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 __FBSDID("$FreeBSD$"); 81 82 #include "opt_inet.h" 83 #include "opt_inet6.h" 84 85 #include <sys/param.h> 86 #include <sys/eventhandler.h> 87 #include <sys/mbuf.h> 88 #include <sys/malloc.h> 89 #include <sys/protosw.h> 90 #include <sys/systm.h> 91 #include <sys/jail.h> 92 #include <sys/time.h> 93 #include <sys/socket.h> /* for net/if.h */ 94 #include <sys/sockio.h> 95 #include <sys/ctype.h> /* string functions */ 96 #include <sys/kernel.h> 97 #include <sys/random.h> 98 #include <sys/syslog.h> 99 #include <sys/sysctl.h> 100 #include <vm/uma.h> 101 #include <sys/module.h> 102 #include <sys/priv.h> 103 #include <sys/proc.h> 104 #include <sys/lock.h> 105 #include <sys/mutex.h> 106 107 #include <net/bpf.h> 108 #include <net/if.h> 109 #include <net/if_clone.h> 110 #include <net/if_dl.h> 111 #include <net/if_types.h> 112 #include <net/if_var.h> 113 #include <net/pfil.h> 114 #include <net/vnet.h> 115 116 #include <netinet/in.h> 117 #include <netinet/in_systm.h> 118 #include <netinet/in_var.h> 119 #include <netinet/ip.h> 120 #include <netinet/ip_var.h> 121 #ifdef INET6 122 #include <netinet/ip6.h> 123 #include <netinet6/ip6_var.h> 124 #include <netinet6/in6_ifattach.h> 125 #endif 126 #if defined(INET) || defined(INET6) 127 #include <netinet/ip_carp.h> 128 #endif 129 #include <machine/in_cksum.h> 130 #include <netinet/if_ether.h> 131 #include <net/bridgestp.h> 132 #include <net/if_bridgevar.h> 133 #include <net/if_llc.h> 134 #include <net/if_vlan_var.h> 135 136 #include <net/route.h> 137 138 #ifdef INET6 139 /* 140 * XXX: declare here to avoid to include many inet6 related files.. 141 * should be more generalized? 142 */ 143 extern void nd6_setmtu(struct ifnet *); 144 #endif 145 146 /* 147 * Size of the route hash table. Must be a power of two. 148 */ 149 #ifndef BRIDGE_RTHASH_SIZE 150 #define BRIDGE_RTHASH_SIZE 1024 151 #endif 152 153 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1) 154 155 /* 156 * Default maximum number of addresses to cache. 157 */ 158 #ifndef BRIDGE_RTABLE_MAX 159 #define BRIDGE_RTABLE_MAX 2000 160 #endif 161 162 /* 163 * Timeout (in seconds) for entries learned dynamically. 164 */ 165 #ifndef BRIDGE_RTABLE_TIMEOUT 166 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */ 167 #endif 168 169 /* 170 * Number of seconds between walks of the route list. 171 */ 172 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD 173 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60) 174 #endif 175 176 /* 177 * List of capabilities to possibly mask on the member interface. 178 */ 179 #define BRIDGE_IFCAPS_MASK (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\ 180 IFCAP_TXCSUM_IPV6) 181 182 /* 183 * List of capabilities to strip 184 */ 185 #define BRIDGE_IFCAPS_STRIP IFCAP_LRO 186 187 /* 188 * Bridge locking 189 */ 190 #define BRIDGE_LOCK_INIT(_sc) do { \ 191 mtx_init(&(_sc)->sc_mtx, "if_bridge", NULL, MTX_DEF); \ 192 } while (0) 193 #define BRIDGE_LOCK_DESTROY(_sc) do { \ 194 mtx_destroy(&(_sc)->sc_mtx); \ 195 } while (0) 196 #define BRIDGE_LOCK(_sc) mtx_lock(&(_sc)->sc_mtx) 197 #define BRIDGE_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_mtx) 198 #define BRIDGE_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_mtx, MA_OWNED) 199 #define BRIDGE_UNLOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_mtx, MA_NOTOWNED) 200 201 /* 202 * Bridge interface list entry. 203 */ 204 struct bridge_iflist { 205 CK_LIST_ENTRY(bridge_iflist) bif_next; 206 struct ifnet *bif_ifp; /* member if */ 207 struct bstp_port bif_stp; /* STP state */ 208 uint32_t bif_flags; /* member if flags */ 209 int bif_savedcaps; /* saved capabilities */ 210 uint32_t bif_addrmax; /* max # of addresses */ 211 uint32_t bif_addrcnt; /* cur. # of addresses */ 212 uint32_t bif_addrexceeded;/* # of address violations */ 213 struct epoch_context bif_epoch_ctx; 214 }; 215 216 /* 217 * Bridge route node. 218 */ 219 struct bridge_rtnode { 220 CK_LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */ 221 CK_LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */ 222 struct bridge_iflist *brt_dst; /* destination if */ 223 unsigned long brt_expire; /* expiration time */ 224 uint8_t brt_flags; /* address flags */ 225 uint8_t brt_addr[ETHER_ADDR_LEN]; 226 uint16_t brt_vlan; /* vlan id */ 227 struct vnet *brt_vnet; 228 struct epoch_context brt_epoch_ctx; 229 }; 230 #define brt_ifp brt_dst->bif_ifp 231 232 /* 233 * Software state for each bridge. 234 */ 235 struct bridge_softc { 236 struct ifnet *sc_ifp; /* make this an interface */ 237 LIST_ENTRY(bridge_softc) sc_list; 238 struct mtx sc_mtx; 239 uint32_t sc_brtmax; /* max # of addresses */ 240 uint32_t sc_brtcnt; /* cur. # of addresses */ 241 uint32_t sc_brttimeout; /* rt timeout in seconds */ 242 struct callout sc_brcallout; /* bridge callout */ 243 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */ 244 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */ 245 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */ 246 uint32_t sc_rthash_key; /* key for hash */ 247 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */ 248 struct bstp_state sc_stp; /* STP state */ 249 uint32_t sc_brtexceeded; /* # of cache drops */ 250 struct ifnet *sc_ifaddr; /* member mac copied from */ 251 struct ether_addr sc_defaddr; /* Default MAC address */ 252 struct epoch_context sc_epoch_ctx; 253 }; 254 255 VNET_DEFINE_STATIC(struct mtx, bridge_list_mtx); 256 #define V_bridge_list_mtx VNET(bridge_list_mtx) 257 static eventhandler_tag bridge_detach_cookie; 258 259 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD; 260 261 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone); 262 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone) 263 264 static int bridge_clone_create(struct if_clone *, int, caddr_t); 265 static void bridge_clone_destroy(struct ifnet *); 266 267 static int bridge_ioctl(struct ifnet *, u_long, caddr_t); 268 static void bridge_mutecaps(struct bridge_softc *); 269 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *, 270 int); 271 static void bridge_ifdetach(void *arg __unused, struct ifnet *); 272 static void bridge_init(void *); 273 static void bridge_dummynet(struct mbuf *, struct ifnet *); 274 static void bridge_stop(struct ifnet *, int); 275 static int bridge_transmit(struct ifnet *, struct mbuf *); 276 static void bridge_qflush(struct ifnet *); 277 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *); 278 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *, 279 struct rtentry *); 280 static int bridge_enqueue(struct bridge_softc *, struct ifnet *, 281 struct mbuf *); 282 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int); 283 284 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *, 285 struct mbuf *m); 286 287 static void bridge_timer(void *); 288 289 static void bridge_broadcast(struct bridge_softc *, struct ifnet *, 290 struct mbuf *, int); 291 static void bridge_span(struct bridge_softc *, struct mbuf *); 292 293 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *, 294 uint16_t, struct bridge_iflist *, int, uint8_t); 295 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *, 296 uint16_t); 297 static void bridge_rttrim(struct bridge_softc *); 298 static void bridge_rtage(struct bridge_softc *); 299 static void bridge_rtflush(struct bridge_softc *, int); 300 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *, 301 uint16_t); 302 303 static void bridge_rtable_init(struct bridge_softc *); 304 static void bridge_rtable_fini(struct bridge_softc *); 305 306 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *); 307 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *, 308 const uint8_t *, uint16_t); 309 static int bridge_rtnode_insert(struct bridge_softc *, 310 struct bridge_rtnode *); 311 static void bridge_rtnode_destroy(struct bridge_softc *, 312 struct bridge_rtnode *); 313 static void bridge_rtable_expire(struct ifnet *, int); 314 static void bridge_state_change(struct ifnet *, int); 315 316 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *, 317 const char *name); 318 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *, 319 struct ifnet *ifp); 320 static void bridge_delete_member(struct bridge_softc *, 321 struct bridge_iflist *, int); 322 static void bridge_delete_span(struct bridge_softc *, 323 struct bridge_iflist *); 324 325 static int bridge_ioctl_add(struct bridge_softc *, void *); 326 static int bridge_ioctl_del(struct bridge_softc *, void *); 327 static int bridge_ioctl_gifflags(struct bridge_softc *, void *); 328 static int bridge_ioctl_sifflags(struct bridge_softc *, void *); 329 static int bridge_ioctl_scache(struct bridge_softc *, void *); 330 static int bridge_ioctl_gcache(struct bridge_softc *, void *); 331 static int bridge_ioctl_gifs(struct bridge_softc *, void *); 332 static int bridge_ioctl_rts(struct bridge_softc *, void *); 333 static int bridge_ioctl_saddr(struct bridge_softc *, void *); 334 static int bridge_ioctl_sto(struct bridge_softc *, void *); 335 static int bridge_ioctl_gto(struct bridge_softc *, void *); 336 static int bridge_ioctl_daddr(struct bridge_softc *, void *); 337 static int bridge_ioctl_flush(struct bridge_softc *, void *); 338 static int bridge_ioctl_gpri(struct bridge_softc *, void *); 339 static int bridge_ioctl_spri(struct bridge_softc *, void *); 340 static int bridge_ioctl_ght(struct bridge_softc *, void *); 341 static int bridge_ioctl_sht(struct bridge_softc *, void *); 342 static int bridge_ioctl_gfd(struct bridge_softc *, void *); 343 static int bridge_ioctl_sfd(struct bridge_softc *, void *); 344 static int bridge_ioctl_gma(struct bridge_softc *, void *); 345 static int bridge_ioctl_sma(struct bridge_softc *, void *); 346 static int bridge_ioctl_sifprio(struct bridge_softc *, void *); 347 static int bridge_ioctl_sifcost(struct bridge_softc *, void *); 348 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *); 349 static int bridge_ioctl_addspan(struct bridge_softc *, void *); 350 static int bridge_ioctl_delspan(struct bridge_softc *, void *); 351 static int bridge_ioctl_gbparam(struct bridge_softc *, void *); 352 static int bridge_ioctl_grte(struct bridge_softc *, void *); 353 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *); 354 static int bridge_ioctl_sproto(struct bridge_softc *, void *); 355 static int bridge_ioctl_stxhc(struct bridge_softc *, void *); 356 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *, 357 int); 358 static int bridge_ip_checkbasic(struct mbuf **mp); 359 #ifdef INET6 360 static int bridge_ip6_checkbasic(struct mbuf **mp); 361 #endif /* INET6 */ 362 static int bridge_fragment(struct ifnet *, struct mbuf **mp, 363 struct ether_header *, int, struct llc *); 364 static void bridge_linkstate(struct ifnet *ifp); 365 static void bridge_linkcheck(struct bridge_softc *sc); 366 367 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */ 368 #define VLANTAGOF(_m) \ 369 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1 370 371 static struct bstp_cb_ops bridge_ops = { 372 .bcb_state = bridge_state_change, 373 .bcb_rtage = bridge_rtable_expire 374 }; 375 376 SYSCTL_DECL(_net_link); 377 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, 378 "Bridge"); 379 380 /* only pass IP[46] packets when pfil is enabled */ 381 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1; 382 #define V_pfil_onlyip VNET(pfil_onlyip) 383 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, 384 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0, 385 "Only pass IP packets when pfil is enabled"); 386 387 /* run pfil hooks on the bridge interface */ 388 VNET_DEFINE_STATIC(int, pfil_bridge) = 1; 389 #define V_pfil_bridge VNET(pfil_bridge) 390 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, 391 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0, 392 "Packet filter on the bridge interface"); 393 394 /* layer2 filter with ipfw */ 395 VNET_DEFINE_STATIC(int, pfil_ipfw); 396 #define V_pfil_ipfw VNET(pfil_ipfw) 397 398 /* layer2 ARP filter with ipfw */ 399 VNET_DEFINE_STATIC(int, pfil_ipfw_arp); 400 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp) 401 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, 402 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0, 403 "Filter ARP packets through IPFW layer2"); 404 405 /* run pfil hooks on the member interface */ 406 VNET_DEFINE_STATIC(int, pfil_member) = 1; 407 #define V_pfil_member VNET(pfil_member) 408 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, 409 