bif, enabled); 1038 } 1039 } 1040 1041 static void 1042 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set) 1043 { 1044 struct ifnet *ifp = bif->bif_ifp; 1045 struct ifreq ifr; 1046 int error, mask, stuck; 1047 1048 bzero(&ifr, sizeof(ifr)); 1049 ifr.ifr_reqcap = set; 1050 1051 if (ifp->if_capenable != set) { 1052 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr); 1053 if (error) 1054 if_printf(sc->sc_ifp, 1055 "error setting capabilities on %s: %d\n", 1056 ifp->if_xname, error); 1057 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP; 1058 stuck = ifp->if_capenable & mask & ~set; 1059 if (stuck != 0) 1060 if_printf(sc->sc_ifp, 1061 "can't disable some capabilities on %s: 0x%x\n", 1062 ifp->if_xname, stuck); 1063 } 1064 } 1065 1066 /* 1067 * bridge_lookup_member: 1068 * 1069 * Lookup a bridge member interface. 1070 */ 1071 static struct bridge_iflist * 1072 bridge_lookup_member(struct bridge_softc *sc, const char *name) 1073 { 1074 struct bridge_iflist *bif; 1075 struct ifnet *ifp; 1076 1077 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc); 1078 1079 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1080 ifp = bif->bif_ifp; 1081 if (strcmp(ifp->if_xname, name) == 0) 1082 return (bif); 1083 } 1084 1085 return (NULL); 1086 } 1087 1088 /* 1089 * bridge_lookup_member_if: 1090 * 1091 * Lookup a bridge member interface by ifnet*. 1092 */ 1093 static struct bridge_iflist * 1094 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp) 1095 { 1096 struct bridge_iflist *bif; 1097 1098 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc); 1099 1100 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1101 if (bif->bif_ifp == member_ifp) 1102 return (bif); 1103 } 1104 1105 return (NULL); 1106 } 1107 1108 static void 1109 bridge_delete_member_cb(struct epoch_context *ctx) 1110 { 1111 struct bridge_iflist *bif; 1112 1113 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx); 1114 1115 free(bif, M_DEVBUF); 1116 } 1117 1118 /* 1119 * bridge_delete_member: 1120 * 1121 * Delete the specified member interface. 1122 */ 1123 static void 1124 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif, 1125 int gone) 1126 { 1127 struct ifnet *ifs = bif->bif_ifp; 1128 struct ifnet *fif = NULL; 1129 struct bridge_iflist *bifl; 1130 1131 BRIDGE_LOCK_ASSERT(sc); 1132 1133 if (bif->bif_flags & IFBIF_STP) 1134 bstp_disable(&bif->bif_stp); 1135 1136 ifs->if_bridge = NULL; 1137 CK_LIST_REMOVE(bif, bif_next); 1138 1139 /* 1140 * If removing the interface that gave the bridge its mac address, set 1141 * the mac address of the bridge to the address of the next member, or 1142 * to its default address if no members are left. 1143 */ 1144 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) { 1145 if (CK_LIST_EMPTY(&sc->sc_iflist)) { 1146 bcopy(&sc->sc_defaddr, 1147 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1148 sc->sc_ifaddr = NULL; 1149 } else { 1150 bifl = CK_LIST_FIRST(&sc->sc_iflist); 1151 fif = bifl->bif_ifp; 1152 bcopy(IF_LLADDR(fif), 1153 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1154 sc->sc_ifaddr = fif; 1155 } 1156 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp); 1157 } 1158 1159 bridge_linkcheck(sc); 1160 bridge_mutecaps(sc); /* recalcuate now this interface is removed */ 1161 BRIDGE_RT_LOCK(sc); 1162 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL); 1163 BRIDGE_RT_UNLOCK(sc); 1164 KASSERT(bif->bif_addrcnt == 0, 1165 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt)); 1166 1167 ifs->if_bridge_output = NULL; 1168 ifs->if_bridge_input = NULL; 1169 ifs->if_bridge_linkstate = NULL; 1170 if (!gone) { 1171 switch (ifs->if_type) { 1172 case IFT_ETHER: 1173 case IFT_L2VLAN: 1174 /* 1175 * Take the interface out of promiscuous mode, but only 1176 * if it was promiscuous in the first place. It might 1177 * not be if we're in the bridge_ioctl_add() error path. 1178 */ 1179 if (ifs->if_flags & IFF_PROMISC) 1180 (void) ifpromisc(ifs, 0); 1181 break; 1182 1183 case IFT_GIF: 1184 break; 1185 1186 default: 1187 #ifdef DIAGNOSTIC 1188 panic("bridge_delete_member: impossible"); 1189 #endif 1190 break; 1191 } 1192 /* reneable any interface capabilities */ 1193 bridge_set_ifcap(sc, bif, bif->bif_savedcaps); 1194 } 1195 bstp_destroy(&bif->bif_stp); /* prepare to free */ 1196 1197 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx); 1198 } 1199 1200 /* 1201 * bridge_delete_span: 1202 * 1203 * Delete the specified span interface. 1204 */ 1205 static void 1206 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif) 1207 { 1208 BRIDGE_LOCK_ASSERT(sc); 1209 1210 KASSERT(bif->bif_ifp->if_bridge == NULL, 1211 ("%s: not a span interface", __func__)); 1212 1213 CK_LIST_REMOVE(bif, bif_next); 1214 1215 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx); 1216 } 1217 1218 static int 1219 bridge_ioctl_add(struct bridge_softc *sc, void *arg) 1220 { 1221 struct ifbreq *req = arg; 1222 struct bridge_iflist *bif = NULL; 1223 struct ifnet *ifs; 1224 int error = 0; 1225 1226 ifs = ifunit(req->ifbr_ifsname); 1227 if (ifs == NULL) 1228 return (ENOENT); 1229 if (ifs->if_ioctl == NULL) /* must be supported */ 1230 return (EINVAL); 1231 1232 /* If it's in the span list, it can't be a member. */ 1233 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1234 if (ifs == bif->bif_ifp) 1235 return (EBUSY); 1236 1237 if (ifs->if_bridge == sc) 1238 return (EEXIST); 1239 1240 if (ifs->if_bridge != NULL) 1241 return (EBUSY); 1242 1243 switch (ifs->if_type) { 1244 case IFT_ETHER: 1245 case IFT_L2VLAN: 1246 case IFT_GIF: 1247 /* permitted interface types */ 1248 break; 1249 default: 1250 return (EINVAL); 1251 } 1252 1253 #ifdef INET6 1254 /* 1255 * Two valid inet6 addresses with link-local scope must not be 1256 * on the parent interface and the member interfaces at the 1257 * same time. This restriction is needed to prevent violation 1258 * of link-local scope zone. Attempts to add a member 1259 * interface which has inet6 addresses when the parent has 1260 * inet6 triggers removal of all inet6 addresses on the member 1261 * interface. 1262 */ 1263 1264 /* Check if the parent interface has a link-local scope addr. */ 1265 if (V_allow_llz_overlap == 0 && 1266 in6ifa_llaonifp(sc->sc_ifp) != NULL) { 1267 /* 1268 * If any, remove all inet6 addresses from the member 1269 * interfaces. 1270 */ 1271 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1272 if (in6ifa_llaonifp(bif->bif_ifp)) { 1273 in6_ifdetach(bif->bif_ifp); 1274 if_printf(sc->sc_ifp, 1275 "IPv6 addresses on %s have been removed " 1276 "before adding it as a member to prevent " 1277 "IPv6 address scope violation.\n", 1278 bif->bif_ifp->if_xname); 1279 } 1280 } 1281 if (in6ifa_llaonifp(ifs)) { 1282 in6_ifdetach(ifs); 1283 if_printf(sc->sc_ifp, 1284 "IPv6 addresses on %s have been removed " 1285 "before adding it as a member to prevent " 1286 "IPv6 address scope violation.\n", 1287 ifs->if_xname); 1288 } 1289 } 1290 #endif 1291 /* Allow the first Ethernet member to define the MTU */ 1292 if (CK_LIST_EMPTY(&sc->sc_iflist)) 1293 sc->sc_ifp->if_mtu = ifs->if_mtu; 1294 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) { 1295 struct ifreq ifr; 1296 1297 snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s", 1298 ifs->if_xname); 1299 ifr.ifr_mtu = sc->sc_ifp->if_mtu; 1300 1301 error = (*ifs->if_ioctl)(ifs, 1302 SIOCSIFMTU, (caddr_t)&ifr); 1303 if (error != 0) { 1304 log(LOG_NOTICE, "%s: invalid MTU: %u for" 1305 " new member %s\n", sc->sc_ifp->if_xname, 1306 ifr.ifr_mtu, 1307 ifs->if_xname); 1308 return (EINVAL); 1309 } 1310 } 1311 1312 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO); 1313 if (bif == NULL) 1314 return (ENOMEM); 1315 1316 bif->bif_ifp = ifs; 1317 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER; 1318 bif->bif_savedcaps = ifs->if_capenable; 1319 1320 /* 1321 * Assign the interface's MAC address to the bridge if it's the first 1322 * member and the MAC address of the bridge has not been changed from 1323 * the default randomly generated one. 1324 */ 1325 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) && 1326 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) { 1327 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1328 sc->sc_ifaddr = ifs; 1329 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp); 1330 } 1331 1332 ifs->if_bridge = sc; 1333 ifs->if_bridge_output = bridge_output; 1334 ifs->if_bridge_input = bridge_input; 1335 ifs->if_bridge_linkstate = bridge_linkstate; 1336 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp); 1337 /* 1338 * XXX: XLOCK HERE!?! 1339 * 1340 * NOTE: insert_***HEAD*** should be safe for the traversals. 