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); 366 367 368 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */ 369 #define VLANTAGOF(_m) \ 370 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1 371 372 static struct bstp_cb_ops bridge_ops = { 373 .bcb_state = bridge_state_change, 374 .bcb_rtage = bridge_rtable_expire 375 }; 376 377 SYSCTL_DECL(_net_link); 378 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, 379 "Bridge"); 380 381 /* only pass IP[46] packets when pfil is enabled */ 382 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1; 383 #define V_pfil_onlyip VNET(pfil_onlyip) 384 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, 385 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0, 386 "Only pass IP packets when pfil is enabled"); 387 388 /* run pfil hooks on the bridge interface */ 389 VNET_DEFINE_STATIC(int, pfil_bridge) = 1; 390 #define V_pfil_bridge VNET(pfil_bridge) 391 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, 392 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0, 393 "Packet filter on the bridge interface"); 394 395 /* layer2 filter with ipfw */ 396 VNET_DEFINE_STATIC(int, pfil_ipfw); 397 #define V_pfil_ipfw VNET(pfil_ipfw) 398 399 /* layer2 ARP filter with ipfw */ 400 VNET_DEFINE_STATIC(int, pfil_ipfw_arp); 401 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp) 402 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, 403 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0, 404 "Filter ARP packets through IPFW layer2"); 405 406 /* run pfil hooks on the member interface */ 407 VNET_DEFINE_STATIC(int, pfil_member) = 1; 408 #define V_pfil_member VNET(pfil_member) 409 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, 410 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0, 411 "Packet filter on the member interface"); 412 413 /* run pfil hooks on the physical interface for locally destined packets */ 414 VNET_DEFINE_STATIC(int, pfil_local_phys); 415 #define V_pfil_local_phys VNET(pfil_local_phys) 416 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, 417 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0, 418 "Packet filter on the physical interface for locally destined packets"); 419 420 /* log STP state changes */ 421 VNET_DEFINE_STATIC(int, log_stp); 422 #define V_log_stp VNET(log_stp) 423 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, 424 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0, 425 "Log STP state changes"); 426 427 /* share MAC with first bridge member */ 428 VNET_DEFINE_STATIC(int, bridge_inherit_mac); 429 #define V_bridge_inherit_mac VNET(bridge_inherit_mac) 430 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, 431 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0, 432 "Inherit MAC address from the first bridge member"); 433 434 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0; 435 #define V_allow_llz_overlap VNET(allow_llz_overlap) 436 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap, 437 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0, 438 "Allow overlap of link-local scope " 439 "zones of a bridge interface and the member interfaces"); 440 441 struct bridge_control { 442 int (*bc_func)(struct bridge_softc *, void *); 443 int bc_argsize; 444 int bc_flags; 445 }; 446 447 #define BC_F_COPYIN 0x01 /* copy