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0, 410 "Packet filter on the member interface"); 411 412 /* run pfil hooks on the physical interface for locally destined packets */ 413 VNET_DEFINE_STATIC(int, pfil_local_phys); 414 #define V_pfil_local_phys VNET(pfil_local_phys) 415 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, 416 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0, 417 "Packet filter on the physical interface for locally destined packets"); 418 419 /* log STP state changes */ 420 VNET_DEFINE_STATIC(int, log_stp); 421 #define V_log_stp VNET(log_stp) 422 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, 423 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0, 424 "Log STP state changes"); 425 426 /* share MAC with first bridge member */ 427 VNET_DEFINE_STATIC(int, bridge_inherit_mac); 428 #define V_bridge_inherit_mac VNET(bridge_inherit_mac) 429 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, 430 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0, 431 "Inherit MAC address from the first bridge member"); 432 433 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0; 434 #define V_allow_llz_overlap VNET(allow_llz_overlap) 435 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap, 436 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0, 437 "Allow overlap of link-local scope " 438 "zones of a bridge interface and the member interfaces"); 439 440 struct bridge_control { 441 int (*bc_func)(struct bridge_softc *, void *); 442 int bc_argsize; 443 int bc_flags; 444 }; 445 446 #define BC_F_COPYIN 0x01 /* copy arguments in */ 447 #define BC_F_COPYOUT 0x02 /* copy arguments out */ 448 #define BC_F_SUSER 0x04 /* do super-user check */ 449 450 const struct bridge_control bridge_control_table[] = { 451 { bridge_ioctl_add, sizeof(struct ifbreq), 452 BC_F_COPYIN|BC_F_SUSER }, 453 { bridge_ioctl_del, sizeof(struct ifbreq), 454 BC_F_COPYIN|BC_F_SUSER }, 455 456 { bridge_ioctl_gifflags, sizeof(struct ifbreq), 457 BC_F_COPYIN|BC_F_COPYOUT }, 458 { bridge_ioctl_sifflags, sizeof(struct ifbreq), 459 BC_F_COPYIN|BC_F_SUSER }, 460 461 { bridge_ioctl_scache, sizeof(struct ifbrparam), 462 BC_F_COPYIN|BC_F_SUSER }, 463 { bridge_ioctl_gcache, sizeof(struct ifbrparam), 464 BC_F_COPYOUT }, 465 466 { bridge_ioctl_gifs, sizeof(struct ifbifconf), 467 BC_F_COPYIN|BC_F_COPYOUT }, 468 { bridge_ioctl_rts, sizeof(struct ifbaconf), 469 BC_F_COPYIN|BC_F_COPYOUT }, 470 471 { bridge_ioctl_saddr, sizeof(struct ifbareq), 472 BC_F_COPYIN|BC_F_SUSER }, 473 474 { bridge_ioctl_sto, sizeof(struct ifbrparam), 475 BC_F_COPYIN|BC_F_SUSER }, 476 { bridge_ioctl_gto, sizeof(struct ifbrparam), 477 BC_F_COPYOUT }, 478 479 { bridge_ioctl_daddr, sizeof(struct ifbareq), 480 BC_F_COPYIN|BC_F_SUSER }, 481 482 { bridge_ioctl_flush, sizeof(struct ifbreq), 483 BC_F_COPYIN|BC_F_SUSER }, 484 485 { bridge_ioctl_gpri, sizeof(struct ifbrparam), 486 BC_F_COPYOUT }, 487 { bridge_ioctl_spri, sizeof(struct ifbrparam), 488 BC_F_COPYIN|BC_F_SUSER }, 489 490 { bridge_ioctl_ght, sizeof(struct ifbrparam), 491 BC_F_COPYOUT }, 492 { bridge_ioctl_sht, sizeof(struct ifbrparam), 493 BC_F_COPYIN|BC_F_SUSER }, 494 495 { bridge_ioctl_gfd, sizeof(struct ifbrparam), 496 BC_F_COPYOUT }, 497 { bridge_ioctl_sfd, sizeof(struct ifbrparam), 498 BC_F_COPYIN|BC_F_SUSER }, 499 500 { bridge_ioctl_gma, sizeof(struct ifbrparam), 501 BC_F_COPYOUT }, 502 { bridge_ioctl_sma, sizeof(struct ifbrparam), 503 BC_F_COPYIN|BC_F_SUSER }, 504 505 { bridge_ioctl_sifprio, sizeof(struct ifbreq), 506 BC_F_COPYIN|BC_F_SUSER }, 507 508 { bridge_ioctl_sifcost, sizeof(struct ifbreq), 509 BC_F_COPYIN|BC_F_SUSER }, 510 511 { bridge_ioctl_addspan, sizeof(struct ifbreq), 512 BC_F_COPYIN|BC_F_SUSER }, 513 { bridge_ioctl_delspan, sizeof(struct ifbreq), 514 BC_F_COPYIN|BC_F_SUSER }, 515 516 { bridge_ioctl_gbparam, sizeof(struct ifbropreq), 517 BC_F_COPYOUT }, 518 519 { bridge_ioctl_grte, sizeof(struct ifbrparam), 520 BC_F_COPYOUT }, 521 522 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf), 523 BC_F_COPYIN|BC_F_COPYOUT }, 524 525 { bridge_ioctl_sproto, sizeof(struct ifbrparam), 526 BC_F_COPYIN|BC_F_SUSER }, 527 528 { bridge_ioctl_stxhc, sizeof(struct ifbrparam), 529 BC_F_COPYIN|BC_F_SUSER }, 530 531 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq), 532 BC_F_COPYIN|BC_F_SUSER }, 533 534 }; 535 const int bridge_control_table_size = nitems(bridge_control_table); 536 537 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list); 538 #define V_bridge_list VNET(bridge_list) 539 #define BRIDGE_LIST_LOCK_INIT(x) mtx_init(&V_bridge_list_mtx, \ 540 "if_bridge list", NULL, MTX_DEF) 541 #define BRIDGE_LIST_LOCK_DESTROY(x) mtx_destroy(&V_bridge_list_mtx) 542 #define BRIDGE_LIST_LOCK(x) mtx_lock(&V_bridge_list_mtx) 543 #define BRIDGE_LIST_UNLOCK(x) mtx_unlock(&V_bridge_list_mtx) 544 545 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner); 546 #define V_bridge_cloner VNET(bridge_cloner) 547 548 static const char bridge_name[] = "bridge"; 549 550 static void 551 vnet_bridge_init(const void *unused __unused) 552 { 553 554 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode", 555 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL, 556 UMA_ALIGN_PTR, 0); 557 BRIDGE_LIST_LOCK_INIT(); 558 LIST_INIT(&V_bridge_list); 559 V_bridge_cloner = if_clone_simple(bridge_name, 560 bridge_clone_create, bridge_clone_destroy, 0); 561 } 562 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY, 563 vnet_bridge_init, NULL); 564 565 static void 566 vnet_bridge_uninit(const void *unused __unused) 567 { 568 569 if_clone_detach(V_bridge_cloner); 570 V_bridge_cloner = NULL; 571 BRIDGE_LIST_LOCK_DESTROY(); 572 573 /* Callbacks may use the UMA zone. */ 574 epoch_drain_callbacks(net_epoch_preempt); 575 576 uma_zdestroy(V_bridge_rtnode_zone); 577 } 578 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY, 579 vnet_bridge_uninit, NULL); 580 581 static int 582 bridge_modevent(module_t mod, int type, void *data) 583 { 584 585 switch (type) { 586 case MOD_LOAD: 587 bridge_dn_p = bridge_dummynet; 588 bridge_detach_cookie = EVENTHANDLER_REGISTER( 589 ifnet_departure_event, bridge_ifdetach, NULL, 590 EVENTHANDLER_PRI_ANY); 591 break; 592 case MOD_UNLOAD: 593 EVENTHANDLER_DEREGISTER(ifnet_departure_event, 594 bridge_detach_cookie); 595 bridge_dn_p = NULL; 596 break; 597 default: 598 return (EOPNOTSUPP); 599 } 600 return (0); 601 } 602 603 static moduledata_t bridge_mod = { 604 "if_bridge", 605 bridge_modevent, 606 0 607 }; 608 609 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY); 610 MODULE_VERSION(if_bridge, 1); 611 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1); 612 613 /* 614 * handler for net.link.bridge.ipfw 615 */ 616 static int 617 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS) 618 { 619 int enable = V_pfil_ipfw; 620 int error; 621 622 error = sysctl_handle_int(oidp, &enable, 0, req); 623 enable &= 1; 624 625 if (enable != V_pfil_ipfw) { 626 V_pfil_ipfw = enable; 627 628 /* 629 * Disable pfil so that ipfw doesnt run twice, if the user 630 * really wants both then they can re-enable pfil_bridge and/or 631 * pfil_member. Also allow non-ip packets as ipfw can filter by 632 * layer2 type. 633 */ 634 if (V_pfil_ipfw) { 635 V_pfil_onlyip = 0; 636 V_pfil_bridge = 0; 637 V_pfil_member = 0; 638 } 639 } 640 641 return (error); 642 } 643 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, 644 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT, 645 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I", 646 "Layer2 filter with IPFW"); 647 648 /* 649 * bridge_clone_create: 650 * 651 * Create a new bridge instance. 652 */ 653 static int 654 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params) 655 { 656 struct bridge_softc *sc; 657 struct ifnet *ifp; 658 659 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO); 660 ifp = sc->sc_ifp = if_alloc(IFT_ETHER); 661 if (ifp == NULL) { 662 free(sc, M_DEVBUF); 663 return (ENOSPC); 664 } 665 666 BRIDGE_LOCK_INIT(sc); 667 sc->sc_brtmax = BRIDGE_RTABLE_MAX; 668 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT; 669 670 /* Initialize our routing table. */ 671 bridge_rtable_init(sc); 672 673 callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0); 674 675 CK_LIST_INIT(&sc->sc_iflist); 676 CK_LIST_INIT(&sc->sc_spanlist); 677 678 ifp->if_softc = sc; 679 if_initname(ifp, bridge_name, unit); 680 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; 681 ifp->if_ioctl = bridge_ioctl; 682 ifp->if_transmit = bridge_transmit; 683 ifp->if_qflush = bridge_qflush; 684 ifp->if_init = bridge_init; 685 ifp->if_type = IFT_BRIDGE; 686 687 ether_gen_addr(ifp, &sc->sc_defaddr); 688 689 bstp_attach(&sc->sc_stp, &bridge_ops); 690 ether_ifattach(ifp, sc->sc_defaddr.octet); 691 /* Now undo some of the damage... */ 692 ifp->if_baudrate = 0; 693 ifp->if_type = IFT_BRIDGE; 694 695 BRIDGE_LIST_LOCK(); 696 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list); 697 BRIDGE_LIST_UNLOCK(); 698 699 return (0); 700 } 701 702 static void 703 bridge_clone_destroy_cb(struct epoch_context *ctx) 704 { 705 struct bridge_softc *sc; 706 707 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx); 708 709 BRIDGE_LOCK_DESTROY(sc); 710 free(sc, M_DEVBUF); 711 } 712 713 /* 714 * bridge_clone_destroy: 715 * 716 * Destroy a bridge instance. 717 */ 718 static void 719 bridge_clone_destroy(struct ifnet *ifp) 720 { 721 struct bridge_softc *sc = ifp->if_softc; 722 struct bridge_iflist *bif; 723 struct epoch_tracker et; 724 725 NET_EPOCH_ENTER(et); 726 BRIDGE_LOCK(sc); 727 728 bridge_stop(ifp, 1); 729 ifp->if_flags &= ~IFF_UP; 730 731 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL) 732 bridge_delete_member(sc, bif, 0); 733 734 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) { 735 bridge_delete_span(sc, bif); 736 } 737 738 /* Tear down the routing table. */ 739 bridge_rtable_fini(sc); 740 741 BRIDGE_UNLOCK(sc); 742 743 callout_drain(&sc->sc_brcallout); 744 745 BRIDGE_LIST_LOCK(); 746 LIST_REMOVE(sc, sc_list); 747 BRIDGE_LIST_UNLOCK(); 748 749 bstp_detach(&sc->sc_stp); 750 NET_EPOCH_EXIT(et); 751 752 ether_ifdetach(ifp); 753 if_free(ifp); 754 755 NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx); 756 } 757 758 /* 759 * bridge_ioctl: 760 * 761 * Handle a control request from the operator. 762 */ 763 static int 764 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data) 765 { 766 struct bridge_softc *sc = ifp->if_softc; 767 struct ifreq *ifr = (struct ifreq *)data; 768 struct bridge_iflist *bif; 769 struct thread *td = curthread; 770 union { 771 struct ifbreq ifbreq; 772 struct ifbifconf ifbifconf; 773 struct ifbareq ifbareq; 774 struct ifbaconf ifbaconf; 775 struct ifbrparam ifbrparam; 776 struct ifbropreq ifbropreq; 777 } args; 778 struct ifdrv *ifd = (struct ifdrv *) data; 779 const struct bridge_control *bc; 780 int error = 0, oldmtu; 781 struct epoch_tracker et; 782 783 NET_EPOCH_ENTER(et); 784 785 switch (cmd) { 786 case SIOCADDMULTI: 787 case SIOCDELMULTI: 788 break; 789 790 case SIOCGDRVSPEC: 791 case SIOCSDRVSPEC: 792 if (ifd->ifd_cmd >= bridge_control_table_size) { 793 error = EINVAL; 794 break; 795 } 796 bc = &bridge_control_table[ifd->ifd_cmd]; 797 798 if (cmd == SIOCGDRVSPEC && 799 (bc->bc_flags & BC_F_COPYOUT) == 0) { 800 error = EINVAL; 801 break; 802 } 803 else if (cmd == SIOCSDRVSPEC && 804 (bc->bc_flags & BC_F_COPYOUT) != 0) { 805 error = EINVAL; 806 break; 807 } 808 809 if (bc->bc_flags & BC_F_SUSER) { 810 error = priv_check(td, PRIV_NET_BRIDGE); 811 if (error) 812 break; 813 } 814 815 if (ifd->ifd_len != bc->bc_argsize || 816 ifd->ifd_len > sizeof(args)) { 817 error = EINVAL; 818 break; 819 } 820 821 bzero(&args, sizeof(args)); 822 if (bc->bc_flags & BC_F_COPYIN) { 823 error = copyin(ifd->ifd_data, &args, ifd->ifd_len); 824 if (error) 825 break; 826 } 827 828 oldmtu = ifp->if_mtu; 829 BRIDGE_LOCK(sc); 830 error = (*bc->bc_func)(sc, &args); 831 BRIDGE_UNLOCK(sc); 832 if (error) 833 break; 834 835 /* 836 * Bridge MTU may change during addition of the first port. 837 * If it did, do network layer specific procedure. 838 */ 839 if (ifp->if_mtu != oldmtu) { 840 #ifdef INET6 841 nd6_setmtu(ifp); 842 #endif 843 rt_updatemtu(ifp); 844 } 845 846 if (bc->bc_flags & BC_F_COPYOUT) 847 error = copyout(&args, ifd->ifd_data, ifd->ifd_len); 848 849 break; 850 851 case SIOCSIFFLAGS: 852 if (!