1341 */ 1342 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next); 1343 1344 /* Set interface capabilities to the intersection set of all members */ 1345 bridge_mutecaps(sc); 1346 bridge_linkcheck(sc); 1347 1348 /* Place the interface into promiscuous mode */ 1349 switch (ifs->if_type) { 1350 case IFT_ETHER: 1351 case IFT_L2VLAN: 1352 error = ifpromisc(ifs, 1); 1353 break; 1354 } 1355 1356 if (error) 1357 bridge_delete_member(sc, bif, 0); 1358 return (error); 1359 } 1360 1361 static int 1362 bridge_ioctl_del(struct bridge_softc *sc, void *arg) 1363 { 1364 struct ifbreq *req = arg; 1365 struct bridge_iflist *bif; 1366 1367 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1368 if (bif == NULL) 1369 return (ENOENT); 1370 1371 bridge_delete_member(sc, bif, 0); 1372 1373 return (0); 1374 } 1375 1376 static int 1377 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg) 1378 { 1379 struct ifbreq *req = arg; 1380 struct bridge_iflist *bif; 1381 struct bstp_port *bp; 1382 1383 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1384 if (bif == NULL) 1385 return (ENOENT); 1386 1387 bp = &bif->bif_stp; 1388 req->ifbr_ifsflags = bif->bif_flags; 1389 req->ifbr_state = bp->bp_state; 1390 req->ifbr_priority = bp->bp_priority; 1391 req->ifbr_path_cost = bp->bp_path_cost; 1392 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff; 1393 req->ifbr_proto = bp->bp_protover; 1394 req->ifbr_role = bp->bp_role; 1395 req->ifbr_stpflags = bp->bp_flags; 1396 req->ifbr_addrcnt = bif->bif_addrcnt; 1397 req->ifbr_addrmax = bif->bif_addrmax; 1398 req->ifbr_addrexceeded = bif->bif_addrexceeded; 1399 1400 /* Copy STP state options as flags */ 1401 if (bp->bp_operedge) 1402 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE; 1403 if (bp->bp_flags & BSTP_PORT_AUTOEDGE) 1404 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE; 1405 if (bp->bp_ptp_link) 1406 req->ifbr_ifsflags |= IFBIF_BSTP_PTP; 1407 if (bp->bp_flags & BSTP_PORT_AUTOPTP) 1408 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP; 1409 if (bp->bp_flags & BSTP_PORT_ADMEDGE) 1410 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE; 1411 if (bp->bp_flags & BSTP_PORT_ADMCOST) 1412 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST; 1413 return (0); 1414 } 1415 1416 static int 1417 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg) 1418 { 1419 struct epoch_tracker et; 1420 struct ifbreq *req = arg; 1421 struct bridge_iflist *bif; 1422 struct bstp_port *bp; 1423 int error; 1424 1425 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1426 if (bif == NULL) 1427 return (ENOENT); 1428 bp = &bif->bif_stp; 1429 1430 if (req->ifbr_ifsflags & IFBIF_SPAN) 1431 /* SPAN is readonly */ 1432 return (EINVAL); 1433 1434 NET_EPOCH_ENTER(et); 1435 1436 if (req->ifbr_ifsflags & IFBIF_STP) { 1437 if ((bif->bif_flags & IFBIF_STP) == 0) { 1438 error = bstp_enable(&bif->bif_stp); 1439 if (error) { 1440 NET_EPOCH_EXIT(et); 1441 return (error); 1442 } 1443 } 1444 } else { 1445 if ((bif->bif_flags & IFBIF_STP) != 0) 1446 bstp_disable(&bif->bif_stp); 1447 } 1448 1449 /* Pass on STP flags */ 1450 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0); 1451 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0); 1452 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0); 1453 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0); 1454 1455 /* Save the bits relating to the bridge */ 1456 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK; 1457 1458 NET_EPOCH_EXIT(et); 1459 1460 return (0); 1461 } 1462 1463 static int 1464 bridge_ioctl_scache(struct bridge_softc *sc, void *arg) 1465 { 1466 struct ifbrparam *param = arg; 1467 1468 sc->sc_brtmax = param->ifbrp_csize; 1469 bridge_rttrim(sc); 1470 1471 return (0); 1472 } 1473 1474 static int 1475 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg) 1476 { 1477 struct ifbrparam *param = arg; 1478 1479 param->ifbrp_csize = sc->sc_brtmax; 1480 1481 return (0); 1482 } 1483 1484 static int 1485 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg) 1486 { 1487 struct ifbifconf *bifc = arg; 1488 struct bridge_iflist *bif; 1489 struct ifbreq breq; 1490 char *buf, *outbuf; 1491 int count, buflen, len, error = 0; 1492 1493 count = 0; 1494 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) 1495 count++; 1496 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1497 count++; 1498 1499 buflen = sizeof(breq) * count; 1500 if (bifc->ifbic_len == 0) { 1501 bifc->ifbic_len = buflen; 1502 return (0); 1503 } 1504 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO); 1505 if (outbuf == NULL) 1506 return (ENOMEM); 1507 1508 count = 0; 1509 buf = outbuf; 1510 len = min(bifc->ifbic_len, buflen); 1511 bzero(&breq, sizeof(breq)); 1512 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1513 if (len < sizeof(breq)) 1514 break; 1515 1516 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname, 1517 sizeof(breq.ifbr_ifsname)); 1518 /* Fill in the ifbreq structure */ 1519 error = bridge_ioctl_gifflags(sc, &breq); 1520 if (error) 1521 break; 1522 memcpy(buf, &breq, sizeof(breq)); 1523 count++; 1524 buf += sizeof(breq); 1525 len -= sizeof(breq); 1526 } 1527 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) { 1528 if (len < sizeof(breq)) 1529 break; 1530 1531 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname, 1532 sizeof(breq.ifbr_ifsname)); 1533 breq.ifbr_ifsflags = bif->bif_flags; 1534 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff; 1535 memcpy(buf, &breq, sizeof(breq)); 1536 count++; 1537 buf += sizeof(breq); 1538 len -= sizeof(breq); 1539 } 1540 1541 bifc->ifbic_len = sizeof(breq) * count; 1542 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len); 1543 free(outbuf, M_TEMP); 1544 return (error); 1545 } 1546 1547 static int 1548 bridge_ioctl_rts(struct bridge_softc *sc, void *arg) 1549 { 1550 struct ifbaconf *bac = arg; 1551 struct bridge_rtnode *brt; 1552 struct ifbareq bareq; 1553 char *buf, *outbuf; 1554 int count, buflen, len, error = 0; 1555 1556 if (bac->ifbac_len == 0) 1557 return (0); 1558 1559 count = 0; 1560 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) 1561 count++; 1562 buflen = sizeof(bareq) * count; 1563 1564 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO); 1565 if (outbuf == NULL) 1566 return (ENOMEM); 1567 1568 count = 0; 1569 buf = outbuf; 1570 len = min(bac->ifbac_len, buflen); 1571 bzero(&bareq, sizeof(bareq)); 1572 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) { 1573 if (len < sizeof(bareq)) 1574 goto out; 1575 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname, 1576 sizeof(bareq.ifba_ifsname)); 1577 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr)); 1578 bareq.ifba_vlan = brt->brt_vlan; 1579 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 1580 time_uptime < brt->brt_expire) 1581 bareq.ifba_expire = brt->brt_expire - time_uptime; 1582 else 1583 bareq.ifba_expire = 0; 1584 bareq.ifba_flags = brt->brt_flags; 1585 1586 memcpy(buf, &bareq, sizeof(bareq)); 1587 count++; 1588 buf += sizeof(bareq); 1589 len -= sizeof(bareq); 1590 } 1591 out: 1592 bac->ifbac_len = sizeof(bareq) * count; 1593 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len); 1594 free(outbuf, M_TEMP); 1595 return (error); 1596 } 1597 1598 static int 1599 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg) 1600 { 1601 struct ifbareq *req = arg; 1602 struct bridge_iflist *bif; 1603 struct epoch_tracker et; 1604 int error; 1605 1606 NET_EPOCH_ENTER(et); 1607 bif = bridge_lookup_member(sc, req->ifba_ifsname); 1608 if (bif == NULL) { 1609 NET_EPOCH_EXIT(et); 1610 return (ENOENT); 1611 } 1612 1613 /* bridge_rtupdate() may acquire the lock. */ 1614 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1, 1615 req->ifba_flags); 1616 NET_EPOCH_EXIT(et); 1617 1618 return (error); 1619 } 1620 1621 static int 1622 bridge_ioctl_sto(struct bridge_softc *sc, void *arg) 1623 { 1624 struct ifbrparam *param = arg; 1625 1626 sc->sc_brttimeout = param->ifbrp_ctime; 1627 return (0); 1628 } 1629 1630 static int 1631 bridge_ioctl_gto(struct bridge_softc *sc, void *arg) 1632 { 1633 struct ifbrparam *param = arg; 1634 1635 param->ifbrp_ctime = sc->sc_brttimeout; 1636 return (0); 1637 } 1638 1639 static int 1640 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg) 1641 { 1642 struct ifbareq *req = arg; 1643 int vlan = req->ifba_vlan; 1644 1645 /* Userspace uses '0' to mean 'any vlan' */ 1646 if (vlan == 0) 1647 vlan = DOT1Q_VID_RSVD_IMPL; 1648 1649 return (bridge_rtdaddr(sc, req->ifba_dst, vlan)); 1650 } 1651 1652 static int 1653 bridge_ioctl_flush(struct bridge_softc *sc, void *arg) 1654 { 1655 struct ifbreq *req = arg; 1656 1657 BRIDGE_RT_LOCK(sc); 1658 bridge_rtflush(sc, req->ifbr_ifsflags); 1659 BRIDGE_RT_UNLOCK(sc); 1660 1661 return (0); 1662 } 1663 1664 static int 1665 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg) 1666 { 1667 struct ifbrparam *param = arg; 1668 struct bstp_state *bs = &sc->sc_stp; 1669 1670 param->ifbrp_prio = bs->bs_bridge_priority; 1671 return (0); 1672 } 1673 1674 static int 1675 bridge_ioctl_spri(struct bridge_softc *sc, void *arg) 1676 { 1677 struct ifbrparam *param = arg; 1678 1679 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio)); 1680 } 1681 1682 static int 1683 bridge_ioctl_ght(struct bridge_softc *sc, void *arg) 1684 { 1685 struct ifbrparam *param = arg; 1686 struct bstp_state *bs = &sc->sc_stp; 1687 1688 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8; 1689 return (0); 1690 } 1691 1692 static int 1693 bridge_ioctl_sht(struct bridge_softc *sc, void *arg) 1694 { 1695 struct ifbrparam *param = arg; 1696 1697 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime)); 1698 } 1699 1700 static int 1701 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg) 1702 { 1703 struct ifbrparam *param = arg; 1704 struct bstp_state *bs = &sc->sc_stp; 1705 1706 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8; 1707 return (0); 1708 } 1709 1710 static int 1711 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg) 1712 { 1713 struct ifbrparam *param = arg; 1714 1715 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay)); 1716 } 1717 1718 static int 1719 bridge_ioctl_gma(struct bridge_softc *sc, void *arg) 1720 { 1721 struct ifbrparam *param = arg; 1722 struct bstp_state *bs = &sc->sc_stp; 1723 1724 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8; 1725 return (0); 1726 } 1727 1728 static int 1729 bridge_ioctl_sma(struct bridge_softc *sc, void *arg) 1730 { 1731 struct ifbrparam *param = arg; 1732 1733 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage)); 1734 } 1735 1736 static int 1737 