arguments in */ 448 #define BC_F_COPYOUT 0x02 /* copy arguments out */ 449 #define BC_F_SUSER 0x04 /* do super-user check */ 450 451 const struct bridge_control bridge_control_table[] = { 452 { bridge_ioctl_add, sizeof(struct ifbreq), 453 BC_F_COPYIN|BC_F_SUSER }, 454 { bridge_ioctl_del, sizeof(struct ifbreq), 455 BC_F_COPYIN|BC_F_SUSER }, 456 457 { bridge_ioctl_gifflags, sizeof(struct ifbreq), 458 BC_F_COPYIN|BC_F_COPYOUT }, 459 { bridge_ioctl_sifflags, sizeof(struct ifbreq), 460 BC_F_COPYIN|BC_F_SUSER }, 461 462 { bridge_ioctl_scache, sizeof(struct ifbrparam), 463 BC_F_COPYIN|BC_F_SUSER }, 464 { bridge_ioctl_gcache, sizeof(struct ifbrparam), 465 BC_F_COPYOUT }, 466 467 { bridge_ioctl_gifs, sizeof(struct ifbifconf), 468 BC_F_COPYIN|BC_F_COPYOUT }, 469 { bridge_ioctl_rts, sizeof(struct ifbaconf), 470 BC_F_COPYIN|BC_F_COPYOUT }, 471 472 { bridge_ioctl_saddr, sizeof(struct ifbareq), 473 BC_F_COPYIN|BC_F_SUSER }, 474 475 { bridge_ioctl_sto, sizeof(struct ifbrparam), 476 BC_F_COPYIN|BC_F_SUSER }, 477 { bridge_ioctl_gto, sizeof(struct ifbrparam), 478 BC_F_COPYOUT }, 479 480 { bridge_ioctl_daddr, sizeof(struct ifbareq), 481 BC_F_COPYIN|BC_F_SUSER }, 482 483 { bridge_ioctl_flush, sizeof(struct ifbreq), 484 BC_F_COPYIN|BC_F_SUSER }, 485 486 { bridge_ioctl_gpri, sizeof(struct ifbrparam), 487 BC_F_COPYOUT }, 488 { bridge_ioctl_spri, sizeof(struct ifbrparam), 489 BC_F_COPYIN|BC_F_SUSER }, 490 491 { bridge_ioctl_ght, sizeof(struct ifbrparam), 492 BC_F_COPYOUT }, 493 { bridge_ioctl_sht, sizeof(struct ifbrparam), 494 BC_F_COPYIN|BC_F_SUSER }, 495 496 { bridge_ioctl_gfd, sizeof(struct ifbrparam), 497 BC_F_COPYOUT }, 498 { bridge_ioctl_sfd, sizeof(struct ifbrparam), 499 BC_F_COPYIN|BC_F_SUSER }, 500 501 { bridge_ioctl_gma, sizeof(struct ifbrparam), 502 BC_F_COPYOUT }, 503 { bridge_ioctl_sma, sizeof(struct ifbrparam), 504 BC_F_COPYIN|BC_F_SUSER }, 505 506 { bridge_ioctl_sifprio, sizeof(struct ifbreq), 507 BC_F_COPYIN|BC_F_SUSER }, 508 509 { bridge_ioctl_sifcost, sizeof(struct ifbreq), 510 BC_F_COPYIN|BC_F_SUSER }, 511 512 { bridge_ioctl_addspan, sizeof(struct ifbreq), 513 BC_F_COPYIN|BC_F_SUSER }, 514 { bridge_ioctl_delspan, sizeof(struct ifbreq), 515 BC_F_COPYIN|BC_F_SUSER }, 516 517 { bridge_ioctl_gbparam, sizeof(struct ifbropreq), 518 BC_F_COPYOUT }, 519 520 { bridge_ioctl_grte, sizeof(struct ifbrparam), 521 BC_F_COPYOUT }, 522 523 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf), 524 BC_F_COPYIN|BC_F_COPYOUT }, 525 526 { bridge_ioctl_sproto, sizeof(struct ifbrparam), 527 BC_F_COPYIN|BC_F_SUSER }, 528 529 { bridge_ioctl_stxhc, sizeof(struct ifbrparam), 530 BC_F_COPYIN|BC_F_SUSER }, 531 532 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq), 533 BC_F_COPYIN|BC_F_SUSER }, 534 535 }; 536 const int bridge_control_table_size = nitems(bridge_control_table); 537 538 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list); 539 #define V_bridge_list VNET(bridge_list) 540 #define BRIDGE_LIST_LOCK_INIT(x) mtx_init(&V_bridge_list_mtx, \ 541 "if_bridge list", NULL, MTX_DEF) 542 #define BRIDGE_LIST_LOCK_DESTROY(x) mtx_destroy(&V_bridge_list_mtx) 543 #define BRIDGE_LIST_LOCK(x) mtx_lock(&V_bridge_list_mtx) 544 #define BRIDGE_LIST_UNLOCK(x) mtx_unlock(&V_bridge_list_mtx) 545 546 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner); 547 #define V_bridge_cloner