(ifp->if_flags & IFF_UP) && 853 (ifp->if_drv_flags & IFF_DRV_RUNNING)) { 854 /* 855 * If interface is marked down and it is running, 856 * then stop and disable it. 857 */ 858 BRIDGE_LOCK(sc); 859 bridge_stop(ifp, 1); 860 BRIDGE_UNLOCK(sc); 861 } else if ((ifp->if_flags & IFF_UP) && 862 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) { 863 /* 864 * If interface is marked up and it is stopped, then 865 * start it. 866 */ 867 (*ifp->if_init)(sc); 868 } 869 break; 870 871 case SIOCSIFMTU: 872 if (ifr->ifr_mtu < 576) { 873 error = EINVAL; 874 break; 875 } 876 if (CK_LIST_EMPTY(&sc->sc_iflist)) { 877 sc->sc_ifp->if_mtu = ifr->ifr_mtu; 878 break; 879 } 880 BRIDGE_LOCK(sc); 881 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 882 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) { 883 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)" 884 " != %d\n", sc->sc_ifp->if_xname, 885 bif->bif_ifp->if_mtu, 886 bif->bif_ifp->if_xname, ifr->ifr_mtu); 887 error = EINVAL; 888 break; 889 } 890 } 891 if (!error) 892 sc->sc_ifp->if_mtu = ifr->ifr_mtu; 893 BRIDGE_UNLOCK(sc); 894 break; 895 default: 896 /* 897 * drop the lock as ether_ioctl() will call bridge_start() and 898 * cause the lock to be recursed. 899 */ 900 error = ether_ioctl(ifp, cmd, data); 901 break; 902 } 903 904 NET_EPOCH_EXIT(et); 905 906 return (error); 907 } 908 909 /* 910 * bridge_mutecaps: 911 * 912 * Clear or restore unwanted capabilities on the member interface 913 */ 914 static void 915 bridge_mutecaps(struct bridge_softc *sc) 916 { 917 struct bridge_iflist *bif; 918 int enabled, mask; 919 920 BRIDGE_LOCK_ASSERT(sc); 921 922 /* Initial bitmask of capabilities to test */ 923 mask = BRIDGE_IFCAPS_MASK; 924 925 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 926 /* Every member must support it or its disabled */ 927 mask &= bif->bif_savedcaps; 928 } 929 930 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 931 enabled = bif->bif_ifp->if_capenable; 932 enabled &= ~BRIDGE_IFCAPS_STRIP; 933 /* strip off mask bits and enable them again if allowed */ 934 enabled &= ~BRIDGE_IFCAPS_MASK; 935 enabled |= mask; 936 BRIDGE_UNLOCK(sc); 937 bridge_set_ifcap(sc, bif, enabled); 938 BRIDGE_LOCK(sc); 939 } 940 } 941 942 static void 943 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set) 944 { 945 struct ifnet *ifp = bif->bif_ifp; 946 struct ifreq ifr; 947 int error, mask, stuck; 948 949 BRIDGE_UNLOCK_ASSERT(sc); 950 951 bzero(&ifr, sizeof(ifr)); 952 ifr.ifr_reqcap = set; 953 954 if (ifp->if_capenable != set) { 955 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr); 956 if (error) 957 if_printf(sc->sc_ifp, 958 "error setting capabilities on %s: %d\n", 959 ifp->if_xname, error); 960 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP; 961 stuck = ifp->if_capenable & mask & ~set; 962 if (stuck != 0) 963 if_printf(sc->sc_ifp, 964 "can't disable some capabilities on %s: 0x%x\n", 965 ifp->if_xname, stuck); 966 } 967 } 968 969 /* 970 * bridge_lookup_member: 971 * 972 * Lookup a bridge member interface. 973 */ 974 static struct bridge_iflist * 975 bridge_lookup_member(struct bridge_softc *sc, const char *name) 976 { 977 struct bridge_iflist *bif; 978 struct ifnet *ifp; 979 980 NET_EPOCH_ASSERT(); 981 982 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 983 ifp = bif->bif_ifp; 984 if (strcmp(ifp->if_xname, name) == 0) 985 return (bif); 986 } 987 988 return (NULL); 989 } 990 991 /* 992 * bridge_lookup_member_if: 993 * 994 * Lookup a bridge member interface by ifnet*. 995 */ 996 static struct bridge_iflist * 997 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp) 998 { 999 struct bridge_iflist *bif; 1000 1001 NET_EPOCH_ASSERT(); 1002 1003 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1004 if (bif->bif_ifp == member_ifp) 1005 return (bif); 1006 } 1007 1008 return (NULL); 1009 } 1010 1011 static void 1012 bridge_delete_member_cb(struct epoch_context *ctx) 1013 { 1014 struct bridge_iflist *bif; 1015 1016 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx); 1017 1018 free(bif, M_DEVBUF); 1019 } 1020 1021 /* 1022 * bridge_delete_member: 1023 * 1024 * Delete the specified member interface. 1025 */ 1026 static void 1027 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif, 1028 int gone) 1029 { 1030 struct ifnet *ifs = bif->bif_ifp; 1031 struct ifnet *fif = NULL; 1032 struct bridge_iflist *bifl; 1033 1034 BRIDGE_LOCK_ASSERT(sc); 1035 1036 if (bif->bif_flags & IFBIF_STP) 1037 bstp_disable(&bif->bif_stp); 1038 1039 ifs->if_bridge = NULL; 1040 CK_LIST_REMOVE(bif, bif_next); 1041 1042 /* 1043 * If removing the interface that gave the bridge its mac address, set 1044 * the mac address of the bridge to the address of the next member, or 1045 * to its default address if no members are left. 1046 */ 1047 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) { 1048 if (CK_LIST_EMPTY(&sc->sc_iflist)) { 1049 bcopy(&sc->sc_defaddr, 1050 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1051 sc->sc_ifaddr = NULL; 1052 } else { 1053 bifl = CK_LIST_FIRST(&sc->sc_iflist); 1054 fif = bifl->bif_ifp; 1055 bcopy(IF_LLADDR(fif), 1056 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1057 sc->sc_ifaddr = fif; 1058 } 1059 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp); 1060 } 1061 1062 bridge_linkcheck(sc); 1063 bridge_mutecaps(sc); /* recalcuate now this interface is removed */ 1064 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL); 1065 KASSERT(bif->bif_addrcnt == 0, 1066 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt)); 1067 1068 ifs->if_bridge_output = NULL; 1069 ifs->if_bridge_input = NULL; 1070 ifs->if_bridge_linkstate = NULL; 1071 BRIDGE_UNLOCK(sc); 1072 if (!gone) { 1073 switch (ifs->if_type) { 1074 case IFT_ETHER: 1075 case IFT_L2VLAN: 1076 /* 1077 * Take the interface out of promiscuous mode, but only 1078 * if it was promiscuous in the first place. It might 1079 * not be if we're in the bridge_ioctl_add() error path. 1080 */ 1081 if (ifs->if_flags & IFF_PROMISC) 1082 (void) ifpromisc(ifs, 0); 1083 break; 1084 1085 case IFT_GIF: 1086 break; 1087 1088 default: 1089 #ifdef DIAGNOSTIC 1090 panic("bridge_delete_member: impossible"); 1091 #endif 1092 break; 1093 } 1094 /* reneable any interface capabilities */ 1095 bridge_set_ifcap(sc, bif, bif->bif_savedcaps); 1096 } 1097 bstp_destroy(&bif->bif_stp); /* prepare to free */ 1098 BRIDGE_LOCK(sc); 1099 1100 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx); 1101 } 1102 1103 /* 1104 * bridge_delete_span: 1105 * 1106 * Delete the specified span interface. 1107 */ 1108 static void 1109 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif) 1110 { 1111 BRIDGE_LOCK_ASSERT(sc); 1112 1113 KASSERT(bif->bif_ifp->if_bridge == NULL, 1114 ("%s: not a span interface", __func__)); 1115 1116 CK_LIST_REMOVE(bif, bif_next); 1117 1118 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx); 1119 } 1120 1121 static int 1122 bridge_ioctl_add(struct bridge_softc *sc, void *arg) 1123 { 1124 struct ifbreq *req = arg; 1125 struct bridge_iflist *bif = NULL; 1126 struct ifnet *ifs; 1127 int error = 0; 1128 1129 ifs = ifunit(req->ifbr_ifsname); 1130 if (ifs == NULL) 1131 return (ENOENT); 1132 if (ifs->if_ioctl == NULL) /* must be supported */ 1133 return (EINVAL); 1134 1135 /* If it's in the span list, it can't be a member. */ 1136 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1137 if (ifs == bif->bif_ifp) 1138 return (EBUSY); 1139 1140 if (ifs->if_bridge == sc) 1141 return (EEXIST); 1142 1143 if (ifs->if_bridge != NULL) 1144 return (EBUSY); 1145 1146 switch (ifs->if_type) { 1147 case IFT_ETHER: 1148 case IFT_L2VLAN: 1149 case IFT_GIF: 1150 /* permitted interface types */ 1151 break; 1152 default: 1153 return (EINVAL); 1154 } 1155 1156 #ifdef INET6 1157 /* 1158 * Two valid inet6 addresses with link-local scope must not be 1159 * on the parent interface and the member interfaces at the 1160 * same time. This restriction is needed to prevent violation 1161 * of link-local scope zone. Attempts to add a member 1162 * interface which has inet6 addresses when the parent has 1163 * inet6 triggers removal of all inet6 addresses on the member 1164 * interface. 1165 */ 1166 1167 /* Check if the parent interface has a link-local scope addr. */ 1168 if (V_allow_llz_overlap == 0 && 1169 in6ifa_llaonifp(sc->sc_ifp) != NULL) { 1170 /* 1171 * If any, remove all inet6 addresses from the member 1172 * interfaces. 1173 */ 1174 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1175 if (in6ifa_llaonifp(bif->bif_ifp)) { 1176 BRIDGE_UNLOCK(sc); 1177 in6_ifdetach(bif->bif_ifp); 1178 BRIDGE_LOCK(sc); 1179 if_printf(sc->sc_ifp, 1180 "IPv6 addresses on %s have been removed " 1181 "before adding it as a member to prevent " 1182 "IPv6 address scope violation.\n", 1183 bif->bif_ifp->if_xname); 1184 } 1185 } 1186 if (in6ifa_llaonifp(ifs)) { 1187 BRIDGE_UNLOCK(sc); 1188 in6_ifdetach(ifs); 1189 BRIDGE_LOCK(sc); 1190 if_printf(sc->sc_ifp, 1191 "IPv6 addresses on %s have been removed " 1192 "before adding it as a member to prevent " 1193 "IPv6 address scope violation.\n", 1194 ifs->if_xname); 1195 } 1196 } 1197 #endif 1198 /* Allow the first Ethernet member to define the MTU */ 1199 if (CK_LIST_EMPTY(&sc->sc_iflist)) 1200 sc->sc_ifp->if_mtu = ifs->if_mtu; 1201 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) { 1202 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n", 1203 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu); 1204 return (EINVAL); 1205 } 1206 1207 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO); 1208 if (bif == NULL) 1209 return (ENOMEM); 1210 1211 bif->bif_ifp = ifs; 1212 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER; 1213 bif->bif_savedcaps = ifs->if_capenable; 1214 1215 /* 1216 * Assign the interface's MAC address to the bridge if it's the first 1217 * member and the MAC address of the bridge has not been changed from 1218 * the default randomly generated one. 1219 */ 1220 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) && 1221 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) { 1222 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1223 sc->sc_ifaddr = ifs; 1224 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp); 1225 } 1226 1227 ifs->if_bridge = sc; 1228 ifs->if_bridge_output = bridge_output; 1229 ifs->if_bridge_input = bridge_input; 1230 ifs->if_bridge_linkstate = bridge_linkstate; 1231 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp); 1232 /* 1233 * XXX: XLOCK HERE!?! 1234 * 1235 * NOTE: insert_***HEAD*** should be safe for the traversals. 1236 */ 1237 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next); 1238 1239 /* Set interface capabilities to the intersection set of all members */ 1240 bridge_mutecaps(sc); 1241 bridge_linkcheck(sc); 1242 1243 /* Place the interface into promiscuous mode */ 1244 switch (ifs->if_type) { 1245 case IFT_ETHER: 1246 case IFT_L2VLAN: 1247 BRIDGE_UNLOCK(sc); 1248 error = ifpromisc(ifs, 1); 1249 BRIDGE_LOCK(sc); 1250 break; 1251 } 1252 1253 if (error) 1254 bridge_delete_member(sc, bif, 0); 1255 return (error); 1256 } 1257 1258 static int 1259 bridge_ioctl_del(struct bridge_softc *sc, void *arg) 1260 { 1261 struct ifbreq *req = arg; 1262 struct bridge_iflist *bif; 1263 1264 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1265 if (bif == NULL) 1266 return (ENOENT); 1267 1268 bridge_delete_member(sc, bif, 0); 1269 1270 return (0); 1271 } 1272 1273 static int 1274 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg) 1275 { 1276 struct ifbreq *req = arg; 1277 struct bridge_iflist *bif; 1278 struct bstp_port *bp; 1279 1280 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1281 if (bif == NULL) 1282 return (ENOENT); 1283 1284 bp = &bif->bif_stp; 1285 req->ifbr_ifsflags = bif->bif_flags; 1286 req->ifbr_state = bp->bp_state; 1287 req->ifbr_priority = bp->bp_priority; 1288 req->ifbr_path_cost = bp->bp_path_cost; 1289 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff; 1290 req->ifbr_proto = bp->bp_protover; 1291 req->ifbr_role = bp->bp_role; 1292 req->ifbr_stpflags = bp->bp_flags; 1293 req->ifbr_addrcnt = bif->bif_addrcnt; 1294 req->ifbr_addrmax = bif->bif_addrmax; 1295 req->ifbr_addrexceeded = bif->bif_addrexceeded; 1296 1297 /* Copy STP state options as flags */ 1298 if (bp->bp_operedge) 1299 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE; 1300 if (bp->bp_flags & BSTP_PORT_AUTOEDGE) 1301 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE; 1302 if (bp->bp_ptp_link) 1303 req->ifbr_ifsflags |= IFBIF_BSTP_PTP; 1304 if (bp->bp_flags & BSTP_PORT_AUTOPTP) 1305 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP; 1306 if (bp->bp_flags & BSTP_PORT_ADMEDGE) 1307 