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg) 1738 { 1739 struct ifbreq *req = arg; 1740 struct bridge_iflist *bif; 1741 1742 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1743 if (bif == NULL) 1744 return (ENOENT); 1745 1746 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority)); 1747 } 1748 1749 static int 1750 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg) 1751 { 1752 struct ifbreq *req = arg; 1753 struct bridge_iflist *bif; 1754 1755 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1756 if (bif == NULL) 1757 return (ENOENT); 1758 1759 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost)); 1760 } 1761 1762 static int 1763 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg) 1764 { 1765 struct ifbreq *req = arg; 1766 struct bridge_iflist *bif; 1767 1768 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1769 if (bif == NULL) 1770 return (ENOENT); 1771 1772 bif->bif_addrmax = req->ifbr_addrmax; 1773 return (0); 1774 } 1775 1776 static int 1777 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg) 1778 { 1779 struct ifbreq *req = arg; 1780 struct bridge_iflist *bif = NULL; 1781 struct ifnet *ifs; 1782 1783 ifs = ifunit(req->ifbr_ifsname); 1784 if (ifs == NULL) 1785 return (ENOENT); 1786 1787 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1788 if (ifs == bif->bif_ifp) 1789 return (EBUSY); 1790 1791 if (ifs->if_bridge != NULL) 1792 return (EBUSY); 1793 1794 switch (ifs->if_type) { 1795 case IFT_ETHER: 1796 case IFT_GIF: 1797 case IFT_L2VLAN: 1798 break; 1799 default: 1800 return (EINVAL); 1801 } 1802 1803 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO); 1804 if (bif == NULL) 1805 return (ENOMEM); 1806 1807 bif->bif_ifp = ifs; 1808 bif->bif_flags = IFBIF_SPAN; 1809 1810 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next); 1811 1812 return (0); 1813 } 1814 1815 static int 1816 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg) 1817 { 1818 struct ifbreq *req = arg; 1819 struct bridge_iflist *bif; 1820 struct ifnet *ifs; 1821 1822 ifs = ifunit(req->ifbr_ifsname); 1823 if (ifs == NULL) 1824 return (ENOENT); 1825 1826 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1827 if (ifs == bif->bif_ifp) 1828 break; 1829 1830 if (bif == NULL) 1831 return (ENOENT); 1832 1833 bridge_delete_span(sc, bif); 1834 1835 return (0); 1836 } 1837 1838 static int 1839 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg) 1840 { 1841 struct ifbropreq *req = arg; 1842 struct bstp_state *bs = &sc->sc_stp; 1843 struct bstp_port *root_port; 1844 1845 req->ifbop_maxage = bs->bs_bridge_max_age >> 8; 1846 req->ifbop_hellotime = bs->bs_bridge_htime >> 8; 1847 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8; 1848 1849 root_port = bs->bs_root_port; 1850 if (root_port == NULL) 1851 req->ifbop_root_port = 0; 1852 else 1853 req->ifbop_root_port = root_port->bp_ifp->if_index; 1854 1855 req->ifbop_holdcount = bs->bs_txholdcount; 1856 req->ifbop_priority = bs->bs_bridge_priority; 1857 req->ifbop_protocol = bs->bs_protover; 1858 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost; 1859 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id; 1860 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id; 1861 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id; 1862 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec; 1863 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec; 1864 1865 return (0); 1866 } 1867 1868 static int 1869 bridge_ioctl_grte(struct bridge_softc *sc, void *arg) 1870 { 1871 struct ifbrparam *param = arg; 1872 1873 param->ifbrp_cexceeded = sc->sc_brtexceeded; 1874 return (0); 1875 } 1876 1877 static int 1878 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg) 1879 { 1880 struct ifbpstpconf *bifstp = arg; 1881 struct bridge_iflist *bif; 1882 struct bstp_port *bp; 1883 struct ifbpstpreq bpreq; 1884 char *buf, *outbuf; 1885 int count, buflen, len, error = 0; 1886 1887 count = 0; 1888 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1889 if ((bif->bif_flags & IFBIF_STP) != 0) 1890 count++; 1891 } 1892 1893 buflen = sizeof(bpreq) * count; 1894 if (bifstp->ifbpstp_len == 0) { 1895 bifstp->ifbpstp_len = buflen; 1896 return (0); 1897 } 1898 1899 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO); 1900 if (outbuf == NULL) 1901 return (ENOMEM); 1902 1903 count = 0; 1904 buf = outbuf; 1905 len = min(bifstp->ifbpstp_len, buflen); 1906 bzero(&bpreq, sizeof(bpreq)); 1907 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1908 if (len < sizeof(bpreq)) 1909 break; 1910 1911 if ((bif->bif_flags & IFBIF_STP) == 0) 1912 continue; 1913 1914 bp = &bif->bif_stp; 1915 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff; 1916 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions; 1917 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost; 1918 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id; 1919 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id; 1920 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id; 1921 1922 memcpy(buf, &bpreq, sizeof(bpreq)); 1923 count++; 1924 buf += sizeof(bpreq); 1925 len -= sizeof(bpreq); 1926 } 1927 1928 bifstp->ifbpstp_len = sizeof(bpreq) * count; 1929 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len); 1930 free(outbuf, M_TEMP); 1931 return (error); 1932 } 1933 1934 static int 1935 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg) 1936 { 1937 struct ifbrparam *param = arg; 1938 1939 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto)); 1940 } 1941 1942 static int 1943 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg) 1944 { 1945 struct ifbrparam *param = arg; 1946 1947 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc)); 1948 } 1949 1950 /* 1951 * bridge_ifdetach: 1952 * 1953 * Detach an interface from a bridge. Called when a member 1954 * interface is detaching. 1955 */ 1956 static void 1957 bridge_ifdetach(void *arg __unused, struct ifnet *ifp) 1958 { 1959 struct bridge_softc *sc = ifp->if_bridge; 1960 struct bridge_iflist *bif; 1961 1962 if (ifp->if_flags & IFF_RENAMING) 1963 return; 1964 if (V_bridge_cloner == NULL) { 1965 /* 1966 * This detach handler can be called after 1967 * vnet_bridge_uninit(). Just return in that case. 1968 */ 1969 return; 1970 } 1971 /* Check if the interface is a bridge member */ 1972 if (sc != NULL) { 1973 BRIDGE_LOCK(sc); 1974 1975 bif = bridge_lookup_member_if(sc, ifp); 1976 if (bif != NULL) 1977 bridge_delete_member(sc, bif, 1); 1978 1979 BRIDGE_UNLOCK(sc); 1980 return; 1981 } 1982 1983 /* Check if the interface is a span port */ 1984 BRIDGE_LIST_LOCK(); 1985 LIST_FOREACH(sc, &V_bridge_list, sc_list) { 1986 BRIDGE_LOCK(sc); 1987 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1988 if (ifp == bif->bif_ifp) { 1989 bridge_delete_span(sc, bif); 1990 break; 1991 } 1992 1993 BRIDGE_UNLOCK(sc); 1994 } 1995 BRIDGE_LIST_UNLOCK(); 1996 } 1997 1998 /* 1999 * bridge_init: 2000 * 2001 * Initialize a bridge interface. 2002 */ 2003 static void 2004 bridge_init(void *xsc) 2005 { 2006 struct bridge_softc *sc = (struct bridge_softc *)xsc; 2007 struct ifnet *ifp = sc->sc_ifp; 2008 2009 if (ifp->if_drv_flags & IFF_DRV_RUNNING) 2010 return; 2011 2012 BRIDGE_LOCK(sc); 2013 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz, 2014 bridge_timer, sc); 2015 2016 ifp->if_drv_flags |= IFF_DRV_RUNNING; 2017 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */ 2018 2019 BRIDGE_UNLOCK(sc); 2020 } 2021 2022 /* 2023 * bridge_stop: 2024 * 2025 * Stop the bridge interface. 2026 */ 2027 static void 2028 bridge_stop(struct ifnet *ifp, int disable) 2029 { 2030 struct bridge_softc *sc = ifp->if_softc; 2031 2032 BRIDGE_LOCK_ASSERT(sc); 2033 2034 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) 2035 return; 2036 2037 BRIDGE_RT_LOCK(sc); 2038 callout_stop(&sc->sc_brcallout); 2039 2040 bstp_stop(&sc->sc_stp); 2041 2042 bridge_rtflush(sc, IFBF_FLUSHDYN); 2043 BRIDGE_RT_UNLOCK(sc); 2044 2045 ifp->if_drv_flags &= ~IFF_DRV_RUNNING; 2046 } 2047 2048 /* 2049 * bridge_enqueue: 2050 * 2051 * Enqueue a packet on a bridge member interface. 2052 * 2053 */ 2054 static int 2055 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m) 2056 { 2057 int len, err = 0; 2058 short mflags; 2059 struct mbuf *m0; 2060 2061 /* We may be sending a fragment so traverse the mbuf */ 2062 for (; m; m = m0) { 2063 m0 = m->m_nextpkt; 2064 m->m_nextpkt = NULL; 2065 len = m->m_pkthdr.len; 2066 mflags = m->m_flags; 2067 2068 /* 2069 * If underlying interface can not do VLAN tag insertion itself 2070 * then attach a packet tag that holds it. 2071 */ 2072 if ((m->m_flags & M_VLANTAG) && 2073 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) { 2074 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag); 2075 if (m == NULL) { 2076 if_printf(dst_ifp, 2077 "unable to prepend VLAN header\n"); 2078 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1); 2079 continue; 2080 } 2081 m->m_flags &= ~M_VLANTAG; 2082 } 2083 2084 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */ 2085 if ((err = dst_ifp->if_transmit(dst_ifp, m))) { 2086 int n; 2087 2088 for (m = m0, n = 1; m != NULL; m = m0, n++) { 2089 m0 = m->m_nextpkt; 2090 m_freem(m); 2091 } 2092 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, n); 2093 break; 2094 } 2095 2096 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1); 2097 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len); 2098 if (mflags & M_MCAST) 2099 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1); 2100 } 2101 2102 return (err); 2103 } 2104 2105 /* 2106 * bridge_dummynet: 2107 * 2108 * Receive a queued packet from dummynet and pass it on to the output 2109 * interface. 