VNET(bridge_cloner) 548 549 static const char bridge_name[] = "bridge"; 550 551 static void 552 vnet_bridge_init(const void *unused __unused) 553 { 554 555 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode", 556 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL, 557 UMA_ALIGN_PTR, 0); 558 BRIDGE_LIST_LOCK_INIT(); 559 LIST_INIT(&V_bridge_list); 560 V_bridge_cloner = if_clone_simple(bridge_name, 561 bridge_clone_create, bridge_clone_destroy, 0); 562 } 563 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY, 564 vnet_bridge_init, NULL); 565 566 static void 567 vnet_bridge_uninit(const void *unused __unused) 568 { 569 570 if_clone_detach(V_bridge_cloner); 571 V_bridge_cloner = NULL; 572 BRIDGE_LIST_LOCK_DESTROY(); 573 574 /* Callbacks may use the UMA zone. */ 575 epoch_drain_callbacks(net_epoch_preempt); 576 577 uma_zdestroy(V_bridge_rtnode_zone); 578 } 579 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY, 580 vnet_bridge_uninit, NULL); 581 582 static int 583 bridge_modevent(module_t mod, int type, void *data) 584 { 585 586 switch (type) { 587 case MOD_LOAD: 588 bridge_dn_p = bridge_dummynet; 589 bridge_detach_cookie = EVENTHANDLER_REGISTER( 590 ifnet_departure_event, bridge_ifdetach, NULL, 591 EVENTHANDLER_PRI_ANY); 592 break; 593 case MOD_UNLOAD: 594 EVENTHANDLER_DEREGISTER(ifnet_departure_event, 595 bridge_detach_cookie); 596 bridge_dn_p = NULL; 597 break; 598 default: 599 return (EOPNOTSUPP); 600 } 601 return (0); 602 } 603 604 static moduledata_t bridge_mod = { 605 "if_bridge", 606 bridge_modevent, 607 0 608 }; 609 610 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY); 611 MODULE_VERSION(if_bridge, 1); 612 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1); 613 614 /* 615 * handler for net.link.bridge.ipfw 616 */ 617 static int 618 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS) 619 { 620 int enable = V_pfil_ipfw; 621 int error; 622 623 error = sysctl_handle_int(oidp, &enable, 0, req); 624 enable &= 1; 625 626 if (enable != V_pfil_ipfw) { 627 V_pfil_ipfw = enable; 628 629 /* 630 * Disable pfil so that ipfw doesnt run twice, if the user 631 * really wants both then they can re-enable pfil_bridge and/or 632 * pfil_member. Also allow non-ip packets as ipfw can filter by 633 * layer2 type. 634 */ 635 if (V_pfil_ipfw) { 636 V_pfil_onlyip = 0; 637 V_pfil_bridge = 0; 638 V_pfil_member = 0; 639 } 640 } 641 642 return (error); 643 } 644 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, 645 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT, 646 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I", 647 "Layer2 filter with IPFW"); 648 649 /* 650 * bridge_clone_create: 651 * 652 * Create a new bridge instance. 653 */ 654 static int 655 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params) 656 { 657 struct bridge_softc *sc; 658 struct ifnet *ifp; 659 660 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO); 661 ifp = sc->sc_ifp = if_alloc(IFT_ETHER); 662 if (ifp == NULL) { 663 free(sc, M_DEVBUF); 664 return (ENOSPC); 665 } 666 667 BRIDGE_LOCK_INIT(sc); 668 sc->sc_brtmax = BRIDGE_RTABLE_MAX; 669 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT; 670 671 /* Initialize our routing table. */ 672 bridge_rtable_init(sc); 673 674 callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0); 