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE; 1308 if (bp->bp_flags & BSTP_PORT_ADMCOST) 1309 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST; 1310 return (0); 1311 } 1312 1313 static int 1314 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg) 1315 { 1316 struct ifbreq *req = arg; 1317 struct bridge_iflist *bif; 1318 struct bstp_port *bp; 1319 int error; 1320 1321 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1322 if (bif == NULL) 1323 return (ENOENT); 1324 bp = &bif->bif_stp; 1325 1326 if (req->ifbr_ifsflags & IFBIF_SPAN) 1327 /* SPAN is readonly */ 1328 return (EINVAL); 1329 1330 if (req->ifbr_ifsflags & IFBIF_STP) { 1331 if ((bif->bif_flags & IFBIF_STP) == 0) { 1332 error = bstp_enable(&bif->bif_stp); 1333 if (error) 1334 return (error); 1335 } 1336 } else { 1337 if ((bif->bif_flags & IFBIF_STP) != 0) 1338 bstp_disable(&bif->bif_stp); 1339 } 1340 1341 /* Pass on STP flags */ 1342 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0); 1343 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0); 1344 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0); 1345 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0); 1346 1347 /* Save the bits relating to the bridge */ 1348 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK; 1349 1350 return (0); 1351 } 1352 1353 static int 1354 bridge_ioctl_scache(struct bridge_softc *sc, void *arg) 1355 { 1356 struct ifbrparam *param = arg; 1357 1358 sc->sc_brtmax = param->ifbrp_csize; 1359 bridge_rttrim(sc); 1360 1361 return (0); 1362 } 1363 1364 static int 1365 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg) 1366 { 1367 struct ifbrparam *param = arg; 1368 1369 param->ifbrp_csize = sc->sc_brtmax; 1370 1371 return (0); 1372 } 1373 1374 static int 1375 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg) 1376 { 1377 struct ifbifconf *bifc = arg; 1378 struct bridge_iflist *bif; 1379 struct ifbreq breq; 1380 char *buf, *outbuf; 1381 int count, buflen, len, error = 0; 1382 1383 count = 0; 1384 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) 1385 count++; 1386 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1387 count++; 1388 1389 buflen = sizeof(breq) * count; 1390 if (bifc->ifbic_len == 0) { 1391 bifc->ifbic_len = buflen; 1392 return (0); 1393 } 1394 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO); 1395 if (outbuf == NULL) 1396 return (ENOMEM); 1397 1398 count = 0; 1399 buf = outbuf; 1400 len = min(bifc->ifbic_len, buflen); 1401 bzero(&breq, sizeof(breq)); 1402 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1403 if (len < sizeof(breq)) 1404 break; 1405 1406 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname, 1407 sizeof(breq.ifbr_ifsname)); 1408 /* Fill in the ifbreq structure */ 1409 error = bridge_ioctl_gifflags(sc, &breq); 1410 if (error) 1411 break; 1412 memcpy(buf, &breq, sizeof(breq)); 1413 count++; 1414 buf += sizeof(breq); 1415 len -= sizeof(breq); 1416 } 1417 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) { 1418 if (len < sizeof(breq)) 1419 break; 1420 1421 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname, 1422 sizeof(breq.ifbr_ifsname)); 1423 breq.ifbr_ifsflags = bif->bif_flags; 1424 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff; 1425 memcpy(buf, &breq, sizeof(breq)); 1426 count++; 1427 buf += sizeof(breq); 1428 len -= sizeof(breq); 1429 } 1430 1431 BRIDGE_UNLOCK(sc); 1432 bifc->ifbic_len = sizeof(breq) * count; 1433 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len); 1434 BRIDGE_LOCK(sc); 1435 free(outbuf, M_TEMP); 1436 return (error); 1437 } 1438 1439 static int 1440 bridge_ioctl_rts(struct bridge_softc *sc, void *arg) 1441 { 1442 struct ifbaconf *bac = arg; 1443 struct bridge_rtnode *brt; 1444 struct ifbareq bareq; 1445 char *buf, *outbuf; 1446 int count, buflen, len, error = 0; 1447 1448 if (bac->ifbac_len == 0) 1449 return (0); 1450 1451 count = 0; 1452 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) 1453 count++; 1454 buflen = sizeof(bareq) * count; 1455 1456 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO); 1457 if (outbuf == NULL) 1458 return (ENOMEM); 1459 1460 count = 0; 1461 buf = outbuf; 1462 len = min(bac->ifbac_len, buflen); 1463 bzero(&bareq, sizeof(bareq)); 1464 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) { 1465 if (len < sizeof(bareq)) 1466 goto out; 1467 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname, 1468 sizeof(bareq.ifba_ifsname)); 1469 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr)); 1470 bareq.ifba_vlan = brt->brt_vlan; 1471 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 1472 time_uptime < brt->brt_expire) 1473 bareq.ifba_expire = brt->brt_expire - time_uptime; 1474 else 1475 bareq.ifba_expire = 0; 1476 bareq.ifba_flags = brt->brt_flags; 1477 1478 memcpy(buf, &bareq, sizeof(bareq)); 1479 count++; 1480 buf += sizeof(bareq); 1481 len -= sizeof(bareq); 1482 } 1483 out: 1484 BRIDGE_UNLOCK(sc); 1485 bac->ifbac_len = sizeof(bareq) * count; 1486 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len); 1487 BRIDGE_LOCK(sc); 1488 free(outbuf, M_TEMP); 1489 return (error); 1490 } 1491 1492 static int 1493 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg) 1494 { 1495 struct ifbareq *req = arg; 1496 struct bridge_iflist *bif; 1497 int error; 1498 1499 NET_EPOCH_ASSERT(); 1500 1501 bif = bridge_lookup_member(sc, req->ifba_ifsname); 1502 if (bif == NULL) 1503 return (ENOENT); 1504 1505 /* bridge_rtupdate() may acquire the lock. */ 1506 BRIDGE_UNLOCK(sc); 1507 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1, 1508 req->ifba_flags); 1509 BRIDGE_LOCK(sc); 1510 1511 return (error); 1512 } 1513 1514 static int 1515 bridge_ioctl_sto(struct bridge_softc *sc, void *arg) 1516 { 1517 struct ifbrparam *param = arg; 1518 1519 sc->sc_brttimeout = param->ifbrp_ctime; 1520 return (0); 1521 } 1522 1523 static int 1524 bridge_ioctl_gto(struct bridge_softc *sc, void *arg) 1525 { 1526 struct ifbrparam *param = arg; 1527 1528 param->ifbrp_ctime = sc->sc_brttimeout; 1529 return (0); 1530 } 1531 1532 static int 1533 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg) 1534 { 1535 struct ifbareq *req = arg; 1536 1537 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan)); 1538 } 1539 1540 static int 1541 bridge_ioctl_flush(struct bridge_softc *sc, void *arg) 1542 { 1543 struct ifbreq *req = arg; 1544 1545 bridge_rtflush(sc, req->ifbr_ifsflags); 1546 return (0); 1547 } 1548 1549 static int 1550 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg) 1551 { 1552 struct ifbrparam *param = arg; 1553 struct bstp_state *bs = &sc->sc_stp; 1554 1555 param->ifbrp_prio = bs->bs_bridge_priority; 1556 return (0); 1557 } 1558 1559 static int 1560 bridge_ioctl_spri(struct bridge_softc *sc, void *arg) 1561 { 1562 struct ifbrparam *param = arg; 1563 1564 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio)); 1565 } 1566 1567 static int 1568 bridge_ioctl_ght(struct bridge_softc *sc, void *arg) 1569 { 1570 struct ifbrparam *param = arg; 1571 struct bstp_state *bs = &sc->sc_stp; 1572 1573 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8; 1574 return (0); 1575 } 1576 1577 static int 1578 bridge_ioctl_sht(struct bridge_softc *sc, void *arg) 1579 { 1580 struct ifbrparam *param = arg; 1581 1582 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime)); 1583 } 1584 1585 static int 1586 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg) 1587 { 1588 struct ifbrparam *param = arg; 1589 struct bstp_state *bs = &sc->sc_stp; 1590 1591 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8; 1592 return (0); 1593 } 1594 1595 static int 1596 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg) 1597 { 1598 struct ifbrparam *param = arg; 1599 1600 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay)); 1601 } 1602 1603 static int 1604 bridge_ioctl_gma(struct bridge_softc *sc, void *arg) 1605 { 1606 struct ifbrparam *param = arg; 1607 struct bstp_state *bs = &sc->sc_stp; 1608 1609 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8; 1610 return (0); 1611 } 1612 1613 static int 1614 bridge_ioctl_sma(struct bridge_softc *sc, void *arg) 1615 { 1616 struct ifbrparam *param = arg; 1617 1618 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage)); 1619 } 1620 1621 static int 1622 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg) 1623 { 1624 struct ifbreq *req = arg; 1625 struct bridge_iflist *bif; 1626 1627 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1628 if (bif == NULL) 1629 return (ENOENT); 1630 1631 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority)); 1632 } 1633 1634 static int 1635 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg) 1636 { 1637 struct ifbreq *req = arg; 1638 struct bridge_iflist *bif; 1639 1640 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1641 if (bif == NULL) 1642 return (ENOENT); 1643 1644 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost)); 1645 } 1646 1647 static int 1648 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg) 1649 { 1650 struct ifbreq *req = arg; 1651 struct bridge_iflist *bif; 1652 1653 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1654 if (bif == NULL) 1655 return (ENOENT); 1656 1657 bif->bif_addrmax = req->ifbr_addrmax; 1658 return (0); 1659 } 1660 1661 static int 1662 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg) 1663 { 1664 struct ifbreq *req = arg; 1665 struct bridge_iflist *bif = NULL; 1666 struct ifnet *ifs; 1667 1668 ifs = ifunit(req->ifbr_ifsname); 1669 if (ifs == NULL) 1670 return (ENOENT); 1671 1672 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1673 if (ifs == bif->bif_ifp) 1674 return (EBUSY); 1675 1676 if (ifs->if_bridge != NULL) 1677 return (EBUSY); 1678 1679 switch (ifs->if_type) { 1680 case IFT_ETHER: 1681 case IFT_GIF: 1682 case IFT_L2VLAN: 1683 break; 1684 default: 1685 return (EINVAL); 1686 } 1687 1688 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO); 1689 if (bif == NULL) 1690 return (ENOMEM); 1691 1692 bif->bif_ifp = ifs; 1693 bif->bif_flags = IFBIF_SPAN; 1694 1695 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next); 1696 1697 return (0); 1698 } 1699 1700 static int 1701 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg) 1702 { 1703 struct ifbreq *req = arg; 1704 struct bridge_iflist *bif; 1705 struct ifnet *ifs; 1706 1707 ifs = ifunit(req->ifbr_ifsname); 1708 if (ifs == NULL) 1709 return (ENOENT); 1710 1711 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1712 if (ifs == bif->bif_ifp) 1713 break; 1714 1715 if (bif == NULL) 1716 return (ENOENT); 1717 1718 bridge_delete_span(sc, bif); 1719 1720 return (0); 1721 } 1722 1723 static int 1724 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg) 1725 { 1726 struct ifbropreq *req = arg; 1727 struct bstp_state *bs = &sc->sc_stp; 1728 struct bstp_port *root_port; 1729 1730 req->ifbop_maxage = bs->bs_bridge_max_age >> 8; 1731 req->ifbop_hellotime = bs->bs_bridge_htime >> 8; 1732 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8; 1733 1734 root_port = bs->bs_root_port; 1735 if (root_port == NULL) 1736 req->ifbop_root_port = 0; 1737 else 1738 req->ifbop_root_port = root_port->bp_ifp->if_index; 1739 1740 req->ifbop_holdcount = bs->bs_txholdcount; 1741 req->ifbop_priority = bs->bs_bridge_priority; 1742 req->ifbop_protocol = bs->bs_protover; 1743 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost; 1744 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id; 1745 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id; 1746 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id; 1747 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec; 1748 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec; 1749 1750 return (0); 1751 } 1752 1753 static int 1754 bridge_ioctl_grte(struct bridge_softc *sc, void *arg) 1755 { 1756 struct ifbrparam *param = arg; 1757 1758 param->ifbrp_cexceeded = sc->sc_brtexceeded; 1759 return (0); 1760 } 1761 1762 static int 1763 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg) 1764 { 1765 struct ifbpstpconf *bifstp = arg; 1766 struct bridge_iflist *bif; 1767 struct bstp_port *bp; 1768 struct ifbpstpreq bpreq; 1769 char *buf, *outbuf; 1770 int count, buflen, len, error = 0; 1771 1772 count = 0; 1773 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1774 if ((bif->bif_flags & IFBIF_STP) != 0) 1775 count++; 1776 } 1777 1778 buflen = sizeof(bpreq) * count; 1779 if (bifstp->ifbpstp_len == 0) { 1780 bifstp->ifbpstp_len = buflen; 1781 return (0); 1782 } 1783 1784 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO); 1785 if (outbuf == NULL) 1786 return (ENOMEM); 1787 1788 count = 0; 1789 buf = outbuf; 1790 len = min(bifstp->ifbpstp_len, buflen); 1791 bzero(&bpreq, sizeof(bpreq)); 1792 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1793 if (len < sizeof(bpreq)) 1794 break; 1795 1796 if ((bif->bif_flags & IFBIF_STP) == 0) 1797 continue; 1798 1799 bp = &bif->bif_stp; 1800 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff; 1801 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions; 1802 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost; 1803 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id; 1804 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id; 1805 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id; 1806 1807 memcpy(buf, &bpreq, sizeof(bpreq)); 1808 count++; 1809 buf += sizeof(bpreq); 1810 len -= sizeof(bpreq); 1811 } 1812 1813 BRIDGE_UNLOCK(sc); 1814 bifstp->ifbpstp_len = sizeof(bpreq) * count; 1815 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len); 1816 BRIDGE_LOCK(sc); 1817 free(outbuf, M_TEMP); 1818 return (error); 1819 } 1820 1821 static int 1822 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg) 1823 { 1824 struct ifbrparam *param = arg; 1825 1826 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto)); 1827 } 1828 1829 static int 1830 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg) 1831 { 1832 struct ifbrparam *param = arg; 1833 1834 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc)); 1835 } 1836 1837 /* 1838 * bridge_ifdetach: 1839 * 1840 * Detach an interface from a bridge. Called when a member 1841 * interface is detaching. 