2110 * 2111 * The mbuf has the Ethernet header already attached. 2112 */ 2113 static void 2114 bridge_dummynet(struct mbuf *m, struct ifnet *ifp) 2115 { 2116 struct bridge_softc *sc; 2117 2118 sc = ifp->if_bridge; 2119 2120 /* 2121 * The packet didnt originate from a member interface. This should only 2122 * ever happen if a member interface is removed while packets are 2123 * queued for it. 2124 */ 2125 if (sc == NULL) { 2126 m_freem(m); 2127 return; 2128 } 2129 2130 if (PFIL_HOOKED_OUT(V_inet_pfil_head) 2131 #ifdef INET6 2132 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2133 #endif 2134 ) { 2135 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0) 2136 return; 2137 if (m == NULL) 2138 return; 2139 } 2140 2141 bridge_enqueue(sc, ifp, m); 2142 } 2143 2144 /* 2145 * bridge_output: 2146 * 2147 * Send output from a bridge member interface. This 2148 * performs the bridging function for locally originated 2149 * packets. 2150 * 2151 * The mbuf has the Ethernet header already attached. We must 2152 * enqueue or free the mbuf before returning. 2153 */ 2154 static int 2155 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa, 2156 struct rtentry *rt) 2157 { 2158 struct ether_header *eh; 2159 struct ifnet *bifp, *dst_if; 2160 struct bridge_softc *sc; 2161 uint16_t vlan; 2162 2163 NET_EPOCH_ASSERT(); 2164 2165 if (m->m_len < ETHER_HDR_LEN) { 2166 m = m_pullup(m, ETHER_HDR_LEN); 2167 if (m == NULL) 2168 return (0); 2169 } 2170 2171 eh = mtod(m, struct ether_header *); 2172 sc = ifp->if_bridge; 2173 vlan = VLANTAGOF(m); 2174 2175 bifp = sc->sc_ifp; 2176 2177 /* 2178 * If bridge is down, but the original output interface is up, 2179 * go ahead and send out that interface. Otherwise, the packet 2180 * is dropped below. 2181 */ 2182 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { 2183 dst_if = ifp; 2184 goto sendunicast; 2185 } 2186 2187 /* 2188 * If the packet is a multicast, or we don't know a better way to 2189 * get there, send to all interfaces. 2190 */ 2191 if (ETHER_IS_MULTICAST(eh->ether_dhost)) 2192 dst_if = NULL; 2193 else 2194 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan); 2195 /* Tap any traffic not passing back out the originating interface */ 2196 if (dst_if != ifp) 2197 ETHER_BPF_MTAP(bifp, m); 2198 if (dst_if == NULL) { 2199 struct bridge_iflist *bif; 2200 struct mbuf *mc; 2201 int used = 0; 2202 2203 bridge_span(sc, m); 2204 2205 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 2206 dst_if = bif->bif_ifp; 2207 2208 if (dst_if->if_type == IFT_GIF) 2209 continue; 2210 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2211 continue; 2212 2213 /* 2214 * If this is not the original output interface, 2215 * and the interface is participating in spanning 2216 * tree, make sure the port is in a state that 2217 * allows forwarding. 2218 */ 2219 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) && 2220 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2221 continue; 2222 2223 if (CK_LIST_NEXT(bif, bif_next) == NULL) { 2224 used = 1; 2225 mc = m; 2226 } else { 2227 mc = m_dup(m, M_NOWAIT); 2228 if (mc == NULL) { 2229 if_inc_counter(bifp, IFCOUNTER_OERRORS, 1); 2230 continue; 2231 } 2232 } 2233 2234 bridge_enqueue(sc, dst_if, mc); 2235 } 2236 if (used == 0) 2237 m_freem(m); 2238 return (0); 2239 } 2240 2241 sendunicast: 2242 /* 2243 * XXX Spanning tree consideration here? 2244 */ 2245 2246 bridge_span(sc, m); 2247 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) { 2248 m_freem(m); 2249 return (0); 2250 } 2251 2252 bridge_enqueue(sc, dst_if, m); 2253 return (0); 2254 } 2255 2256 /* 2257 * bridge_transmit: 2258 * 2259 * Do output on a bridge. 2260 * 2261 */ 2262 static int 2263 bridge_transmit(struct ifnet *ifp, struct mbuf *m) 2264 { 2265 struct bridge_softc *sc; 2266 struct ether_header *eh; 2267 struct ifnet *dst_if; 2268 int error = 0; 2269 2270 sc = ifp->if_softc; 2271 2272 ETHER_BPF_MTAP(ifp, m); 2273 2274 eh = mtod(m, struct ether_header *); 2275 2276 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) && 2277 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, DOT1Q_VID_NULL)) != 2278 NULL) { 2279 error = bridge_enqueue(sc, dst_if, m); 2280 } else 2281 bridge_broadcast(sc, ifp, m, 0); 2282 2283 return (error); 2284 } 2285 2286 #ifdef ALTQ 2287 static void 2288 bridge_altq_start(if_t ifp) 2289 { 2290 struct ifaltq *ifq = &ifp->if_snd; 2291 struct mbuf *m; 2292 2293 IFQ_LOCK(ifq); 2294 IFQ_DEQUEUE_NOLOCK(ifq, m); 2295 while (m != NULL) { 2296 bridge_transmit(ifp, m); 2297 IFQ_DEQUEUE_NOLOCK(ifq, m); 2298 } 2299 IFQ_UNLOCK(ifq); 2300 } 2301 2302 static int 2303 bridge_altq_transmit(if_t ifp, struct mbuf *m) 2304 { 2305 int err; 2306 2307 if (ALTQ_IS_ENABLED(&ifp->if_snd)) { 2308 IFQ_ENQUEUE(&ifp->if_snd, m, err); 2309 if (err == 0) 2310 bridge_altq_start(ifp); 2311 } else 2312 err = bridge_transmit(ifp, m); 2313 2314 return (err); 2315 } 2316 #endif /* ALTQ */ 2317 2318 /* 2319 * The ifp->if_qflush entry point for if_bridge(4) is no-op. 2320 */ 2321 static void 2322 bridge_qflush(struct ifnet *ifp __unused) 2323 { 2324 } 2325 2326 /* 2327 * bridge_forward: 2328 * 2329 * The forwarding function of the bridge. 2330 * 2331 * NOTE: Releases the lock on return. 2332 */ 2333 static void 2334 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif, 2335 struct mbuf *m) 2336 { 2337 struct bridge_iflist *dbif; 2338 struct ifnet *src_if, *dst_if, *ifp; 2339 struct ether_header *eh; 2340 uint16_t vlan; 2341 uint8_t *dst; 2342 int error; 2343 2344 NET_EPOCH_ASSERT(); 2345 2346 src_if = m->m_pkthdr.rcvif; 2347 ifp = sc->sc_ifp; 2348 2349 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1); 2350 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); 2351 vlan = VLANTAGOF(m); 2352 2353 if ((sbif->bif_flags & IFBIF_STP) && 2354 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2355 goto drop; 2356 2357 eh = mtod(m, struct ether_header *); 2358 dst = eh->ether_dhost; 2359 2360 /* If the interface is learning, record the address. */ 2361 if (sbif->bif_flags & IFBIF_LEARNING) { 2362 error = bridge_rtupdate(sc, eh->ether_shost, vlan, 2363 sbif, 0, IFBAF_DYNAMIC); 2364 /* 2365 * If the interface has addresses limits then deny any source 2366 * that is not in the cache. 2367 */ 2368 if (error && sbif->bif_addrmax) 2369 goto drop; 2370 } 2371 2372 if ((sbif->bif_flags & IFBIF_STP) != 0 && 2373 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING) 2374 goto drop; 2375 2376 #ifdef DEV_NETMAP 2377 /* 2378 * Hand the packet to netmap only if it wasn't injected by netmap 2379 * itself. 2380 */ 2381 if ((m->m_flags & M_BRIDGE_INJECT) == 0 && 2382 (if_getcapenable(ifp) & IFCAP_NETMAP) != 0) { 2383 ifp->if_input(ifp, m); 2384 return; 2385 } 2386 m->m_flags &= ~M_BRIDGE_INJECT; 2387 #endif 2388 2389 /* 2390 * At this point, the port either doesn't participate 2391 * in spanning tree or it is in the forwarding state. 2392 */ 2393 2394 /* 2395 * If the packet is unicast, destined for someone on 2396 * "this" side of the bridge, drop it. 2397 */ 2398 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) { 2399 dst_if = bridge_rtlookup(sc, dst, vlan); 2400 if (src_if == dst_if) 2401 goto drop; 2402 } else { 2403 /* 2404 * Check if its a reserved multicast address, any address 2405 * listed in 802.1D section 7.12.6 may not be forwarded by the 2406 * bridge. 2407 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F 2408 */ 2409 if (dst[0] == 0x01 && dst[1] == 0x80 && 2410 dst[2] == 0xc2 && dst[3] == 0x00 && 2411 dst[4] == 0x00 && dst[5] <= 0x0f) 2412 goto drop; 2413 2414 /* ...forward it to all interfaces. */ 2415 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1); 2416 dst_if = NULL; 2417 } 2418 2419 /* 2420 * If we have a destination interface which is a member of our bridge, 2421 * OR this is a unicast packet, push it through the bpf(4) machinery. 2422 * For broadcast or multicast packets, don't bother because it will 2423 * be reinjected into ether_input. We do this before we pass the packets 2424 * through the pfil(9) framework, as it is possible that pfil(9) will 2425 * drop the packet, or possibly modify it, making it difficult to debug 2426 * firewall issues on the bridge. 2427 */ 2428 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0) 2429 ETHER_BPF_MTAP(ifp, m); 2430 2431 /* run the packet filter */ 2432 if (PFIL_HOOKED_IN(V_inet_pfil_head) 2433 #ifdef INET6 2434 || PFIL_HOOKED_IN(V_inet6_pfil_head) 2435 #endif 2436 ) { 2437 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0) 2438 return; 2439 if (m == NULL) 2440 return; 2441 } 2442 2443 if (dst_if == NULL) { 2444 bridge_broadcast(sc, src_if, m, 1); 2445 return; 2446 } 2447 2448 /* 2449 * At this point, we're dealing with a unicast frame 2450 * going to a different interface. 2451 */ 2452 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2453 goto drop; 2454 2455 dbif = bridge_lookup_member_if(sc, dst_if); 2456 if (dbif == NULL) 2457 /* Not a member of the bridge (anymore?) */ 2458 goto drop; 2459 2460 /* Private segments can not talk to each other */ 2461 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE) 2462 goto drop; 2463 2464 if ((dbif->bif_flags & IFBIF_STP) && 2465 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2466 goto drop; 2467 2468 if (PFIL_HOOKED_OUT(V_inet_pfil_head) 2469 #ifdef INET6 2470 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2471 #endif 2472 ) { 2473 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0) 2474 return; 2475 if (m == NULL) 2476 return; 2477 } 2478 2479 bridge_enqueue(sc, dst_if, m); 2480 return; 2481 2482 drop: 2483 m_freem(m); 2484 } 2485 2486 /* 2487 * bridge_input: 2488 * 2489 * Receive input from a member interface. Queue the packet for 2490 * bridging if it is not for us. 2491 */ 2492 static struct mbuf * 2493 bridge_input(struct ifnet *ifp, struct mbuf *m) 2494 { 2495 struct bridge_softc *sc; 2496 struct bridge_iflist *bif, *bif2; 2497 struct ifnet *bifp; 2498 struct ether_header *eh; 2499 struct mbuf *mc, *mc2; 2500 uint16_t vlan; 2501 int error; 2502 2503 NET_EPOCH_ASSERT(); 2504 2505 eh = mtod(m, struct ether_header *); 2506 vlan = VLANTAGOF(m); 2507 2508 sc = ifp->if_bridge; 2509 if (sc == NULL) { 2510 /* 2511 * This packet originated from the bridge itself, so it must 2512 * have been transmitted by netmap. Derive the "source" 2513 * interface from the source address and drop the packet if the 2514 * source address isn't known. 