675 676 CK_LIST_INIT(&sc->sc_iflist); 677 CK_LIST_INIT(&sc->sc_spanlist); 678 679 ifp->if_softc = sc; 680 if_initname(ifp, bridge_name, unit); 681 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; 682 ifp->if_ioctl = bridge_ioctl; 683 ifp->if_transmit = bridge_transmit; 684 ifp->if_qflush = bridge_qflush; 685 ifp->if_init = bridge_init; 686 ifp->if_type = IFT_BRIDGE; 687 688 ether_gen_addr(ifp, &sc->sc_defaddr); 689 690 bstp_attach(&sc->sc_stp, &bridge_ops); 691 ether_ifattach(ifp, sc->sc_defaddr.octet); 692 /* Now undo some of the damage... */ 693 ifp->if_baudrate = 0; 694 ifp->if_type = IFT_BRIDGE; 695 696 BRIDGE_LIST_LOCK(); 697 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list); 698 BRIDGE_LIST_UNLOCK(); 699 700 return (0); 701 } 702 703 static void 704 bridge_clone_destroy_cb(struct epoch_context *ctx) 705 { 706 struct bridge_softc *sc; 707 708 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx); 709 710 BRIDGE_LOCK_DESTROY(sc); 711 free(sc, M_DEVBUF); 712 } 713 714 /* 715 * bridge_clone_destroy: 716 * 717 * Destroy a bridge instance. 718 */ 719 static void 720 bridge_clone_destroy(struct ifnet *ifp) 721 { 722 struct bridge_softc *sc = ifp->if_softc; 723 struct bridge_iflist *bif; 724 struct epoch_tracker et; 725 726 NET_EPOCH_ENTER(et); 727 BRIDGE_LOCK(sc); 728 729 bridge_stop(ifp, 1); 730 ifp->if_flags &= ~IFF_UP; 731 732 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL) 733 bridge_delete_member(sc, bif, 0); 734 735 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) { 736 bridge_delete_span(sc, bif); 737 } 738 739 /* Tear down the routing table. */ 740 bridge_rtable_fini(sc); 741 742 BRIDGE_UNLOCK(sc); 743 744 callout_drain(&sc->sc_brcallout); 745 746 BRIDGE_LIST_LOCK(); 747 LIST_REMOVE(sc, sc_list); 748 BRIDGE_LIST_UNLOCK(); 749 750 bstp_detach(&sc->sc_stp); 751 NET_EPOCH_EXIT(et); 752 753 ether_ifdetach(ifp); 754 if_free(ifp); 755 756 NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx); 757 } 758 759 /* 760 * bridge_ioctl: 761 * 762 * Handle a control request from the operator. 763 */ 764 static int 765 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data) 766 { 767 struct bridge_softc *sc = ifp->if_softc; 768 struct ifreq *ifr = (struct ifreq *)data; 769 struct bridge_iflist *bif; 770 struct thread *td = curthread; 771 union { 772 struct ifbreq ifbreq; 773 struct ifbifconf ifbifconf; 774 struct ifbareq ifbareq; 775 struct ifbaconf ifbaconf; 776 struct ifbrparam ifbrparam; 777 struct ifbropreq ifbropreq; 778 } args; 779 struct ifdrv *ifd = (struct ifdrv *) data; 780 const struct bridge_control *bc; 781 int error = 0, oldmtu; 782 struct epoch_tracker et; 783 784 NET_EPOCH_ENTER(et); 785 786 switch (cmd) { 787 788 case SIOCADDMULTI: 789 case SIOCDELMULTI: 790 break; 791 792 case SIOCGDRVSPEC: 793 case SIOCSDRVSPEC: 794 if (ifd->ifd_cmd >= bridge_control_table_size) { 795 error = EINVAL; 796 break; 797 } 798 bc = &bridge_control_table[ifd->ifd_cmd]; 799 800 if (cmd == SIOCGDRVSPEC && 801 (bc->bc_flags & BC_F_COPYOUT) == 0) { 802 error = EINVAL; 803 break; 804 } 805 else if (cmd == SIOCSDRVSPEC && 806 (bc->bc_flags & BC_F_COPYOUT) != 0) { 807 error = EINVAL; 808 break; 809 } 810 811 if (bc->bc_flags & BC_F_SUSER) { 812 error = priv_check(td, PRIV_NET_BRIDGE); 813 if (error) 814 break; 815 } 816 817 if (ifd->ifd_len != bc->bc_argsize || 818 ifd->ifd_len > sizeof(args)) { 819 error = EINVAL; 820 break; 821 } 822 823 bzero(&args, sizeof(args)); 824 if (bc->bc_flags & BC_F_COPYIN) { 825 error = copyin(ifd->ifd_data, &args, ifd->ifd_len); 826 if (error) 827 break; 828 } 829 830 oldmtu = ifp->if_mtu; 831 BRIDGE_LOCK(sc); 832 error = (*bc->bc_func)(sc, &args); 833 BRIDGE_UNLOCK(sc); 834 if (error) 835 break; 836 837 /* 838 * Bridge MTU may change during addition of the first port. 839 * If it did, do network layer specific procedure. 840 */ 841 if (ifp->if_mtu != oldmtu) { 842 #ifdef INET6 843 nd6_setmtu(ifp); 844 #endif 845 rt_updatemtu(ifp); 846 } 847 848 if (bc->bc_flags & BC_F_COPYOUT) 849 error = copyout(&args, ifd->ifd_data, ifd->ifd_len); 850 851 break; 852 853 case SIOCSIFFLAGS: 854 if (!(ifp->if_flags & IFF_UP) && 855 (ifp->if_drv_flags & IFF_DRV_RUNNING)) { 856 /* 857 * If interface is marked down and it is running, 858 * then stop and disable it. 859 */ 860 BRIDGE_LOCK(sc); 861 bridge_stop(ifp, 1); 862 BRIDGE_UNLOCK(sc); 863 } else if ((ifp->if_flags & IFF_UP) && 864 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) { 865 /* 866 * If interface is marked up and it is stopped, then 867 * start it. 868 */ 869 (*ifp->if_init)(sc); 870 } 871 break; 872 873 case SIOCSIFMTU: 874 if (ifr->ifr_mtu < 576) { 875 error = EINVAL; 876 break; 877 } 878 if (CK_LIST_EMPTY(&sc->sc_iflist)) { 879 sc->sc_ifp->if_mtu = ifr->ifr_mtu; 880 break; 881 } 882 BRIDGE_LOCK(sc); 883 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 884 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) { 885 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)" 886 " != %d\n", sc->sc_ifp->if_xname, 887 bif->bif_ifp->if_mtu, 888 bif->bif_ifp->if_xname, ifr->ifr_mtu); 889 error = EINVAL; 890 break; 891 } 892 } 893 if (!error) 894 sc->sc_ifp->if_mtu = ifr->ifr_mtu; 895 BRIDGE_UNLOCK(sc); 896 break; 897 default: 898 /* 899 * drop the lock as ether_ioctl() will call bridge_start() and 900 * cause the lock to be recursed. 901 */ 902 error = ether_ioctl(ifp, cmd, data); 903 break; 904 } 905 906 NET_EPOCH_EXIT(et); 907 908 return (error); 909 } 910 911 /* 912 * bridge_mutecaps: 913 * 914 * Clear or restore unwanted capabilities on the member interface 915 */ 916 static void 917 bridge_mutecaps(struct bridge_softc *sc) 918 { 919 struct bridge_iflist *bif; 920 int enabled, mask; 921 922 BRIDGE_LOCK_ASSERT(sc); 923 924 /* Initial bitmask of capabilities to test */ 925 mask = BRIDGE_IFCAPS_MASK; 926 927 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 928 /* Every member must support it or its disabled */ 929 mask &= bif->bif_savedcaps; 930 } 931 932 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 933 enabled = bif->bif_ifp->if_capenable; 934 enabled &= ~BRIDGE_IFCAPS_STRIP; 935 /* strip off mask bits and enable them again if allowed */ 936 enabled &= ~BRIDGE_IFCAPS_MASK; 937 enabled |= mask; 938 BRIDGE_UNLOCK(sc); 939 bridge_set_ifcap(sc, bif, enabled); 940 BRIDGE_LOCK(sc); 941 } 942 } 943 944 static void 945 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set) 946 { 947 struct ifnet *ifp = bif->bif_ifp; 948 struct ifreq ifr; 949 int error, mask, stuck; 950 951 BRIDGE_UNLOCK_ASSERT(sc); 952 953 bzero(&ifr, sizeof(ifr)); 954 ifr.ifr_reqcap = set; 955 956 if (ifp->if_capenable != set) { 957 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr); 958 if (error) 959 if_printf(sc->sc_ifp, 960 "error setting capabilities on %s: %d\n", 961 ifp->if_xname, error); 962 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP; 963 stuck = ifp->if_capenable & mask & ~set; 964 if (stuck != 0) 965 if_printf(sc->sc_ifp, 966 "can't disable some capabilities on %s: 0x%x\n", 967 ifp->if_xname, stuck); 968 } 969 } 970 971 /* 972 * bridge_lookup_member: 973 * 974 * Lookup a bridge member interface. 975 */ 976 static struct bridge_iflist * 977 bridge_lookup_member(struct bridge_softc *sc, const char *name) 978 { 979 struct bridge_iflist *bif; 980 struct ifnet *ifp; 981 982 NET_EPOCH_ASSERT(); 983 984 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 985 ifp = bif->bif_ifp; 986 if (strcmp(ifp->if_xname, name) == 0) 987 return (bif); 988 } 989 990 return (NULL); 991 } 992 993 /* 994 * bridge_lookup_member_if: 995 * 996 * Lookup a bridge member interface by ifnet*. 997 */ 998 static struct bridge_iflist * 999 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp) 1000 { 1001 struct bridge_iflist *bif; 1002 1003 NET_EPOCH_ASSERT(); 1004 1005 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1006 if (bif->bif_ifp == member_ifp) 1007 return (bif); 1008 } 1009 1010 return (NULL); 1011 } 1012 1013 static void 1014 bridge_delete_member_cb(struct epoch_context *ctx) 1015 { 1016 struct bridge_iflist *bif; 1017 1018 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx); 1019 1020 free(bif, M_DEVBUF); 1021 } 1022 1023 /* 1024 * bridge_delete_member: 1025 * 1026 * Delete the specified member interface. 1027 */ 1028 static void 1029 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif, 1030 int gone) 1031 { 1032 struct ifnet *ifs = bif->bif_ifp; 1033 struct ifnet *fif = NULL; 1034 struct bridge_iflist *bifl; 1035 1036 BRIDGE_LOCK_ASSERT(sc); 1037 1038 if (bif->bif_flags & IFBIF_STP) 1039 bstp_disable(&bif->bif_stp); 1040 1041 ifs->if_bridge = NULL; 1042 CK_LIST_REMOVE(bif, bif_next); 1043 1044 /* 1045 * If removing the interface that gave the bridge its mac address, set 1046 * the mac address of the bridge to the address of the next member, or 1047 * to its default address if no members are left. 1048 */ 1049 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) { 1050 if (CK_LIST_EMPTY(&sc->sc_iflist)) { 1051 bcopy(&sc->sc_defaddr, 1052 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1053 sc->sc_ifaddr = NULL; 1054 } else { 1055 bifl = CK_LIST_FIRST(&sc->sc_iflist); 1056 fif = bifl->bif_ifp; 1057 bcopy(IF_LLADDR(fif), 1058 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1059 sc->sc_ifaddr = fif; 1060 } 1061 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp); 1062 } 1063 1064 bridge_linkcheck(sc); 1065 bridge_mutecaps(sc); /* recalcuate now this interface is removed */ 1066 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL); 1067 KASSERT(bif->bif_addrcnt == 0, 1068 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt)); 1069 1070 ifs->if_bridge_output = NULL; 1071 ifs->if_bridge_input = NULL; 1072 ifs->if_bridge_linkstate = NULL; 1073 BRIDGE_UNLOCK(sc); 1074 if (!gone) { 1075 switch (ifs->if_type) { 1076 case IFT_ETHER: 1077 case IFT_L2VLAN: 1078 /* 1079 * Take the interface out of promiscuous mode, but only 1080 * if it was promiscuous in the first place. It might 1081 * not be if we're in the bridge_ioctl_add() error path. 1082 */ 1083 if (ifs->if_flags & IFF_PROMISC) 1084 (void) ifpromisc(ifs, 0); 1085 break; 1086 1087 case IFT_GIF: 1088 break; 1089 1090 default: 1091 #ifdef DIAGNOSTIC 1092 panic("bridge_delete_member: impossible"); 1093 #endif 1094 break; 1095 } 1096 /* reneable any interface capabilities */ 1097 bridge_set_ifcap(sc, bif, bif->bif_savedcaps); 1098 } 1099 bstp_destroy(&bif->bif_stp); /* prepare to free */ 1100 BRIDGE_LOCK(sc); 1101 1102 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx); 1103 } 1104 1105 /* 1106 * bridge_delete_span: 1107 * 1108 * Delete the specified span interface. 1109 */ 1110 static void 1111 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif) 1112 { 1113 BRIDGE_LOCK_ASSERT(sc); 1114 1115 KASSERT(bif->bif_ifp->if_bridge == NULL, 1116 ("%s: not a span interface", __func__)); 1117 1118 CK_LIST_REMOVE(bif, bif_next); 1119 1120 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx); 1121 } 1122 1123 static int 1124 bridge_ioctl_add(struct bridge_softc *sc, void *arg) 1125 { 1126 struct ifbreq *req = arg; 1127 struct bridge_iflist *bif = NULL; 1128 struct ifnet *ifs; 1129 int error = 0; 1130 1131 ifs = ifunit(req->ifbr_ifsname); 1132 if (ifs == NULL) 1133 return (ENOENT); 1134 if (ifs->if_ioctl == NULL) /* must be supported */ 1135 return (EINVAL); 1136 1137 /* If it's in the span list, it can't be a member. */ 1138 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1139 if (ifs == bif->bif_ifp) 1140 return (EBUSY); 1141 1142 if (ifs->if_bridge == sc) 1143 return (EEXIST); 1144 1145 if (ifs->if_bridge != NULL) 1146 return (EBUSY); 1147 1148 switch (ifs->if_type) { 1149 case IFT_ETHER: 1150 case IFT_L2VLAN: 1151 case IFT_GIF: 1152 /* permitted interface types */ 1153 break; 1154 default: 1155 return (EINVAL); 1156 } 1157 1158 #ifdef INET6 1159 /* 1160 * Two valid inet6 addresses with link-local scope must not be 1161 * on the parent interface and the member interfaces at the 1162 * same time. This restriction is needed to prevent violation 1163 * of link-local scope zone. Attempts to add a member 1164 * interface which has inet6 addresses when the parent has 1165 * inet6 triggers removal of all inet6 addresses on the member 1166 * interface. 1167 */ 1168 1169 /* Check if the parent interface has a link-local scope addr. */ 1170 if (V_allow_llz_overlap == 0 && 1171 in6ifa_llaonifp(sc->sc_ifp) != NULL) { 1172 /* 1173 * If any, remove all inet6 addresses from the member 1174 * interfaces. 1175 */ 1176 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1177 if (in6ifa_llaonifp(bif->bif_ifp)) { 1178 BRIDGE_UNLOCK(sc); 1179 in6_ifdetach(bif->bif_ifp); 1180 BRIDGE_LOCK(sc); 1181 if_printf(sc->sc_ifp, 1182 "IPv6 addresses on %s have been removed " 1183 "before adding it as a member to prevent " 1184 "IPv6 address scope violation.\n", 1185 bif->bif_ifp->if_xname); 1186 } 1187 } 1188 if (in6ifa_llaonifp(ifs)) { 1189 BRIDGE_UNLOCK(sc); 1190 in6_ifdetach(ifs); 1191 BRIDGE_LOCK(sc); 1192 if_printf(sc->sc_ifp, 1193 "IPv6 addresses on %s have been removed " 1194 "before adding it as a member to prevent " 1195 "IPv6 address scope violation.