1842 */ 1843 static void 1844 bridge_ifdetach(void *arg __unused, struct ifnet *ifp) 1845 { 1846 struct bridge_softc *sc = ifp->if_bridge; 1847 struct bridge_iflist *bif; 1848 struct epoch_tracker et; 1849 1850 if (ifp->if_flags & IFF_RENAMING) 1851 return; 1852 if (V_bridge_cloner == NULL) { 1853 /* 1854 * This detach handler can be called after 1855 * vnet_bridge_uninit(). Just return in that case. 1856 */ 1857 return; 1858 } 1859 NET_EPOCH_ENTER(et); 1860 /* Check if the interface is a bridge member */ 1861 if (sc != NULL) { 1862 BRIDGE_LOCK(sc); 1863 1864 bif = bridge_lookup_member_if(sc, ifp); 1865 if (bif != NULL) 1866 bridge_delete_member(sc, bif, 1); 1867 1868 BRIDGE_UNLOCK(sc); 1869 NET_EPOCH_EXIT(et); 1870 return; 1871 } 1872 1873 /* Check if the interface is a span port */ 1874 BRIDGE_LIST_LOCK(); 1875 LIST_FOREACH(sc, &V_bridge_list, sc_list) { 1876 BRIDGE_LOCK(sc); 1877 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1878 if (ifp == bif->bif_ifp) { 1879 bridge_delete_span(sc, bif); 1880 break; 1881 } 1882 1883 BRIDGE_UNLOCK(sc); 1884 } 1885 BRIDGE_LIST_UNLOCK(); 1886 NET_EPOCH_EXIT(et); 1887 } 1888 1889 /* 1890 * bridge_init: 1891 * 1892 * Initialize a bridge interface. 1893 */ 1894 static void 1895 bridge_init(void *xsc) 1896 { 1897 struct bridge_softc *sc = (struct bridge_softc *)xsc; 1898 struct ifnet *ifp = sc->sc_ifp; 1899 1900 if (ifp->if_drv_flags & IFF_DRV_RUNNING) 1901 return; 1902 1903 BRIDGE_LOCK(sc); 1904 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz, 1905 bridge_timer, sc); 1906 1907 ifp->if_drv_flags |= IFF_DRV_RUNNING; 1908 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */ 1909 1910 BRIDGE_UNLOCK(sc); 1911 } 1912 1913 /* 1914 * bridge_stop: 1915 * 1916 * Stop the bridge interface. 1917 */ 1918 static void 1919 bridge_stop(struct ifnet *ifp, int disable) 1920 { 1921 struct bridge_softc *sc = ifp->if_softc; 1922 1923 NET_EPOCH_ASSERT(); 1924 BRIDGE_LOCK_ASSERT(sc); 1925 1926 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) 1927 return; 1928 1929 callout_stop(&sc->sc_brcallout); 1930 bstp_stop(&sc->sc_stp); 1931 1932 bridge_rtflush(sc, IFBF_FLUSHDYN); 1933 1934 ifp->if_drv_flags &= ~IFF_DRV_RUNNING; 1935 } 1936 1937 /* 1938 * bridge_enqueue: 1939 * 1940 * Enqueue a packet on a bridge member interface. 1941 * 1942 */ 1943 static int 1944 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m) 1945 { 1946 int len, err = 0; 1947 short mflags; 1948 struct mbuf *m0; 1949 1950 /* We may be sending a fragment so traverse the mbuf */ 1951 for (; m; m = m0) { 1952 m0 = m->m_nextpkt; 1953 m->m_nextpkt = NULL; 1954 len = m->m_pkthdr.len; 1955 mflags = m->m_flags; 1956 1957 /* 1958 * If underlying interface can not do VLAN tag insertion itself 1959 * then attach a packet tag that holds it. 1960 */ 1961 if ((m->m_flags & M_VLANTAG) && 1962 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) { 1963 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag); 1964 if (m == NULL) { 1965 if_printf(dst_ifp, 1966 "unable to prepend VLAN header\n"); 1967 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1); 1968 continue; 1969 } 1970 m->m_flags &= ~M_VLANTAG; 1971 } 1972 1973 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */ 1974 if ((err = dst_ifp->if_transmit(dst_ifp, m))) { 1975 m_freem(m0); 1976 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 1977 break; 1978 } 1979 1980 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1); 1981 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len); 1982 if (mflags & M_MCAST) 1983 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1); 1984 } 1985 1986 return (err); 1987 } 1988 1989 /* 1990 * bridge_dummynet: 1991 * 1992 * Receive a queued packet from dummynet and pass it on to the output 1993 * interface. 1994 * 1995 * The mbuf has the Ethernet header already attached. 1996 */ 1997 static void 1998 bridge_dummynet(struct mbuf *m, struct ifnet *ifp) 1999 { 2000 struct bridge_softc *sc; 2001 2002 sc = ifp->if_bridge; 2003 2004 /* 2005 * The packet didnt originate from a member interface. This should only 2006 * ever happen if a member interface is removed while packets are 2007 * queued for it. 2008 */ 2009 if (sc == NULL) { 2010 m_freem(m); 2011 return; 2012 } 2013 2014 if (PFIL_HOOKED_OUT(V_inet_pfil_head) 2015 #ifdef INET6 2016 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2017 #endif 2018 ) { 2019 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0) 2020 return; 2021 if (m == NULL) 2022 return; 2023 } 2024 2025 bridge_enqueue(sc, ifp, m); 2026 } 2027 2028 /* 2029 * bridge_output: 2030 * 2031 * Send output from a bridge member interface. This 2032 * performs the bridging function for locally originated 2033 * packets. 2034 * 2035 * The mbuf has the Ethernet header already attached. We must 2036 * enqueue or free the mbuf before returning. 2037 */ 2038 static int 2039 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa, 2040 struct rtentry *rt) 2041 { 2042 struct ether_header *eh; 2043 struct ifnet *bifp, *dst_if; 2044 struct bridge_softc *sc; 2045 uint16_t vlan; 2046 2047 NET_EPOCH_ASSERT(); 2048 2049 if (m->m_len < ETHER_HDR_LEN) { 2050 m = m_pullup(m, ETHER_HDR_LEN); 2051 if (m == NULL) 2052 return (0); 2053 } 2054 2055 eh = mtod(m, struct ether_header *); 2056 sc = ifp->if_bridge; 2057 vlan = VLANTAGOF(m); 2058 2059 bifp = sc->sc_ifp; 2060 2061 /* 2062 * If bridge is down, but the original output interface is up, 2063 * go ahead and send out that interface. Otherwise, the packet 2064 * is dropped below. 2065 */ 2066 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { 2067 dst_if = ifp; 2068 goto sendunicast; 2069 } 2070 2071 /* 2072 * If the packet is a multicast, or we don't know a better way to 2073 * get there, send to all interfaces. 2074 */ 2075 if (ETHER_IS_MULTICAST(eh->ether_dhost)) 2076 dst_if = NULL; 2077 else 2078 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan); 2079 /* Tap any traffic not passing back out the originating interface */ 2080 if (dst_if != ifp) 2081 ETHER_BPF_MTAP(bifp, m); 2082 if (dst_if == NULL) { 2083 struct bridge_iflist *bif; 2084 struct mbuf *mc; 2085 int used = 0; 2086 2087 bridge_span(sc, m); 2088 2089 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 2090 dst_if = bif->bif_ifp; 2091 2092 if (dst_if->if_type == IFT_GIF) 2093 continue; 2094 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2095 continue; 2096 2097 /* 2098 * If this is not the original output interface, 2099 * and the interface is participating in spanning 2100 * tree, make sure the port is in a state that 2101 * allows forwarding. 2102 */ 2103 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) && 2104 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2105 continue; 2106 2107 if (CK_LIST_NEXT(bif, bif_next) == NULL) { 2108 used = 1; 2109 mc = m; 2110 } else { 2111 mc = m_copypacket(m, M_NOWAIT); 2112 if (mc == NULL) { 2113 if_inc_counter(bifp, IFCOUNTER_OERRORS, 1); 2114 continue; 2115 } 2116 } 2117 2118 bridge_enqueue(sc, dst_if, mc); 2119 } 2120 if (used == 0) 2121 m_freem(m); 2122 return (0); 2123 } 2124 2125 sendunicast: 2126 /* 2127 * XXX Spanning tree consideration here? 2128 */ 2129 2130 bridge_span(sc, m); 2131 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) { 2132 m_freem(m); 2133 return (0); 2134 } 2135 2136 bridge_enqueue(sc, dst_if, m); 2137 return (0); 2138 } 2139 2140 /* 2141 * bridge_transmit: 2142 * 2143 * Do output on a bridge. 2144 * 2145 */ 2146 static int 2147 bridge_transmit(struct ifnet *ifp, struct mbuf *m) 2148 { 2149 struct bridge_softc *sc; 2150 struct ether_header *eh; 2151 struct ifnet *dst_if; 2152 int error = 0; 2153 2154 sc = ifp->if_softc; 2155 2156 ETHER_BPF_MTAP(ifp, m); 2157 2158 eh = mtod(m, struct ether_header *); 2159 2160 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) && 2161 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) { 2162 error = bridge_enqueue(sc, dst_if, m); 2163 } else 2164 bridge_broadcast(sc, ifp, m, 0); 2165 2166 return (error); 2167 } 2168 2169 /* 2170 * The ifp->if_qflush entry point for if_bridge(4) is no-op. 2171 */ 2172 static void 2173 bridge_qflush(struct ifnet *ifp __unused) 2174 { 2175 } 2176 2177 /* 2178 * bridge_forward: 2179 * 2180 * The forwarding function of the bridge. 2181 * 2182 * NOTE: Releases the lock on return. 2183 */ 2184 static void 2185 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif, 2186 struct mbuf *m) 2187 { 2188 struct bridge_iflist *dbif; 2189 struct ifnet *src_if, *dst_if, *ifp; 2190 struct ether_header *eh; 2191 uint16_t vlan; 2192 uint8_t *dst; 2193 int error; 2194 2195 NET_EPOCH_ASSERT(); 2196 2197 src_if = m->m_pkthdr.rcvif; 2198 ifp = sc->sc_ifp; 2199 2200 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1); 2201 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); 2202 vlan = VLANTAGOF(m); 2203 2204 if ((sbif->bif_flags & IFBIF_STP) && 2205 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2206 goto drop; 2207 2208 eh = mtod(m, struct ether_header *); 2209 dst = eh->ether_dhost; 2210 2211 /* If the interface is learning, record the address. */ 2212 if (sbif->bif_flags & IFBIF_LEARNING) { 2213 error = bridge_rtupdate(sc, eh->ether_shost, vlan, 2214 sbif, 0, IFBAF_DYNAMIC); 2215 /* 2216 * If the interface has addresses limits then deny any source 2217 * that is not in the cache. 2218 */ 2219 if (error && sbif->bif_addrmax) 2220 goto drop; 2221 } 2222 2223 if ((sbif->bif_flags & IFBIF_STP) != 0 && 2224 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING) 2225 goto drop; 2226 2227 /* 2228 * At this point, the port either doesn't participate 2229 * in spanning tree or it is in the forwarding state. 2230 */ 2231 2232 /* 2233 * If the packet is unicast, destined for someone on 2234 * "this" side of the bridge, drop it. 2235 */ 2236 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) { 2237 dst_if = bridge_rtlookup(sc, dst, vlan); 2238 if (src_if == dst_if) 2239 goto drop; 2240 } else { 2241 /* 2242 * Check if its a reserved multicast address, any address 2243 * listed in 802.1D section 7.12.6 may not be forwarded by the 2244 * bridge. 2245 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F 2246 */ 2247 if (dst[0] == 0x01 && dst[1] == 0x80 && 2248 dst[2] == 0xc2 && dst[3] == 0x00 && 2249 dst[4] == 0x00 && dst[5] <= 0x0f) 2250 goto drop; 2251 2252 /* ...forward it to all interfaces. */ 2253 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1); 2254 dst_if = NULL; 2255 } 2256 2257 /* 2258 * If we have a destination interface which is a member of our bridge, 2259 * OR this is a unicast packet, push it through the bpf(4) machinery. 2260 * For broadcast or multicast packets, don't bother because it will 2261 * be reinjected into ether_input. We do this before we pass the packets 2262 * through the pfil(9) framework, as it is possible that pfil(9) will 2263 * drop the packet, or possibly modify it, making it difficult to debug 2264 * firewall issues on the bridge. 