2515 */ 2516 KASSERT((m->m_flags & M_BRIDGE_INJECT) != 0, 2517 ("%s: ifnet %p missing a bridge softc", __func__, ifp)); 2518 sc = if_getsoftc(ifp); 2519 ifp = bridge_rtlookup(sc, eh->ether_shost, vlan); 2520 if (ifp == NULL) { 2521 if_inc_counter(sc->sc_ifp, IFCOUNTER_IERRORS, 1); 2522 m_freem(m); 2523 return (NULL); 2524 } 2525 m->m_pkthdr.rcvif = ifp; 2526 } 2527 bifp = sc->sc_ifp; 2528 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) 2529 return (m); 2530 2531 /* 2532 * Implement support for bridge monitoring. If this flag has been 2533 * set on this interface, discard the packet once we push it through 2534 * the bpf(4) machinery, but before we do, increment the byte and 2535 * packet counters associated with this interface. 2536 */ 2537 if ((bifp->if_flags & IFF_MONITOR) != 0) { 2538 m->m_pkthdr.rcvif = bifp; 2539 ETHER_BPF_MTAP(bifp, m); 2540 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); 2541 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); 2542 m_freem(m); 2543 return (NULL); 2544 } 2545 bif = bridge_lookup_member_if(sc, ifp); 2546 if (bif == NULL) { 2547 return (m); 2548 } 2549 2550 bridge_span(sc, m); 2551 2552 if (m->m_flags & (M_BCAST|M_MCAST)) { 2553 /* Tap off 802.1D packets; they do not get forwarded. */ 2554 if (memcmp(eh->ether_dhost, bstp_etheraddr, 2555 ETHER_ADDR_LEN) == 0) { 2556 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */ 2557 return (NULL); 2558 } 2559 2560 if ((bif->bif_flags & IFBIF_STP) && 2561 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) { 2562 return (m); 2563 } 2564 2565 /* 2566 * Make a deep copy of the packet and enqueue the copy 2567 * for bridge processing; return the original packet for 2568 * local processing. 2569 */ 2570 mc = m_dup(m, M_NOWAIT); 2571 if (mc == NULL) { 2572 return (m); 2573 } 2574 2575 /* Perform the bridge forwarding function with the copy. */ 2576 bridge_forward(sc, bif, mc); 2577 2578 #ifdef DEV_NETMAP 2579 /* 2580 * If netmap is enabled and has not already seen this packet, 2581 * then it will be consumed by bridge_forward(). 2582 */ 2583 if ((if_getcapenable(bifp) & IFCAP_NETMAP) != 0 && 2584 (m->m_flags & M_BRIDGE_INJECT) == 0) { 2585 m_freem(m); 2586 return (NULL); 2587 } 2588 #endif 2589 2590 /* 2591 * Reinject the mbuf as arriving on the bridge so we have a 2592 * chance at claiming multicast packets. We can not loop back 2593 * here from ether_input as a bridge is never a member of a 2594 * bridge. 2595 */ 2596 KASSERT(bifp->if_bridge == NULL, 2597 ("loop created in bridge_input")); 2598 mc2 = m_dup(m, M_NOWAIT); 2599 if (mc2 != NULL) { 2600 /* Keep the layer3 header aligned */ 2601 int i = min(mc2->m_pkthdr.len, max_protohdr); 2602 mc2 = m_copyup(mc2, i, ETHER_ALIGN); 2603 } 2604 if (mc2 != NULL) { 2605 mc2->m_pkthdr.rcvif = bifp; 2606 mc2->m_flags &= ~M_BRIDGE_INJECT; 2607 sc->sc_if_input(bifp, mc2); 2608 } 2609 2610 /* Return the original packet for local processing. */ 2611 return (m); 2612 } 2613 2614 if ((bif->bif_flags & IFBIF_STP) && 2615 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) { 2616 return (m); 2617 } 2618 2619 #if defined(INET) || defined(INET6) 2620 #define CARP_CHECK_WE_ARE_DST(iface) \ 2621 ((iface)->if_carp && (*carp_forus_p)((iface), eh->ether_dhost)) 2622 #define CARP_CHECK_WE_ARE_SRC(iface) \ 2623 ((iface)->if_carp && (*carp_forus_p)((iface), eh->ether_shost)) 2624 #else 2625 #define CARP_CHECK_WE_ARE_DST(iface) false 2626 #define CARP_CHECK_WE_ARE_SRC(iface) false 2627 #endif 2628 2629 #ifdef INET6 2630 #define PFIL_HOOKED_INET6 PFIL_HOOKED_IN(V_inet6_pfil_head) 2631 #else 2632 #define PFIL_HOOKED_INET6 false 2633 #endif 2634 2635 #ifdef DEV_NETMAP 2636 #define GRAB_FOR_NETMAP(ifp, m) do { \ 2637 if ((if_getcapenable(ifp) & IFCAP_NETMAP) != 0 && \ 2638 ((m)->m_flags & M_BRIDGE_INJECT) == 0) { \ 2639 (ifp)->if_input(ifp, m); \ 2640 return (NULL); \ 2641 } \ 2642 } while (0) 2643 #else 2644 #define GRAB_FOR_NETMAP(ifp, m) 2645 #endif 2646 2647 #define GRAB_OUR_PACKETS(iface) \ 2648 if ((iface)->if_type == IFT_GIF) \ 2649 continue; \ 2650 /* It is destined for us. */ \ 2651 if (memcmp(IF_LLADDR(iface), eh->ether_dhost, ETHER_ADDR_LEN) == 0 || \ 2652 CARP_CHECK_WE_ARE_DST(iface)) { \ 2653 if (bif->bif_flags & IFBIF_LEARNING) { \ 2654 error = bridge_rtupdate(sc, eh->ether_shost, \ 2655 vlan, bif, 0, IFBAF_DYNAMIC); \ 2656 if (error && bif->bif_addrmax) { \ 2657 m_freem(m); \ 2658 return (NULL); \ 2659 } \ 2660 } \ 2661 m->m_pkthdr.rcvif = iface; \ 2662 if ((iface) == ifp) { \ 2663 /* Skip bridge processing... src == dest */ \ 2664 return (m); \ 2665 } \ 2666 /* It's passing over or to the bridge, locally. */ \ 2667 ETHER_BPF_MTAP(bifp, m); \ 2668 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \ 2669 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);\ 2670 /* Hand the packet over to netmap if necessary. */ \ 2671 GRAB_FOR_NETMAP(bifp, m); \ 2672 /* Filter on the physical interface. */ \ 2673 if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) || \ 2674 PFIL_HOOKED_INET6)) { \ 2675 if (bridge_pfil(&m, NULL, ifp, \ 2676 PFIL_IN) != 0 || m == NULL) { \ 2677 return (NULL); \ 2678 } \ 2679 } \ 2680 if ((iface) != bifp) \ 2681 ETHER_BPF_MTAP(iface, m); \ 2682 return (m); \ 2683 } \ 2684 \ 2685 /* We just received a packet that we sent out. */ \ 2686 if (memcmp(IF_LLADDR(iface), eh->ether_shost, ETHER_ADDR_LEN) == 0 || \ 2687 CARP_CHECK_WE_ARE_SRC(iface)) { \ 2688 m_freem(m); \ 2689 return (NULL); \ 2690 } 2691 2692 /* 2693 * Unicast. Make sure it's not for the bridge. 2694 */ 2695 do { GRAB_OUR_PACKETS(bifp) } while (0); 2696 2697 /* 2698 * Give a chance for ifp at first priority. This will help when the 2699 * packet comes through the interface like VLAN's with the same MACs 2700 * on several interfaces from the same bridge. This also will save 2701 * some CPU cycles in case the destination interface and the input 2702 * interface (eq ifp) are the same. 2703 */ 2704 do { GRAB_OUR_PACKETS(ifp) } while (0); 2705 2706 /* Now check the all bridge members. */ 2707 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) { 2708 GRAB_OUR_PACKETS(bif2->bif_ifp) 2709 } 2710 2711 #undef CARP_CHECK_WE_ARE_DST 2712 #undef CARP_CHECK_WE_ARE_SRC 2713 #undef PFIL_HOOKED_INET6 2714 #undef GRAB_FOR_NETMAP 2715 #undef GRAB_OUR_PACKETS 2716 2717 /* Perform the bridge forwarding function. */ 2718 bridge_forward(sc, bif, m); 2719 2720 return (NULL); 2721 } 2722 2723 /* 2724 * Inject a packet back into the host ethernet stack. This will generally only 2725 * be used by netmap when an application writes to the host TX ring. The 2726 * M_BRIDGE_INJECT flag ensures that the packet is re-routed to the bridge 2727 * interface after ethernet processing. 2728 */ 2729 static void 2730 bridge_inject(struct ifnet *ifp, struct mbuf *m) 2731 { 2732 struct bridge_softc *sc; 2733 2734 KASSERT((if_getcapenable(ifp) & IFCAP_NETMAP) != 0, 2735 ("%s: iface %s is not running in netmap mode", 2736 __func__, if_name(ifp))); 2737 KASSERT((m->m_flags & M_BRIDGE_INJECT) == 0, 2738 ("%s: mbuf %p has M_BRIDGE_INJECT set", __func__, m)); 2739 2740 m->m_flags |= M_BRIDGE_INJECT; 2741 sc = if_getsoftc(ifp); 2742 sc->sc_if_input(ifp, m); 2743 } 2744 2745 /* 2746 * bridge_broadcast: 2747 * 2748 * Send a frame to all interfaces that are members of 2749 * the bridge, except for the one on which the packet 2750 * arrived. 2751 * 2752 * NOTE: Releases the lock on return. 2753 */ 2754 static void 2755 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if, 2756 struct mbuf *m, int runfilt) 2757 { 2758 struct bridge_iflist *dbif, *sbif; 2759 struct mbuf *mc; 2760 struct ifnet *dst_if; 2761 int used = 0, i; 2762 2763 NET_EPOCH_ASSERT(); 2764 2765 sbif = bridge_lookup_member_if(sc, src_if); 2766 2767 /* Filter on the bridge interface before broadcasting */ 2768 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head) 2769 #ifdef INET6 2770 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2771 #endif 2772 )) { 2773 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0) 2774 return; 2775 if (m == NULL) 2776 return; 2777 } 2778 2779 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) { 2780 dst_if = dbif->bif_ifp; 2781 if (dst_if == src_if) 2782 continue; 2783 2784 /* Private segments can not talk to each other */ 2785 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)) 2786 continue; 2787 2788 if ((dbif->bif_flags & IFBIF_STP) && 2789 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2790 continue; 2791 2792 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 && 2793 (m->m_flags & (M_BCAST|M_MCAST)) == 0) 2794 continue; 2795 2796 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2797 continue; 2798 2799 if (CK_LIST_NEXT(dbif, bif_next) == NULL) { 2800 mc = m; 2801 used = 1; 2802 } else { 2803 mc = m_dup(m, M_NOWAIT); 2804 if (mc == NULL) { 2805 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 2806 continue; 2807 } 2808 } 2809 2810 /* 2811 * Filter on the output interface. Pass a NULL bridge interface 2812 * pointer so we do not redundantly filter on the bridge for 2813 * each interface we broadcast on. 2814 */ 2815 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head) 2816 #ifdef INET6 2817 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2818 #endif 2819 )) { 2820 if (used == 0) { 2821 /* Keep the layer3 header aligned */ 2822 i = min(mc->m_pkthdr.len, max_protohdr); 2823 mc = m_copyup(mc, i, ETHER_ALIGN); 2824 if (mc == NULL) { 2825 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 2826 continue; 2827 } 2828 } 2829 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0) 2830 continue; 2831 if (mc == NULL) 2832 continue; 2833 } 2834 2835 bridge_enqueue(sc, dst_if, mc); 2836 } 2837 if (used == 0) 2838 m_freem(m); 2839 } 2840 2841 /* 2842 * bridge_span: 2843 * 2844 * Duplicate a packet out one or more interfaces that are in span mode, 2845 * the original mbuf is unmodified. 