2265 */ 2266 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0) 2267 ETHER_BPF_MTAP(ifp, m); 2268 2269 /* run the packet filter */ 2270 if (PFIL_HOOKED_IN(V_inet_pfil_head) 2271 #ifdef INET6 2272 || PFIL_HOOKED_IN(V_inet6_pfil_head) 2273 #endif 2274 ) { 2275 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0) 2276 return; 2277 if (m == NULL) 2278 return; 2279 } 2280 2281 if (dst_if == NULL) { 2282 bridge_broadcast(sc, src_if, m, 1); 2283 return; 2284 } 2285 2286 /* 2287 * At this point, we're dealing with a unicast frame 2288 * going to a different interface. 2289 */ 2290 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2291 goto drop; 2292 2293 dbif = bridge_lookup_member_if(sc, dst_if); 2294 if (dbif == NULL) 2295 /* Not a member of the bridge (anymore?) */ 2296 goto drop; 2297 2298 /* Private segments can not talk to each other */ 2299 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE) 2300 goto drop; 2301 2302 if ((dbif->bif_flags & IFBIF_STP) && 2303 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2304 goto drop; 2305 2306 if (PFIL_HOOKED_OUT(V_inet_pfil_head) 2307 #ifdef INET6 2308 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2309 #endif 2310 ) { 2311 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0) 2312 return; 2313 if (m == NULL) 2314 return; 2315 } 2316 2317 bridge_enqueue(sc, dst_if, m); 2318 return; 2319 2320 drop: 2321 m_freem(m); 2322 } 2323 2324 /* 2325 * bridge_input: 2326 * 2327 * Receive input from a member interface. Queue the packet for 2328 * bridging if it is not for us. 2329 */ 2330 static struct mbuf * 2331 bridge_input(struct ifnet *ifp, struct mbuf *m) 2332 { 2333 struct bridge_softc *sc = ifp->if_bridge; 2334 struct bridge_iflist *bif, *bif2; 2335 struct ifnet *bifp; 2336 struct ether_header *eh; 2337 struct mbuf *mc, *mc2; 2338 uint16_t vlan; 2339 int error; 2340 2341 NET_EPOCH_ASSERT(); 2342 2343 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) 2344 return (m); 2345 2346 bifp = sc->sc_ifp; 2347 vlan = VLANTAGOF(m); 2348 2349 /* 2350 * Implement support for bridge monitoring. If this flag has been 2351 * set on this interface, discard the packet once we push it through 2352 * the bpf(4) machinery, but before we do, increment the byte and 2353 * packet counters associated with this interface. 2354 */ 2355 if ((bifp->if_flags & IFF_MONITOR) != 0) { 2356 m->m_pkthdr.rcvif = bifp; 2357 ETHER_BPF_MTAP(bifp, m); 2358 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); 2359 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); 2360 m_freem(m); 2361 return (NULL); 2362 } 2363 bif = bridge_lookup_member_if(sc, ifp); 2364 if (bif == NULL) { 2365 return (m); 2366 } 2367 2368 eh = mtod(m, struct ether_header *); 2369 2370 bridge_span(sc, m); 2371 2372 if (m->m_flags & (M_BCAST|M_MCAST)) { 2373 /* Tap off 802.1D packets; they do not get forwarded. */ 2374 if (memcmp(eh->ether_dhost, bstp_etheraddr, 2375 ETHER_ADDR_LEN) == 0) { 2376 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */ 2377 return (NULL); 2378 } 2379 2380 if ((bif->bif_flags & IFBIF_STP) && 2381 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) { 2382 return (m); 2383 } 2384 2385 /* 2386 * Make a deep copy of the packet and enqueue the copy 2387 * for bridge processing; return the original packet for 2388 * local processing. 2389 */ 2390 mc = m_dup(m, M_NOWAIT); 2391 if (mc == NULL) { 2392 return (m); 2393 } 2394 2395 /* Perform the bridge forwarding function with the copy. */ 2396 bridge_forward(sc, bif, mc); 2397 2398 /* 2399 * Reinject the mbuf as arriving on the bridge so we have a 2400 * chance at claiming multicast packets. We can not loop back 2401 * here from ether_input as a bridge is never a member of a 2402 * bridge. 2403 */ 2404 KASSERT(bifp->if_bridge == NULL, 2405 ("loop created in bridge_input")); 2406 mc2 = m_dup(m, M_NOWAIT); 2407 if (mc2 != NULL) { 2408 /* Keep the layer3 header aligned */ 2409 int i = min(mc2->m_pkthdr.len, max_protohdr); 2410 mc2 = m_copyup(mc2, i, ETHER_ALIGN); 2411 } 2412 if (mc2 != NULL) { 2413 mc2->m_pkthdr.rcvif = bifp; 2414 (*bifp->if_input)(bifp, mc2); 2415 } 2416 2417 /* Return the original packet for local processing. */ 2418 return (m); 2419 } 2420 2421 if ((bif->bif_flags & IFBIF_STP) && 2422 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) { 2423 return (m); 2424 } 2425 2426 #if (defined(INET) || defined(INET6)) 2427 # define OR_CARP_CHECK_WE_ARE_DST(iface) \ 2428 || ((iface)->if_carp \ 2429 && (*carp_forus_p)((iface), eh->ether_dhost)) 2430 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \ 2431 || ((iface)->if_carp \ 2432 && (*carp_forus_p)((iface), eh->ether_shost)) 2433 #else 2434 # define OR_CARP_CHECK_WE_ARE_DST(iface) 2435 # define OR_CARP_CHECK_WE_ARE_SRC(iface) 2436 #endif 2437 2438 #ifdef INET6 2439 # define OR_PFIL_HOOKED_INET6 \ 2440 || PFIL_HOOKED_IN(V_inet6_pfil_head) 2441 #else 2442 # define OR_PFIL_HOOKED_INET6 2443 #endif 2444 2445 #define GRAB_OUR_PACKETS(iface) \ 2446 if ((iface)->if_type == IFT_GIF) \ 2447 continue; \ 2448 /* It is destined for us. */ \ 2449 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \ 2450 OR_CARP_CHECK_WE_ARE_DST((iface)) \ 2451 ) { \ 2452 if (bif->bif_flags & IFBIF_LEARNING) { \ 2453 error = bridge_rtupdate(sc, eh->ether_shost, \ 2454 vlan, bif, 0, IFBAF_DYNAMIC); \ 2455 if (error && bif->bif_addrmax) { \ 2456 m_freem(m); \ 2457 return (NULL); \ 2458 } \ 2459 } \ 2460 m->m_pkthdr.rcvif = iface; \ 2461 if ((iface) == ifp) { \ 2462 /* Skip bridge processing... src == dest */ \ 2463 return (m); \ 2464 } \ 2465 /* It's passing over or to the bridge, locally. */ \ 2466 ETHER_BPF_MTAP(bifp, m); \ 2467 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \ 2468 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \ 2469 /* Filter on the physical interface. */ \ 2470 if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \ 2471 OR_PFIL_HOOKED_INET6)) { \ 2472 if (bridge_pfil(&m, NULL, ifp, \ 2473 PFIL_IN) != 0 || m == NULL) { \ 2474 return (NULL); \ 2475 } \ 2476 } \ 2477 if ((iface) != bifp) \ 2478 ETHER_BPF_MTAP(iface, m); \ 2479 return (m); \ 2480 } \ 2481 \ 2482 /* We just received a packet that we sent out. */ \ 2483 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \ 2484 OR_CARP_CHECK_WE_ARE_SRC((iface)) \ 2485 ) { \ 2486 m_freem(m); \ 2487 return (NULL); \ 2488 } 2489 2490 /* 2491 * Unicast. Make sure it's not for the bridge. 2492 */ 2493 do { GRAB_OUR_PACKETS(bifp) } while (0); 2494 2495 /* 2496 * Give a chance for ifp at first priority. This will help when the 2497 * packet comes through the interface like VLAN's with the same MACs 2498 * on several interfaces from the same bridge. This also will save 2499 * some CPU cycles in case the destination interface and the input 2500 * interface (eq ifp) are the same. 2501 */ 2502 do { GRAB_OUR_PACKETS(ifp) } while (0); 2503 2504 /* Now check the all bridge members. */ 2505 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) { 2506 GRAB_OUR_PACKETS(bif2->bif_ifp) 2507 } 2508 2509 #undef OR_CARP_CHECK_WE_ARE_DST 2510 #undef OR_CARP_CHECK_WE_ARE_SRC 2511 #undef OR_PFIL_HOOKED_INET6 2512 #undef GRAB_OUR_PACKETS 2513 2514 /* Perform the bridge forwarding function. */ 2515 bridge_forward(sc, bif, m); 2516 2517 return (NULL); 2518 } 2519 2520 /* 2521 * bridge_broadcast: 2522 * 2523 * Send a frame to all interfaces that are members of 2524 * the bridge, except for the one on which the packet 2525 * arrived. 2526 * 2527 * NOTE: Releases the lock on return. 2528 */ 2529 static void 2530 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if, 2531 struct mbuf *m, int runfilt) 2532 { 2533 struct bridge_iflist *dbif, *sbif; 2534 struct mbuf *mc; 2535 struct ifnet *dst_if; 2536 int used = 0, i; 2537 2538 NET_EPOCH_ASSERT(); 2539 2540 sbif = bridge_lookup_member_if(sc, src_if); 2541 2542 /* Filter on the bridge interface before broadcasting */ 2543 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head) 2544 #ifdef INET6 2545 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2546 #endif 2547 )) { 2548 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0) 2549 return; 2550 if (m == NULL) 2551 return; 2552 } 2553 2554 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) { 2555 dst_if = dbif->bif_ifp; 2556 if (dst_if == src_if) 2557 continue; 2558 2559 /* Private segments can not talk to each other */ 2560 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)) 2561 continue; 2562 2563 if ((dbif->bif_flags & IFBIF_STP) && 2564 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2565 continue; 2566 2567 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 && 2568 (m->m_flags & (M_BCAST|M_MCAST)) == 0) 2569 continue; 2570 2571 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2572 continue; 2573 2574 if (CK_LIST_NEXT(dbif, bif_next) == NULL) { 2575 mc = m; 2576 used = 1; 2577 } else { 2578 mc = m_dup(m, M_NOWAIT); 2579 if (mc == NULL) { 2580 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 2581 continue; 2582 } 2583 } 2584 2585 /* 2586 * Filter on the output interface. Pass a NULL bridge interface 2587 * pointer so we do not redundantly filter on the bridge for 2588 * each interface we broadcast on. 2589 */ 2590 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head) 2591 #ifdef INET6 2592 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2593 #endif 2594 )) { 2595 if (used == 0) { 2596 /* Keep the layer3 header aligned */ 2597 i = min(mc->m_pkthdr.len, max_protohdr); 2598 mc = m_copyup(mc, i, ETHER_ALIGN); 2599 if (mc == NULL) { 2600 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 2601 continue; 2602 } 2603 } 2604 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0) 2605 continue; 2606 if (mc == NULL) 2607 continue; 2608 } 2609 2610 bridge_enqueue(sc, dst_if, mc); 2611 } 2612 if (used == 0) 2613 m_freem(m); 2614 } 2615 2616 /* 2617 * bridge_span: 2618 * 2619 * Duplicate a packet out one or more interfaces that are in span mode, 2620 * the original mbuf is unmodified. 2621 */ 2622 static void 2623 bridge_span(struct bridge_softc *sc, struct mbuf *m) 2624 { 2625 struct bridge_iflist *bif; 2626 struct ifnet *dst_if; 2627 struct mbuf *mc; 2628 2629 NET_EPOCH_ASSERT(); 2630 2631 if (CK_LIST_EMPTY(&sc->sc_spanlist)) 2632 return; 2633 2634 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) { 2635 dst_if = bif->bif_ifp; 2636 2637 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2638 continue; 2639 2640 mc = m_copypacket(m, M_NOWAIT); 2641 if (mc == NULL) { 2642 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 2643 continue; 2644 } 2645 2646 bridge_enqueue(sc, dst_if, mc); 2647 } 2648 } 2649 2650 /* 2651 * bridge_rtupdate: 2652 * 2653 * Add a bridge routing entry. 2654 */ 2655 static int 2656 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan, 2657 struct bridge_iflist *bif, int setflags, uint8_t flags) 2658 { 2659 struct bridge_rtnode *brt; 2660 int error; 2661 2662 NET_EPOCH_ASSERT(); 2663 BRIDGE_UNLOCK_ASSERT(sc); 2664 2665 /* Check the source address is valid and not multicast. */ 2666 if (ETHER_IS_MULTICAST(dst) || 2667 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 && 2668 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0) 2669 return (EINVAL); 2670 2671 /* 802.1p frames map to vlan 1 */ 2672 if (vlan == 0) 2673 vlan = 1; 2674 2675 /* 2676 * A route for this destination might already exist. If so, 2677 * update it, otherwise create a new one. 2678 */ 2679 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) { 2680 BRIDGE_LOCK(sc); 2681 2682 /* Check again, now that we have the lock. There could have 2683 * been a race and we only want to insert this once. */ 2684 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) { 2685 BRIDGE_UNLOCK(sc); 2686 return (0); 2687 } 2688 2689 if (sc->sc_brtcnt >= sc->sc_brtmax) { 2690 sc->sc_brtexceeded++; 2691 BRIDGE_UNLOCK(sc); 2692 return (ENOSPC); 2693 } 2694 /* Check per interface address limits (if enabled) */ 2695 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) { 2696 bif->bif_addrexceeded++; 2697 BRIDGE_UNLOCK(sc); 2698 return (ENOSPC); 2699 } 2700 2701 /* 2702 * Allocate a new bridge forwarding node, and 2703 * initialize the expiration time and Ethernet 2704 * address. 2705 */ 2706 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO); 2707 if (brt == NULL) { 2708 BRIDGE_UNLOCK(sc); 2709 return (ENOMEM); 2710 } 2711 brt->brt_vnet = curvnet; 2712 2713 if (bif->bif_flags & IFBIF_STICKY) 2714 brt->brt_flags = IFBAF_STICKY; 2715 else 2716 brt->brt_flags = IFBAF_DYNAMIC; 2717 2718 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN); 2719 brt->brt_vlan = vlan; 2720 2721 if ((error = bridge_rtnode_insert(sc, brt)) != 0) { 2722 uma_zfree(V_bridge_rtnode_zone, brt); 2723 BRIDGE_UNLOCK(sc); 2724 return (error); 2725 } 2726 brt->brt_dst = bif; 2727 bif->bif_addrcnt++; 2728 2729 BRIDGE_UNLOCK(sc); 2730 } 2731 2732 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 2733 brt->brt_dst != bif) { 2734 BRIDGE_LOCK(sc); 2735 brt->brt_dst->bif_addrcnt--; 2736 brt->brt_dst = bif; 2737 brt->brt_dst->bif_addrcnt++; 2738 BRIDGE_UNLOCK(sc); 2739 } 2740 2741 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 2742 brt->brt_expire = time_uptime + sc->sc_brttimeout; 2743 if (setflags) 2744 brt->brt_flags = flags; 2745 2746 return (0); 2747 } 2748 2749 /* 2750 * bridge_rtlookup: 2751 * 2752 * Lookup the destination interface for an address. 2753 */ 2754 static struct ifnet * 2755 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) 2756 { 2757 struct bridge_rtnode *brt; 2758 2759 NET_EPOCH_ASSERT(); 2760 2761 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL) 2762 return (NULL); 2763 2764 return (brt->brt_ifp); 2765 } 2766 2767 /* 2768 * bridge_rttrim: 2769 * 2770 * Trim the routine table so that we have a number 2771 * of routing entries less than or equal to the 2772 * maximum number. 2773 */ 2774 static void 2775 bridge_rttrim(struct bridge_softc *sc) 2776 { 2777 struct bridge_rtnode *brt, *nbrt; 2778 2779 NET_EPOCH_ASSERT(); 2780 BRIDGE_LOCK_ASSERT(sc); 2781 2782 /* Make sure we actually need to do this. */ 2783 if (sc->sc_brtcnt <= sc->sc_brtmax) 2784 return; 2785 2786 /* Force an aging cycle; this might trim enough addresses. */ 2787 bridge_rtage(sc); 2788 if (sc->sc_brtcnt <= sc->sc_brtmax) 2789 return; 2790 2791 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 2792 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { 2793 bridge_rtnode_destroy(sc, brt); 2794 if (sc->sc_brtcnt <= sc->sc_brtmax) 2795 return; 2796 } 2797 } 2798 } 2799 2800 /* 2801 * bridge_timer: 2802 * 2803 * Aging timer for the bridge. 2804 */ 2805 static void 2806 bridge_timer(void *arg) 2807 { 2808 struct bridge_softc *sc = arg; 2809 struct epoch_tracker et; 2810 2811 NET_EPOCH_ENTER(et); 2812 BRIDGE_LOCK_ASSERT(sc); 2813 2814 /* Destruction of rtnodes requires a proper vnet context */ 2815 CURVNET_SET(sc->sc_ifp->if_vnet); 2816 bridge_rtage(sc); 2817 2818 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) 2819 callout_reset(&sc->sc_brcallout, 2820 bridge_rtable_prune_period * hz, bridge_timer, sc); 2821 CURVNET_RESTORE(); 2822 NET_EPOCH_EXIT(et); 2823 } 2824 2825 /* 2826 * bridge_rtage: 2827 * 2828 * Perform an aging cycle. 2829 */ 2830 static void 2831 bridge_rtage(struct bridge_softc *sc) 2832 { 2833 struct bridge_rtnode *brt, *nbrt; 2834 2835 NET_EPOCH_ASSERT(); 2836 BRIDGE_LOCK_ASSERT(sc); 2837 2838 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 2839 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { 2840 if (time_uptime >= brt->brt_expire) 2841 bridge_rtnode_destroy(sc, brt); 2842 } 2843 } 2844 } 2845 2846 /* 2847 * bridge_rtflush: 2848 * 2849 * Remove all dynamic addresses from the bridge. 2850 */ 2851 static void 2852 bridge_rtflush(struct bridge_softc *sc, int full) 2853 { 2854 struct bridge_rtnode *brt, *nbrt; 2855 2856 NET_EPOCH_ASSERT(); 2857 BRIDGE_LOCK_ASSERT(sc); 2858 2859 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 2860 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 2861 bridge_rtnode_destroy(sc, brt); 2862 } 2863 } 2864 2865 /* 2866 * bridge_rtdaddr: 2867 * 2868 * Remove an address from the table. 2869 */ 2870 static int 2871 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) 2872 { 2873 struct bridge_rtnode *brt; 2874 int found = 0; 2875 2876 NET_EPOCH_ASSERT(); 2877 BRIDGE_LOCK_ASSERT(sc); 2878 2879 /* 2880 * If vlan is zero then we want to delete for all vlans so the lookup 2881 * may return more than one. 2882 */ 2883 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) { 2884 bridge_rtnode_destroy(sc, brt); 2885 found = 1; 2886 } 2887 2888 return (found ? 0 : ENOENT); 2889 } 2890 2891 /* 2892 * bridge_rtdelete: 2893 * 2894 * Delete routes to a speicifc member interface. 2895 */ 2896 static void 2897 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full) 2898 { 2899 struct bridge_rtnode *brt, *nbrt; 2900 2901 NET_EPOCH_ASSERT(); 2902 BRIDGE_LOCK_ASSERT(sc); 2903 2904 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 2905 if (brt->brt_ifp == ifp && (full || 2906 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)) 2907 bridge_rtnode_destroy(sc, brt); 2908 } 2909 } 2910 2911 /* 2912 * bridge_rtable_init: 2913 * 2914 * Initialize the route table for this bridge. 2915 */ 2916 static void 2917 bridge_rtable_init(struct bridge_softc *sc) 2918 { 2919 int i; 2920 2921 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE, 2922 M_DEVBUF, M_WAITOK); 2923 2924 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++) 2925 CK_LIST_INIT(&sc->sc_rthash[i]); 2926 2927 sc->sc_rthash_key = arc4random(); 2928 CK_LIST_INIT(&sc->sc_rtlist); 2929 } 2930 2931 /* 2932 * bridge_rtable_fini: 2933 * 2934 * Deconstruct the route table for this bridge. 2935 */ 2936 static void 2937 bridge_rtable_fini(struct bridge_softc *sc) 2938 { 2939 2940 KASSERT(sc->sc_brtcnt == 0, 2941 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt)); 2942 free(sc->sc_rthash, M_DEVBUF); 2943 } 2944 2945 /* 2946 * The following hash function is adapted from "Hash Functions" by Bob Jenkins 2947 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997). 2948 */ 2949 #define mix(a, b, c) \ 2950 do { \ 2951 a -= b; a -= c; a ^= (c >> 13); \ 2952 b -= c; b -= a; b ^= (a << 8); \ 2953 c -= a; c -= b; c ^= (b >> 13); \ 2954 a -= b; a -= c; a ^= (c >> 12); \ 2955 b -= c; b -= a; b ^= (a << 16); \ 2956 c -= a; c -= b; c ^= (b >> 5); \ 2957 a -= b; a -= c; a ^= (c >> 3); \ 2958 b -= c; b -= a; b ^= (a << 10); \ 2959 c -= a; c -= b; c ^= (b >> 15); \ 2960 } while (/*CONSTCOND*/0) 2961 2962 static __inline uint32_t 2963 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr) 2964 { 2965 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key; 2966 2967 b += addr[5] << 8; 2968 b += addr[4]; 2969 a += addr[3] << 24; 2970 a += addr[2] << 16; 2971 a += addr[1] << 8; 2972 a += addr[0]; 2973 2974 mix(a, b, c); 2975 2976 return (c & BRIDGE_RTHASH_MASK); 2977 } 2978 2979 #undef mix 2980 2981 static int 2982 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b) 2983 { 2984 int i, d; 2985 2986 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) { 2987 d = ((int)a[i]) - ((int)b[i]); 2988 } 2989 2990 return (d); 2991 } 2992 2993 /* 2994 * bridge_rtnode_lookup: 2995 * 2996 * Look up a bridge route node for the specified destination. Compare the 2997 * vlan id or if zero then just return the first match. 2998 */ 2999 static struct bridge_rtnode * 3000 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) 3001 { 3002 struct bridge_rtnode *brt; 3003 uint32_t hash; 3004 int dir; 3005 3006 NET_EPOCH_ASSERT(); 3007 3008 hash = bridge_rthash(sc, addr); 3009 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) { 3010 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr); 3011 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0)) 3012 return (brt); 3013 if (dir > 0) 3014 return (NULL); 3015 } 3016 3017 return (NULL); 3018 } 3019 3020 /* 3021 * bridge_rtnode_insert: 3022 * 3023 * Insert the specified bridge node into the route table. We 3024 * assume the entry is not already in the table. 3025 */ 3026 static int 3027 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt) 3028 { 3029 struct bridge_rtnode *lbrt; 3030 uint32_t hash; 3031 int dir; 3032 3033 BRIDGE_LOCK_ASSERT(sc); 3034 3035 hash = bridge_rthash(sc, brt->brt_addr); 3036 3037 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]); 3038 if (lbrt == NULL) { 3039 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash); 3040 goto out; 3041 } 3042 3043 do { 3044 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr); 3045 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan) 3046 return (EEXIST); 3047 if (dir > 0) { 3048 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash); 3049 goto out; 3050 } 3051 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) { 3052 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash); 3053 goto out; 3054 } 3055 lbrt = CK_LIST_NEXT(lbrt, brt_hash); 3056 } while (lbrt != NULL); 3057 3058 #ifdef DIAGNOSTIC 3059 panic("bridge_rtnode_insert: impossible"); 3060 #endif 3061 3062 out: 3063 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list); 3064 sc->sc_brtcnt++; 3065 3066 return (0); 3067 } 3068 3069 static void 3070 bridge_rtnode_destroy_cb(struct epoch_context *ctx) 3071 { 3072 struct bridge_rtnode *brt; 3073 3074 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx); 3075 3076 CURVNET_SET(brt->brt_vnet); 3077 uma_zfree(V_bridge_rtnode_zone, brt); 3078 CURVNET_RESTORE(); 3079 } 3080 3081 /* 3082 * bridge_rtnode_destroy: 3083 * 3084 * Destroy a bridge rtnode. 3085 */ 3086 static void 3087 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt) 3088 { 3089 NET_EPOCH_ASSERT(); 3090 BRIDGE_LOCK_ASSERT(sc); 3091 3092 CK_LIST_REMOVE(brt, brt_hash); 3093 3094 CK_LIST_REMOVE(brt, brt_list); 3095 sc->sc_brtcnt--; 3096 brt->brt_dst->bif_addrcnt--; 3097 3098 NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx); 3099 } 3100 3101 /* 3102 * bridge_rtable_expire: 3103 * 3104 * Set the expiry time for all routes on an interface. 3105 */ 3106 static void 3107 bridge_rtable_expire(struct ifnet *ifp, int age) 3108 { 3109 struct bridge_softc *sc = ifp->if_bridge; 3110 struct bridge_rtnode *brt; 3111 struct epoch_tracker et; 3112 3113 NET_EPOCH_ENTER(et); 3114 CURVNET_SET(ifp->if_vnet); 3115 BRIDGE_LOCK(sc); 3116 3117 /* 3118 * If the age is zero then flush, otherwise set all the expiry times to 3119 * age for the interface 3120 */ 3121 if (age == 0) 3122 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN); 3123 else { 3124 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) { 3125 /* Cap the expiry time to 'age' */ 3126 if (brt->brt_ifp == ifp && 3127 brt->brt_expire > time_uptime + age && 3128 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 3129 brt->brt_expire = time_uptime + age; 3130 } 3131 } 3132 BRIDGE_UNLOCK(sc); 3133 CURVNET_RESTORE(); 3134 NET_EPOCH_EXIT(et); 3135 } 3136 3137 /* 3138 * bridge_state_change: 3139 * 3140 * Callback from the bridgestp code when a port changes states. 3141 */ 3142 static void 3143 bridge_state_change(struct ifnet *ifp, int state) 3144 { 3145 struct bridge_softc *sc = ifp->if_bridge; 3146 static const char *stpstates[] = { 3147 "disabled", 3148 "listening", 3149 "learning", 3150 "forwarding", 3151 "blocking", 3152 "discarding" 3153 }; 3154 3155 CURVNET_SET(ifp->if_vnet); 3156 if (V_log_stp) 3157 log(LOG_NOTICE, "%s: state changed to %s on %s\n", 3158 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname); 3159 CURVNET_RESTORE(); 3160 } 3161 3162 /* 3163 * Send bridge packets through pfil if they are one of the types pfil can deal 3164 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without 3165 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for 3166 * that interface. 3167 */ 3168 static int 3169 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir) 3170 { 3171 int snap, error, i, hlen; 3172 struct ether_header *eh1, eh2; 3173 struct ip *ip; 3174 struct llc llc1; 3175 u_int16_t ether_type; 3176 pfil_return_t rv; 3177 3178 snap = 0; 3179 error = -1; /* Default error if not error == 0 */ 3180 3181 #if 0 3182 /* we may return with the IP fields swapped, ensure its not shared */ 3183 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__)); 3184 #endif 3185 3186 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0) 3187 return (0); /* filtering is disabled */ 3188 3189 i = min((*mp)->m_pkthdr.len, max_protohdr); 3190 if ((*mp)->m_len < i) { 3191 *mp = m_pullup(*mp, i); 3192 if (*mp == NULL) { 3193 printf("%s: m_pullup failed\n", __func__); 3194 return (-1); 3195 } 3196 } 3197 3198 eh1 = mtod(*mp, struct ether_header *); 3199 ether_type = ntohs(eh1->ether_type); 3200 3201 /* 3202 * Check for SNAP/LLC. 3203 */ 3204 if (ether_type < ETHERMTU) { 3205 struct llc *llc2 = (struct llc *)(eh1 + 1); 3206 3207 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 && 3208 llc2->llc_dsap == LLC_SNAP_LSAP && 3209 llc2->llc_ssap == LLC_SNAP_LSAP && 3210 llc2->llc_control == LLC_UI) { 3211 ether_type = htons(llc2->llc_un.type_snap.ether_type); 3212 snap = 1; 3213 } 3214 } 3215 3216 /* 3217 * If we're trying to filter bridge traffic, don't look at anything 3218 * other than IP and ARP traffic. If the filter doesn't understand 3219 * IPv6, don't allow IPv6 through the bridge either. This is lame 3220 * since if we really wanted, say, an AppleTalk filter, we are hosed, 3221 * but of course we don't have an AppleTalk filter to begin with. 3222 * (Note that since pfil doesn't understand ARP it will pass *ALL* 3223 * ARP traffic.) 