2846 */ 2847 static void 2848 bridge_span(struct bridge_softc *sc, struct mbuf *m) 2849 { 2850 struct bridge_iflist *bif; 2851 struct ifnet *dst_if; 2852 struct mbuf *mc; 2853 2854 NET_EPOCH_ASSERT(); 2855 2856 if (CK_LIST_EMPTY(&sc->sc_spanlist)) 2857 return; 2858 2859 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) { 2860 dst_if = bif->bif_ifp; 2861 2862 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2863 continue; 2864 2865 mc = m_dup(m, M_NOWAIT); 2866 if (mc == NULL) { 2867 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 2868 continue; 2869 } 2870 2871 bridge_enqueue(sc, dst_if, mc); 2872 } 2873 } 2874 2875 /* 2876 * bridge_rtupdate: 2877 * 2878 * Add a bridge routing entry. 2879 */ 2880 static int 2881 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan, 2882 struct bridge_iflist *bif, int setflags, uint8_t flags) 2883 { 2884 struct bridge_rtnode *brt; 2885 struct bridge_iflist *obif; 2886 int error; 2887 2888 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc); 2889 2890 /* Check the source address is valid and not multicast. */ 2891 if (ETHER_IS_MULTICAST(dst) || 2892 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 && 2893 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0) 2894 return (EINVAL); 2895 2896 /* 2897 * A route for this destination might already exist. If so, 2898 * update it, otherwise create a new one. 2899 */ 2900 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) { 2901 BRIDGE_RT_LOCK(sc); 2902 2903 /* Check again, now that we have the lock. There could have 2904 * been a race and we only want to insert this once. */ 2905 if (bridge_rtnode_lookup(sc, dst, vlan) != NULL) { 2906 BRIDGE_RT_UNLOCK(sc); 2907 return (0); 2908 } 2909 2910 if (sc->sc_brtcnt >= sc->sc_brtmax) { 2911 sc->sc_brtexceeded++; 2912 BRIDGE_RT_UNLOCK(sc); 2913 return (ENOSPC); 2914 } 2915 /* Check per interface address limits (if enabled) */ 2916 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) { 2917 bif->bif_addrexceeded++; 2918 BRIDGE_RT_UNLOCK(sc); 2919 return (ENOSPC); 2920 } 2921 2922 /* 2923 * Allocate a new bridge forwarding node, and 2924 * initialize the expiration time and Ethernet 2925 * address. 2926 */ 2927 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO); 2928 if (brt == NULL) { 2929 BRIDGE_RT_UNLOCK(sc); 2930 return (ENOMEM); 2931 } 2932 brt->brt_vnet = curvnet; 2933 2934 if (bif->bif_flags & IFBIF_STICKY) 2935 brt->brt_flags = IFBAF_STICKY; 2936 else 2937 brt->brt_flags = IFBAF_DYNAMIC; 2938 2939 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN); 2940 brt->brt_vlan = vlan; 2941 2942 brt->brt_dst = bif; 2943 if ((error = bridge_rtnode_insert(sc, brt)) != 0) { 2944 uma_zfree(V_bridge_rtnode_zone, brt); 2945 BRIDGE_RT_UNLOCK(sc); 2946 return (error); 2947 } 2948 bif->bif_addrcnt++; 2949 2950 BRIDGE_RT_UNLOCK(sc); 2951 } 2952 2953 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 2954 (obif = brt->brt_dst) != bif) { 2955 MPASS(obif != NULL); 2956 2957 BRIDGE_RT_LOCK(sc); 2958 brt->brt_dst->bif_addrcnt--; 2959 brt->brt_dst = bif; 2960 brt->brt_dst->bif_addrcnt++; 2961 BRIDGE_RT_UNLOCK(sc); 2962 2963 if (V_log_mac_flap && 2964 ppsratecheck(&V_log_last, &V_log_count, V_log_interval)) { 2965 log(LOG_NOTICE, 2966 "%s: mac address %6D vlan %d moved from %s to %s\n", 2967 sc->sc_ifp->if_xname, 2968 &brt->brt_addr[0], ":", 2969 brt->brt_vlan, 2970 obif->bif_ifp->if_xname, 2971 bif->bif_ifp->if_xname); 2972 } 2973 } 2974 2975 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 2976 brt->brt_expire = time_uptime + sc->sc_brttimeout; 2977 if (setflags) 2978 brt->brt_flags = flags; 2979 2980 return (0); 2981 } 2982 2983 /* 2984 * bridge_rtlookup: 2985 * 2986 * Lookup the destination interface for an address. 2987 */ 2988 static struct ifnet * 2989 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) 2990 { 2991 struct bridge_rtnode *brt; 2992 2993 NET_EPOCH_ASSERT(); 2994 2995 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL) 2996 return (NULL); 2997 2998 return (brt->brt_ifp); 2999 } 3000 3001 /* 3002 * bridge_rttrim: 3003 * 3004 * Trim the routine table so that we have a number 3005 * of routing entries less than or equal to the 3006 * maximum number. 3007 */ 3008 static void 3009 bridge_rttrim(struct bridge_softc *sc) 3010 { 3011 struct bridge_rtnode *brt, *nbrt; 3012 3013 NET_EPOCH_ASSERT(); 3014 BRIDGE_RT_LOCK_ASSERT(sc); 3015 3016 /* Make sure we actually need to do this. */ 3017 if (sc->sc_brtcnt <= sc->sc_brtmax) 3018 return; 3019 3020 /* Force an aging cycle; this might trim enough addresses. */ 3021 bridge_rtage(sc); 3022 if (sc->sc_brtcnt <= sc->sc_brtmax) 3023 return; 3024 3025 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 3026 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { 3027 bridge_rtnode_destroy(sc, brt); 3028 if (sc->sc_brtcnt <= sc->sc_brtmax) 3029 return; 3030 } 3031 } 3032 } 3033 3034 /* 3035 * bridge_timer: 3036 * 3037 * Aging timer for the bridge. 3038 */ 3039 static void 3040 bridge_timer(void *arg) 3041 { 3042 struct bridge_softc *sc = arg; 3043 3044 BRIDGE_RT_LOCK_ASSERT(sc); 3045 3046 /* Destruction of rtnodes requires a proper vnet context */ 3047 CURVNET_SET(sc->sc_ifp->if_vnet); 3048 bridge_rtage(sc); 3049 3050 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) 3051 callout_reset(&sc->sc_brcallout, 3052 bridge_rtable_prune_period * hz, bridge_timer, sc); 3053 CURVNET_RESTORE(); 3054 } 3055 3056 /* 3057 * bridge_rtage: 3058 * 3059 * Perform an aging cycle. 3060 */ 3061 static void 3062 bridge_rtage(struct bridge_softc *sc) 3063 { 3064 struct bridge_rtnode *brt, *nbrt; 3065 3066 BRIDGE_RT_LOCK_ASSERT(sc); 3067 3068 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 3069 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { 3070 if (time_uptime >= brt->brt_expire) 3071 bridge_rtnode_destroy(sc, brt); 3072 } 3073 } 3074 } 3075 3076 /* 3077 * bridge_rtflush: 3078 * 3079 * Remove all dynamic addresses from the bridge. 3080 */ 3081 static void 3082 bridge_rtflush(struct bridge_softc *sc, int full) 3083 { 3084 struct bridge_rtnode *brt, *nbrt; 3085 3086 BRIDGE_RT_LOCK_ASSERT(sc); 3087 3088 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 3089 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 3090 bridge_rtnode_destroy(sc, brt); 3091 } 3092 } 3093 3094 /* 3095 * bridge_rtdaddr: 3096 * 3097 * Remove an address from the table. 3098 */ 3099 static int 3100 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) 3101 { 3102 struct bridge_rtnode *brt; 3103 int found = 0; 3104 3105 BRIDGE_RT_LOCK(sc); 3106 3107 /* 3108 * If vlan is DOT1Q_VID_RSVD_IMPL then we want to delete for all vlans 3109 * so the lookup may return more than one. 3110 */ 3111 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) { 3112 bridge_rtnode_destroy(sc, brt); 3113 found = 1; 3114 } 3115 3116 BRIDGE_RT_UNLOCK(sc); 3117 3118 return (found ? 0 : ENOENT); 3119 } 3120 3121 /* 3122 * bridge_rtdelete: 3123 * 3124 * Delete routes to a speicifc member interface. 3125 */ 3126 static void 3127 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full) 3128 { 3129 struct bridge_rtnode *brt, *nbrt; 3130 3131 BRIDGE_RT_LOCK_ASSERT(sc); 3132 3133 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 3134 if (brt->brt_ifp == ifp && (full || 3135 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)) 3136 bridge_rtnode_destroy(sc, brt); 3137 } 3138 } 3139 3140 /* 3141 * bridge_rtable_init: 3142 * 3143 * Initialize the route table for this bridge. 3144 */ 3145 static void 3146 bridge_rtable_init(struct bridge_softc *sc) 3147 { 3148 int i; 3149 3150 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE, 3151 M_DEVBUF, M_WAITOK); 3152 3153 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++) 3154 CK_LIST_INIT(&sc->sc_rthash[i]); 3155 3156 sc->sc_rthash_key = arc4random(); 3157 CK_LIST_INIT(&sc->sc_rtlist); 3158 } 3159 3160 /* 3161 * bridge_rtable_fini: 3162 * 3163 * Deconstruct the route table for this bridge. 3164 */ 3165 static void 3166 bridge_rtable_fini(struct bridge_softc *sc) 3167 { 3168 3169 KASSERT(sc->sc_brtcnt == 0, 3170 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt)); 3171 free(sc->sc_rthash, M_DEVBUF); 3172 } 3173 3174 /* 3175 * The following hash function is adapted from "Hash Functions" by Bob Jenkins 3176 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997). 