3224 */ 3225 switch (ether_type) { 3226 case ETHERTYPE_ARP: 3227 case ETHERTYPE_REVARP: 3228 if (V_pfil_ipfw_arp == 0) 3229 return (0); /* Automatically pass */ 3230 break; 3231 3232 case ETHERTYPE_IP: 3233 #ifdef INET6 3234 case ETHERTYPE_IPV6: 3235 #endif /* INET6 */ 3236 break; 3237 default: 3238 /* 3239 * Check to see if the user wants to pass non-ip 3240 * packets, these will not be checked by pfil(9) and 3241 * passed unconditionally so the default is to drop. 3242 */ 3243 if (V_pfil_onlyip) 3244 goto bad; 3245 } 3246 3247 /* Run the packet through pfil before stripping link headers */ 3248 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 && 3249 dir == PFIL_OUT && ifp != NULL) { 3250 switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) { 3251 case PFIL_DROPPED: 3252 return (EACCES); 3253 case PFIL_CONSUMED: 3254 return (0); 3255 } 3256 } 3257 3258 /* Strip off the Ethernet header and keep a copy. */ 3259 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2); 3260 m_adj(*mp, ETHER_HDR_LEN); 3261 3262 /* Strip off snap header, if present */ 3263 if (snap) { 3264 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1); 3265 m_adj(*mp, sizeof(struct llc)); 3266 } 3267 3268 /* 3269 * Check the IP header for alignment and errors 3270 */ 3271 if (dir == PFIL_IN) { 3272 switch (ether_type) { 3273 case ETHERTYPE_IP: 3274 error = bridge_ip_checkbasic(mp); 3275 break; 3276 #ifdef INET6 3277 case ETHERTYPE_IPV6: 3278 error = bridge_ip6_checkbasic(mp); 3279 break; 3280 #endif /* INET6 */ 3281 default: 3282 error = 0; 3283 } 3284 if (error) 3285 goto bad; 3286 } 3287 3288 error = 0; 3289 3290 /* 3291 * Run the packet through pfil 3292 */ 3293 rv = PFIL_PASS; 3294 switch (ether_type) { 3295 case ETHERTYPE_IP: 3296 /* 3297 * Run pfil on the member interface and the bridge, both can 3298 * be skipped by clearing pfil_member or pfil_bridge. 3299 * 3300 * Keep the order: 3301 * in_if -> bridge_if -> out_if 3302 */ 3303 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv = 3304 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) != 3305 PFIL_PASS) 3306 break; 3307 3308 if (V_pfil_member && ifp != NULL && (rv = 3309 pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) != 3310 PFIL_PASS) 3311 break; 3312 3313 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv = 3314 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) != 3315 PFIL_PASS) 3316 break; 3317 3318 /* check if we need to fragment the packet */ 3319 /* bridge_fragment generates a mbuf chain of packets */ 3320 /* that already include eth headers */ 3321 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) { 3322 i = (*mp)->m_pkthdr.len; 3323 if (i > ifp->if_mtu) { 3324 error = bridge_fragment(ifp, mp, &eh2, snap, 3325 &llc1); 3326 return (error); 3327 } 3328 } 3329 3330 /* Recalculate the ip checksum. */ 3331 ip = mtod(*mp, struct ip *); 3332 hlen = ip->ip_hl << 2; 3333 if (hlen < sizeof(struct ip)) 3334 goto bad; 3335 if (hlen > (*mp)->m_len) { 3336 if ((*mp = m_pullup(*mp, hlen)) == NULL) 3337 goto bad; 3338 ip = mtod(*mp, struct ip *); 3339 if (ip == NULL) 3340 goto bad; 3341 } 3342 ip->ip_sum = 0; 3343 if (hlen == sizeof(struct ip)) 3344 ip->ip_sum = in_cksum_hdr(ip); 3345 else 3346 ip->ip_sum = in_cksum(*mp, hlen); 3347 3348 break; 3349 #ifdef INET6 3350 case ETHERTYPE_IPV6: 3351 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv = 3352 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) != 3353 PFIL_PASS) 3354 break; 3355 3356 if (V_pfil_member && ifp != NULL && (rv = 3357 pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) != 3358 PFIL_PASS) 3359 break; 3360 3361 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv = 3362 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) != 3363 PFIL_PASS) 3364 break; 3365 break; 3366 #endif 3367 } 3368 3369 switch (rv) { 3370 case PFIL_CONSUMED: 3371 return (0); 3372 case PFIL_DROPPED: 3373 return (EACCES); 3374 default: 3375 break; 3376 } 3377 3378 error = -1; 3379 3380 /* 3381 * Finally, put everything back the way it was and return 3382 */ 3383 if (snap) { 3384 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT); 3385 if (*mp == NULL) 3386 return (error); 3387 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc)); 3388 } 3389 3390 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT); 3391 if (*mp == NULL) 3392 return (error); 3393 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN); 3394 3395 return (0); 3396 3397 bad: 3398 m_freem(*mp); 3399 *mp = NULL; 3400 return (error); 3401 } 3402 3403 /* 3404 * Perform basic checks on header size since 3405 * pfil assumes ip_input has already processed 3406 * it for it. Cut-and-pasted from ip_input.c. 3407 * Given how simple the IPv6 version is, 3408 * does the IPv4 version really need to be 3409 * this complicated? 3410 * 3411 * XXX Should we update ipstat here, or not? 3412 * XXX Right now we update ipstat but not 3413 * XXX csum_counter. 3414 */ 3415 static int 3416 bridge_ip_checkbasic(struct mbuf **mp) 3417 { 3418 struct mbuf *m = *mp; 3419 struct ip *ip; 3420 int len, hlen; 3421 u_short sum; 3422 3423 if (*mp == NULL) 3424 return (-1); 3425 3426 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 3427 if ((m = m_copyup(m, sizeof(struct ip), 3428 (max_linkhdr + 3) & ~3)) == NULL) { 3429 /* XXXJRT new stat, please */ 3430 KMOD_IPSTAT_INC(ips_toosmall); 3431 goto bad; 3432 } 3433 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 3434 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 3435 KMOD_IPSTAT_INC(ips_toosmall); 3436 goto bad; 3437 } 3438 } 3439 ip = mtod(m, struct ip *); 3440 if (ip == NULL) goto bad; 3441 3442 if (ip->ip_v != IPVERSION) { 3443 KMOD_IPSTAT_INC(ips_badvers); 3444 goto bad; 3445 } 3446 hlen = ip->ip_hl << 2; 3447 if (hlen < sizeof(struct ip)) { /* minimum header length */ 3448 KMOD_IPSTAT_INC(ips_badhlen); 3449 goto bad; 3450 } 3451 if (hlen > m->m_len) { 3452 if ((m = m_pullup(m, hlen)) == NULL) { 3453 KMOD_IPSTAT_INC(ips_badhlen); 3454 goto bad; 3455 } 3456 ip = mtod(m, struct ip *); 3457 if (ip == NULL) goto bad; 3458 } 3459 3460 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) { 3461 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID); 3462 } else { 3463 if (hlen == sizeof(struct ip)) { 3464 sum = in_cksum_hdr(ip); 3465 } else { 3466 sum = in_cksum(m, hlen); 3467 } 3468 } 3469 if (sum) { 3470 KMOD_IPSTAT_INC(ips_badsum); 3471 goto bad; 3472 } 3473 3474 /* Retrieve the packet length. */ 3475 len = ntohs(ip->ip_len); 3476 3477 /* 3478 * Check for additional length bogosity 3479 */ 3480 if (len < hlen) { 3481 KMOD_IPSTAT_INC(ips_badlen); 3482 goto bad; 3483 } 3484 3485 /* 3486 * Check that the amount of data in the buffers 3487 * is as at least much as the IP header would have us expect. 3488 * Drop packet if shorter than we expect. 3489 */ 3490 if (m->m_pkthdr.len < len) { 3491 KMOD_IPSTAT_INC(ips_tooshort); 3492 goto bad; 3493 } 3494 3495 /* Checks out, proceed */ 3496 *mp = m; 3497 return (0); 3498 3499 bad: 3500 *mp = m; 3501 return (-1); 3502 } 3503 3504 #ifdef INET6 3505 /* 3506 * Same as above, but for IPv6. 3507 * Cut-and-pasted from ip6_input.c. 3508 * XXX Should we update ip6stat, or not? 3509 */ 3510 static int 3511 bridge_ip6_checkbasic(struct mbuf **mp) 3512 { 3513 struct mbuf *m = *mp; 3514 struct ip6_hdr *ip6; 3515 3516 /* 3517 * If the IPv6 header is not aligned, slurp it up into a new 3518 * mbuf with space for link headers, in the event we forward 3519 * it. Otherwise, if it is aligned, make sure the entire base 3520 * IPv6 header is in the first mbuf of the chain. 3521 */ 3522 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 3523 struct ifnet *inifp = m->m_pkthdr.rcvif; 3524 if ((m = m_copyup(m, sizeof(struct ip6_hdr), 3525 (max_linkhdr + 3) & ~3)) == NULL) { 3526 /* XXXJRT new stat, please */ 3527 IP6STAT_INC(ip6s_toosmall); 3528 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 3529 goto bad; 3530 } 3531 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) { 3532 struct ifnet *inifp = m->m_pkthdr.rcvif; 3533 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 3534 IP6STAT_INC(ip6s_toosmall); 3535 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 3536 goto bad; 3537 } 3538 } 3539 3540 ip6 = mtod(m, struct ip6_hdr *); 3541 3542 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 3543 IP6STAT_INC(ip6s_badvers); 3544 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 3545 goto bad; 3546 } 3547 3548 /* Checks out, proceed */ 3549 *mp = m; 3550 return (0); 3551 3552 bad: 3553 *mp = m; 3554 return (-1); 3555 } 3556 #endif /* INET6 */ 3557 3558 /* 3559 * bridge_fragment: 3560 * 3561 * Fragment mbuf chain in multiple packets and prepend ethernet header. 3562 */ 3563 static int 3564 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh, 3565 int snap, struct llc *llc) 3566 { 3567 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL; 3568 struct ip *ip; 3569 int error = -1; 3570 3571 if (m->m_len < sizeof(struct ip) && 3572 (m = m_pullup(m, sizeof(struct ip))) == NULL) 3573 goto dropit; 3574 ip = mtod(m, struct ip *); 3575 3576 m->m_pkthdr.csum_flags |= CSUM_IP; 3577 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist); 3578 if (error) 3579 goto dropit; 3580 3581 /* 3582 * Walk the chain and re-add the Ethernet header for 3583 * each mbuf packet. 3584 */ 3585 for (mcur = m; mcur; mcur = mcur->m_nextpkt) { 3586 nextpkt = mcur->m_nextpkt; 3587 mcur->m_nextpkt = NULL; 3588 if (snap) { 3589 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT); 3590 if (mcur == NULL) { 3591 error = ENOBUFS; 3592 if (mprev != NULL) 3593 mprev->m_nextpkt = nextpkt; 3594 goto dropit; 3595 } 3596 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc)); 3597 } 3598 3599 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT); 3600 if (mcur == NULL) { 3601 error = ENOBUFS; 3602 if (mprev != NULL) 3603 mprev->m_nextpkt = nextpkt; 3604 goto dropit; 3605 } 3606 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN); 3607 3608 /* 3609 * The previous two M_PREPEND could have inserted one or two 3610 * mbufs in front so we have to update the previous packet's 3611 * m_nextpkt. 3612 */ 3613 mcur->m_nextpkt = nextpkt; 3614 if (mprev != NULL) 3615 mprev->m_nextpkt = mcur; 3616 else { 3617 /* The first mbuf in the original chain needs to be 3618 * updated. */ 3619 *mp = mcur; 3620 } 3621 mprev = mcur; 3622 } 3623 3624 KMOD_IPSTAT_INC(ips_fragmented); 3625 return (error); 3626 3627 dropit: 3628 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */ 3629 m = mcur->m_nextpkt; 3630 m_freem(mcur); 3631 } 3632 return (error); 3633 } 3634 3635 static void 3636 bridge_linkstate(struct ifnet *ifp) 3637 { 3638 struct bridge_softc *sc = ifp->if_bridge; 3639 struct bridge_iflist *bif; 3640 struct epoch_tracker et; 3641 3642 NET_EPOCH_ENTER(et); 3643 3644 bif = bridge_lookup_member_if(sc, ifp); 3645 if (bif == NULL) { 3646 NET_EPOCH_EXIT(et); 3647 return; 3648 } 3649 bridge_linkcheck(sc); 3650 3651 bstp_linkstate(&bif->bif_stp); 3652 3653 NET_EPOCH_EXIT(et); 3654 } 3655 3656 static void 3657 bridge_linkcheck(struct bridge_softc *sc) 3658 { 3659 struct bridge_iflist *bif; 3660 int new_link, hasls; 3661 3662 NET_EPOCH_ASSERT(); 3663 3664 new_link = LINK_STATE_DOWN; 3665 hasls = 0; 3666 /* Our link is considered up if at least one of our ports is active */ 3667 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 3668 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE) 3669 hasls++; 3670 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) { 3671 new_link = LINK_STATE_UP; 3672 break; 3673 } 3674 } 3675 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) { 3676 /* If no interfaces support link-state then we default to up */ 3677 new_link = LINK_STATE_UP; 3678 } 3679 if_link_state_change(sc->sc_ifp, new_link); 3680 } 3681