3177 */ 3178 #define mix(a, b, c) \ 3179 do { \ 3180 a -= b; a -= c; a ^= (c >> 13); \ 3181 b -= c; b -= a; b ^= (a << 8); \ 3182 c -= a; c -= b; c ^= (b >> 13); \ 3183 a -= b; a -= c; a ^= (c >> 12); \ 3184 b -= c; b -= a; b ^= (a << 16); \ 3185 c -= a; c -= b; c ^= (b >> 5); \ 3186 a -= b; a -= c; a ^= (c >> 3); \ 3187 b -= c; b -= a; b ^= (a << 10); \ 3188 c -= a; c -= b; c ^= (b >> 15); \ 3189 } while (/*CONSTCOND*/0) 3190 3191 static __inline uint32_t 3192 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr) 3193 { 3194 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key; 3195 3196 b += addr[5] << 8; 3197 b += addr[4]; 3198 a += addr[3] << 24; 3199 a += addr[2] << 16; 3200 a += addr[1] << 8; 3201 a += addr[0]; 3202 3203 mix(a, b, c); 3204 3205 return (c & BRIDGE_RTHASH_MASK); 3206 } 3207 3208 #undef mix 3209 3210 static int 3211 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b) 3212 { 3213 int i, d; 3214 3215 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) { 3216 d = ((int)a[i]) - ((int)b[i]); 3217 } 3218 3219 return (d); 3220 } 3221 3222 /* 3223 * bridge_rtnode_lookup: 3224 * 3225 * Look up a bridge route node for the specified destination. Compare the 3226 * vlan id or if zero then just return the first match. 3227 */ 3228 static struct bridge_rtnode * 3229 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) 3230 { 3231 struct bridge_rtnode *brt; 3232 uint32_t hash; 3233 int dir; 3234 3235 BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc); 3236 3237 hash = bridge_rthash(sc, addr); 3238 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) { 3239 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr); 3240 if (dir == 0 && (brt->brt_vlan == vlan || vlan == DOT1Q_VID_RSVD_IMPL)) 3241 return (brt); 3242 if (dir > 0) 3243 return (NULL); 3244 } 3245 3246 return (NULL); 3247 } 3248 3249 /* 3250 * bridge_rtnode_insert: 3251 * 3252 * Insert the specified bridge node into the route table. We 3253 * assume the entry is not already in the table. 3254 */ 3255 static int 3256 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt) 3257 { 3258 struct bridge_rtnode *lbrt; 3259 uint32_t hash; 3260 int dir; 3261 3262 BRIDGE_RT_LOCK_ASSERT(sc); 3263 3264 hash = bridge_rthash(sc, brt->brt_addr); 3265 3266 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]); 3267 if (lbrt == NULL) { 3268 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash); 3269 goto out; 3270 } 3271 3272 do { 3273 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr); 3274 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan) 3275 return (EEXIST); 3276 if (dir > 0) { 3277 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash); 3278 goto out; 3279 } 3280 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) { 3281 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash); 3282 goto out; 3283 } 3284 lbrt = CK_LIST_NEXT(lbrt, brt_hash); 3285 } while (lbrt != NULL); 3286 3287 #ifdef DIAGNOSTIC 3288 panic("bridge_rtnode_insert: impossible"); 3289 #endif 3290 3291 out: 3292 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list); 3293 sc->sc_brtcnt++; 3294 3295 return (0); 3296 } 3297 3298 static void 3299 bridge_rtnode_destroy_cb(struct epoch_context *ctx) 3300 { 3301 struct bridge_rtnode *brt; 3302 3303 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx); 3304 3305 CURVNET_SET(brt->brt_vnet); 3306 uma_zfree(V_bridge_rtnode_zone, brt); 3307 CURVNET_RESTORE(); 3308 } 3309 3310 /* 3311 * bridge_rtnode_destroy: 3312 * 3313 * Destroy a bridge rtnode. 3314 */ 3315 static void 3316 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt) 3317 { 3318 BRIDGE_RT_LOCK_ASSERT(sc); 3319 3320 CK_LIST_REMOVE(brt, brt_hash); 3321 3322 CK_LIST_REMOVE(brt, brt_list); 3323 sc->sc_brtcnt--; 3324 brt->brt_dst->bif_addrcnt--; 3325 3326 NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx); 3327 } 3328 3329 /* 3330 * bridge_rtable_expire: 3331 * 3332 * Set the expiry time for all routes on an interface. 3333 */ 3334 static void 3335 bridge_rtable_expire(struct ifnet *ifp, int age) 3336 { 3337 struct bridge_softc *sc = ifp->if_bridge; 3338 struct bridge_rtnode *brt; 3339 3340 CURVNET_SET(ifp->if_vnet); 3341 BRIDGE_RT_LOCK(sc); 3342 3343 /* 3344 * If the age is zero then flush, otherwise set all the expiry times to 3345 * age for the interface 3346 */ 3347 if (age == 0) 3348 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN); 3349 else { 3350 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) { 3351 /* Cap the expiry time to 'age' */ 3352 if (brt->brt_ifp == ifp && 3353 brt->brt_expire > time_uptime + age && 3354 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 3355 brt->brt_expire = time_uptime + age; 3356 } 3357 } 3358 BRIDGE_RT_UNLOCK(sc); 3359 CURVNET_RESTORE(); 3360 } 3361 3362 /* 3363 * bridge_state_change: 3364 * 3365 * Callback from the bridgestp code when a port changes states. 3366 */ 3367 static void 3368 bridge_state_change(struct ifnet *ifp, int state) 3369 { 3370 struct bridge_softc *sc = ifp->if_bridge; 3371 static const char *stpstates[] = { 3372 "disabled", 3373 "listening", 3374 "learning", 3375 "forwarding", 3376 "blocking", 3377 "discarding" 3378 }; 3379 3380 CURVNET_SET(ifp->if_vnet); 3381 if (V_log_stp) 3382 log(LOG_NOTICE, "%s: state changed to %s on %s\n", 3383 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname); 3384 CURVNET_RESTORE(); 3385 } 3386 3387 /* 3388 * Send bridge packets through pfil if they are one of the types pfil can deal 3389 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without 3390 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for 3391 * that interface. 3392 */ 3393 static int 3394 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir) 3395 { 3396 int snap, error, i, hlen; 3397 struct ether_header *eh1, eh2; 3398 struct ip *ip; 3399 struct llc llc1; 3400 u_int16_t ether_type; 3401 pfil_return_t rv; 3402 3403 snap = 0; 3404 error = -1; /* Default error if not error == 0 */ 3405 3406 #if 0 3407 /* we may return with the IP fields swapped, ensure its not shared */ 3408 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__)); 3409 #endif 3410 3411 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0) 3412 return (0); /* filtering is disabled */ 3413 3414 i = min((*mp)->m_pkthdr.len, max_protohdr); 3415 if ((*mp)->m_len < i) { 3416 *mp = m_pullup(*mp, i); 3417 if (*mp == NULL) { 3418 printf("%s: m_pullup failed\n", __func__); 3419 return (-1); 3420 } 3421 } 3422 3423 eh1 = mtod(*mp, struct ether_header *); 3424 ether_type = ntohs(eh1->ether_type); 3425 3426 /* 3427 * Check for SNAP/LLC. 3428 */ 3429 if (ether_type < ETHERMTU) { 3430 struct llc *llc2 = (struct llc *)(eh1 + 1); 3431 3432 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 && 3433 llc2->llc_dsap == LLC_SNAP_LSAP && 3434 llc2->llc_ssap == LLC_SNAP_LSAP && 3435 llc2->llc_control == LLC_UI) { 3436 ether_type = htons(llc2->llc_un.type_snap.ether_type); 3437 snap = 1; 3438 } 3439 } 3440 3441 /* 3442 * If we're trying to filter bridge traffic, don't look at anything 3443 * other than IP and ARP traffic. If the filter doesn't understand 3444 * IPv6, don't allow IPv6 through the bridge either. This is lame 3445 * since if we really wanted, say, an AppleTalk filter, we are hosed, 3446 * but of course we don't have an AppleTalk filter to begin with. 3447 * (Note that since pfil doesn't understand ARP it will pass *ALL* 3448 * ARP traffic.) 3449 */ 3450 switch (ether_type) { 3451 case ETHERTYPE_ARP: 3452 case ETHERTYPE_REVARP: 3453 if (V_pfil_ipfw_arp == 0) 3454 return (0); /* Automatically pass */ 3455 break; 3456 3457 case ETHERTYPE_IP: 3458 #ifdef INET6 3459 case ETHERTYPE_IPV6: 3460 #endif /* INET6 */ 3461 break; 3462 default: 3463 /* 3464 * Check to see if the user wants to pass non-ip 3465 * packets, these will not be checked by pfil(9) and 3466 * passed unconditionally so the default is to drop. 3467 */ 3468 if (V_pfil_onlyip) 3469 goto bad; 3470 } 3471 3472 /* Run the packet through pfil before stripping link headers */ 3473 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 && 3474 dir == PFIL_OUT && ifp != NULL) { 3475 switch (pfil_mbuf_out(V_link_pfil_head, mp, ifp, NULL)) { 3476 case PFIL_DROPPED: 3477 return (EACCES); 3478 case PFIL_CONSUMED: 3479 return (0); 3480 } 3481 } 3482 3483 /* Strip off the Ethernet header and keep a copy. */ 3484 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2); 3485 m_adj(*mp, ETHER_HDR_LEN); 3486 3487 /* Strip off snap header, if present */ 3488 if (snap) { 3489 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1); 3490 m_adj(*mp, sizeof(struct llc)); 3491 } 3492 3493 /* 3494 * Check the IP header for alignment and errors 3495 */ 3496 if (dir == PFIL_IN) { 3497 switch (ether_type) { 3498 case ETHERTYPE_IP: 3499 error = bridge_ip_checkbasic(mp); 3500 break; 3501 #ifdef INET6 3502 case ETHERTYPE_IPV6: 3503 error = bridge_ip6_checkbasic(mp); 3504 break; 3505 #endif /* INET6 */ 3506 default: 3507 error = 0; 3508 } 3509 if (error) 3510 goto bad; 3511 } 3512 3513 error = 0; 3514 3515 /* 3516 * Run the packet through pfil 3517 */ 3518 rv = PFIL_PASS; 3519 switch (ether_type) { 3520 case ETHERTYPE_IP: 3521 /* 3522 * Run pfil on the member interface and the bridge, both can 3523 * be skipped by clearing pfil_member or pfil_bridge. 3524 * 3525 * Keep the order: 3526 * in_if -> bridge_if -> out_if 3527 */ 3528 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv = 3529 pfil_mbuf_out(V_inet_pfil_head, mp, bifp, NULL)) != 3530 PFIL_PASS) 3531 break; 3532 3533 if (V_pfil_member && ifp != NULL) { 3534 rv = (dir == PFIL_OUT) ? 3535 pfil_mbuf_out(V_inet_pfil_head, mp, ifp, NULL) : 3536 pfil_mbuf_in(V_inet_pfil_head, mp, ifp, NULL); 3537 if (rv != PFIL_PASS) 3538 break; 3539 } 3540 3541 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv = 3542 pfil_mbuf_in(V_inet_pfil_head, mp, bifp, NULL)) != 3543 PFIL_PASS) 3544 break; 3545 3546 /* check if we need to fragment the packet */ 3547 /* bridge_fragment generates a mbuf chain of packets */ 3548 /* that already include eth headers */ 3549 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) { 3550 i = (*mp)->m_pkthdr.len; 3551 if (i > ifp->if_mtu) { 3552 error = bridge_fragment(ifp, mp, &eh2, snap, 3553 &llc1); 3554 return (error); 3555 } 3556 } 3557 3558 /* Recalculate the ip checksum. */ 3559 ip = mtod(*mp, struct ip *); 3560 hlen = ip->ip_hl << 2; 3561 if (hlen < sizeof(struct ip)) 3562 goto bad; 3563 if (hlen > (*mp)->m_len) { 3564 if ((*mp = m_pullup(*mp, hlen)) == NULL) 3565 goto bad; 3566 ip = mtod(*mp, struct ip *); 3567 if (ip == NULL) 3568 goto bad; 3569 } 3570 ip->ip_sum = 0; 3571 if (hlen == sizeof(struct ip)) 3572 ip->ip_sum = in_cksum_hdr(ip); 3573 else 3574 ip->ip_sum = in_cksum(*mp, hlen); 3575 3576 break; 3577 #ifdef INET6 3578 case ETHERTYPE_IPV6: 3579 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv = 3580 pfil_mbuf_out(V_inet6_pfil_head, mp, bifp, NULL)) != 3581 PFIL_PASS) 3582 break; 3583 3584 if (V_pfil_member && ifp != NULL) { 3585 rv = (dir == PFIL_OUT) ? 3586 pfil_mbuf_out(V_inet6_pfil_head, mp, ifp, NULL) : 3587 pfil_mbuf_in(V_inet6_pfil_head, mp, ifp, NULL); 3588 if (rv != PFIL_PASS) 3589 break; 3590 } 3591 3592 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv = 3593 pfil_mbuf_in(V_inet6_pfil_head, mp, bifp, NULL)) != 3594 PFIL_PASS) 3595 break; 3596 break; 3597 #endif 3598 } 3599 3600 switch (rv) { 3601 case PFIL_CONSUMED: 3602 return (0); 3603 case PFIL_DROPPED: 3604 return (EACCES); 3605 default: 3606 break; 3607 } 3608 3609 error = -1; 3610 3611 /* 3612 * Finally, put everything back the way it was and return 3613 */ 3614 if (snap) { 3615 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT); 3616 if (*mp == NULL) 3617 return (error); 3618 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc)); 3619 } 3620 3621 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT); 3622 if (*mp == NULL) 3623 return (error); 3624 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN); 3625 3626 return (0); 3627 3628 bad: 3629 m_freem(*mp); 3630 *mp = NULL; 3631 return (error); 3632 } 3633 3634 /* 3635 * Perform basic checks on header size since 3636 * pfil assumes ip_input has already processed 3637 * it for it. Cut-and-pasted from ip_input.c. 3638 * Given how simple the IPv6 version is, 3639 * does the IPv4 version really need to be 3640 * this complicated? 3641 * 3642 * XXX Should we update ipstat here, or not? 3643 * XXX Right now we update ipstat but not 3644 * XXX csum_counter. 3645 */ 3646 static int 3647 bridge_ip_checkbasic(struct mbuf **mp) 3648 { 3649 struct mbuf *m = *mp; 3650 struct ip *ip; 3651 int len, hlen; 3652 u_short sum; 3653 3654 if (*mp == NULL) 3655 return (-1); 3656 3657 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 3658 if ((m = m_copyup(m, sizeof(struct ip), 3659 (max_linkhdr + 3) & ~3)) == NULL) { 3660 /* XXXJRT new stat, please */ 3661 KMOD_IPSTAT_INC(ips_toosmall); 3662 goto bad; 3663 } 3664 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 3665 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 3666 KMOD_IPSTAT_INC(ips_toosmall); 3667 goto bad; 3668 } 3669 } 3670 ip = mtod(m, struct ip *); 3671 if (ip == NULL) goto bad; 3672 3673 if (ip->ip_v != IPVERSION) { 3674 KMOD_IPSTAT_INC(ips_badvers); 3675 goto bad; 3676 } 3677 hlen = ip->ip_hl << 2; 3678 if (hlen < sizeof(struct ip)) { /* minimum header length */ 3679 KMOD_IPSTAT_INC(ips_badhlen); 3680 goto bad; 3681 } 3682 if (hlen > m->m_len) { 3683 if ((m = m_pullup(m, hlen)) == NULL) { 3684 KMOD_IPSTAT_INC(ips_badhlen); 3685 goto bad; 3686 } 3687 ip = mtod(m, struct ip *); 3688 if (ip == NULL) goto bad; 3689 } 3690 3691 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) { 3692 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID); 3693 } else { 3694 if (hlen == sizeof(struct ip)) { 3695 sum = in_cksum_hdr(ip); 3696 } else { 3697 sum = in_cksum(m, hlen); 3698 } 3699 } 3700 if (sum) { 3701 KMOD_IPSTAT_INC(ips_badsum); 3702 goto bad; 3703 } 3704 3705 /* Retrieve the packet length. */ 3706 len = ntohs(ip->ip_len); 3707 3708 /* 3709 * Check for additional length bogosity 3710 */ 3711 if (len < hlen) { 3712 KMOD_IPSTAT_INC(ips_badlen); 3713 goto bad; 3714 } 3715 3716 /* 3717 * Check that the amount of data in the buffers 3718 * is as at least much as the IP header would have us expect. 3719 * Drop packet if shorter than we expect. 3720 */ 3721 if (m->m_pkthdr.len < len) { 3722 KMOD_IPSTAT_INC(ips_tooshort); 3723 goto bad; 3724 } 3725 3726 /* Checks out, proceed */ 3727 *mp = m; 3728 return (0); 3729 3730 bad: 3731 *mp = m; 3732 return (-1); 3733 } 3734 3735 #ifdef INET6 3736 /* 3737 * Same as above, but for IPv6. 3738 * Cut-and-pasted from ip6_input.c. 3739 * XXX Should we update ip6stat, or not? 3740 */ 3741 static int 3742 bridge_ip6_checkbasic(struct mbuf **mp) 3743 { 3744 struct mbuf *m = *mp; 3745 struct ip6_hdr *ip6; 3746 3747 /* 3748 * If the IPv6 header is not aligned, slurp it up into a new 3749 * mbuf with space for link headers, in the event we forward 3750 * it. Otherwise, if it is aligned, make sure the entire base 3751 * IPv6 header is in the first mbuf of the chain. 3752 */ 3753 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 3754 struct ifnet *inifp = m->m_pkthdr.rcvif; 3755 if ((m = m_copyup(m, sizeof(struct ip6_hdr), 3756 (max_linkhdr + 3) & ~3)) == NULL) { 3757 /* XXXJRT new stat, please */ 3758 IP6STAT_INC(ip6s_toosmall); 3759 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 3760 goto bad; 3761 } 3762 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) { 3763 struct ifnet *inifp = m->m_pkthdr.rcvif; 3764 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 3765 IP6STAT_INC(ip6s_toosmall); 3766 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 3767 goto bad; 3768 } 3769 } 3770 3771 ip6 = mtod(m, struct ip6_hdr *); 3772 3773 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 3774 IP6STAT_INC(ip6s_badvers); 3775 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 3776 goto bad; 3777 } 3778 3779 /* Checks out, proceed */ 3780 *mp = m; 3781 return (0); 3782 3783 bad: 3784 *mp = m; 3785 return (-1); 3786 } 3787 #endif /* INET6 */ 3788 3789 /* 3790 * bridge_fragment: 3791 * 3792 * Fragment mbuf chain in multiple packets and prepend ethernet header. 3793 */ 3794 static int 3795 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh, 3796 int snap, struct llc *llc) 3797 { 3798 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL; 3799 struct ip *ip; 3800 int error = -1; 3801 3802 if (m->m_len < sizeof(struct ip) && 3803 (m = m_pullup(m, sizeof(struct ip))) == NULL) 3804 goto dropit; 3805 ip = mtod(m, struct ip *); 3806 3807 m->m_pkthdr.csum_flags |= CSUM_IP; 3808 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist); 3809 if (error) 3810 goto dropit; 3811 3812 /* 3813 * Walk the chain and re-add the Ethernet header for 3814 * each mbuf packet. 3815 */ 3816 for (mcur = m; mcur; mcur = mcur->m_nextpkt) { 3817 nextpkt = mcur->m_nextpkt; 3818 mcur->m_nextpkt = NULL; 3819 if (snap) { 3820 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT); 3821 if (mcur == NULL) { 3822 error = ENOBUFS; 3823 if (mprev != NULL) 3824 mprev->m_nextpkt = nextpkt; 3825 goto dropit; 3826 } 3827 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc)); 3828 } 3829 3830 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT); 3831 if (mcur == NULL) { 3832 error = ENOBUFS; 3833 if (mprev != NULL) 3834 mprev->m_nextpkt = nextpkt; 3835 goto dropit; 3836 } 3837 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN); 3838 3839 /* 3840 * The previous two M_PREPEND could have inserted one or two 3841 * mbufs in front so we have to update the previous packet's 3842 * m_nextpkt. 3843 */ 3844 mcur->m_nextpkt = nextpkt; 3845 if (mprev != NULL) 3846 mprev->m_nextpkt = mcur; 3847 else { 3848 /* The first mbuf in the original chain needs to be 3849 * updated. */ 3850 *mp = mcur; 3851 } 3852 mprev = mcur; 3853 } 3854 3855 KMOD_IPSTAT_INC(ips_fragmented); 3856 return (error); 3857 3858 dropit: 3859 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */ 3860 m = mcur->m_nextpkt; 3861 m_freem(mcur); 3862 } 3863 return (error); 3864 } 3865 3866 static void 3867 bridge_linkstate(struct ifnet *ifp) 3868 { 3869 struct bridge_softc *sc = ifp->if_bridge; 3870 struct bridge_iflist *bif; 3871 struct epoch_tracker et; 3872 3873 NET_EPOCH_ENTER(et); 3874 3875 bif = bridge_lookup_member_if(sc, ifp); 3876 if (bif == NULL) { 3877 NET_EPOCH_EXIT(et); 3878 return; 3879 } 3880 bridge_linkcheck(sc); 3881 3882 bstp_linkstate(&bif->bif_stp); 3883 3884 NET_EPOCH_EXIT(et); 3885 } 3886 3887 static void 3888 bridge_linkcheck(struct bridge_softc *sc) 3889 { 3890 struct bridge_iflist *bif; 3891 int new_link, hasls; 3892 3893 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc); 3894 3895 new_link = LINK_STATE_DOWN; 3896 hasls = 0; 3897 /* Our link is considered up if at least one of our ports is active */ 3898 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 3899 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE) 3900 hasls++; 3901 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) { 3902 new_link = LINK_STATE_UP; 3903 break; 3904 } 3905 } 3906 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) { 3907 /* If no interfaces support link-state then we default to up */ 3908 new_link = LINK_STATE_UP; 3909 } 3910 if_link_state_change(sc->sc_ifp, new_link); 3911 } 3912