\n", 1196 ifs->if_xname); 1197 } 1198 } 1199 #endif 1200 /* Allow the first Ethernet member to define the MTU */ 1201 if (CK_LIST_EMPTY(&sc->sc_iflist)) 1202 sc->sc_ifp->if_mtu = ifs->if_mtu; 1203 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) { 1204 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n", 1205 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu); 1206 return (EINVAL); 1207 } 1208 1209 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO); 1210 if (bif == NULL) 1211 return (ENOMEM); 1212 1213 bif->bif_ifp = ifs; 1214 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER; 1215 bif->bif_savedcaps = ifs->if_capenable; 1216 1217 /* 1218 * Assign the interface's MAC address to the bridge if it's the first 1219 * member and the MAC address of the bridge has not been changed from 1220 * the default randomly generated one. 1221 */ 1222 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) && 1223 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) { 1224 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1225 sc->sc_ifaddr = ifs; 1226 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp); 1227 } 1228 1229 ifs->if_bridge = sc; 1230 ifs->if_bridge_output = bridge_output; 1231 ifs->if_bridge_input = bridge_input; 1232 ifs->if_bridge_linkstate = bridge_linkstate; 1233 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp); 1234 /* 1235 * XXX: XLOCK HERE!?! 1236 * 1237 * NOTE: insert_***HEAD*** should be safe for the traversals. 1238 */ 1239 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next); 1240 1241 /* Set interface capabilities to the intersection set of all members */ 1242 bridge_mutecaps(sc); 1243 bridge_linkcheck(sc); 1244 1245 /* Place the interface into promiscuous mode */ 1246 switch (ifs->if_type) { 1247 case IFT_ETHER: 1248 case IFT_L2VLAN: 1249 BRIDGE_UNLOCK(sc); 1250 error = ifpromisc(ifs, 1); 1251 BRIDGE_LOCK(sc); 1252 break; 1253 } 1254 1255 if (error) 1256 bridge_delete_member(sc, bif, 0); 1257 return (error); 1258 } 1259 1260 static int 1261 bridge_ioctl_del(struct bridge_softc *sc, void *arg) 1262 { 1263 struct ifbreq *req = arg; 1264 struct bridge_iflist *bif; 1265 1266 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1267 if (bif == NULL) 1268 return (ENOENT); 1269 1270 bridge_delete_member(sc, bif, 0); 1271 1272 return (0); 1273 } 1274 1275 static int 1276 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg) 1277 { 1278 struct ifbreq *req = arg; 1279 struct bridge_iflist *bif; 1280 struct bstp_port *bp; 1281 1282 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1283 if (bif == NULL) 1284 return (ENOENT); 1285 1286 bp = &bif->bif_stp; 1287 req->ifbr_ifsflags = bif->bif_flags; 1288 req->ifbr_state = bp->bp_state; 1289 req->ifbr_priority = bp->bp_priority; 1290 req->ifbr_path_cost = bp->bp_path_cost; 1291 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff; 1292 req->ifbr_proto = bp->bp_protover; 1293 req->ifbr_role = bp->bp_role; 1294 req->ifbr_stpflags = bp->bp_flags; 1295 req->ifbr_addrcnt = bif->bif_addrcnt; 1296 req->ifbr_addrmax = bif->bif_addrmax; 1297 req->ifbr_addrexceeded = bif->bif_addrexceeded; 1298 1299 /* Copy STP state options as flags */ 1300 if (bp->bp_operedge) 1301 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE; 1302 if (bp->bp_flags & BSTP_PORT_AUTOEDGE) 1303 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE; 1304 if (bp->bp_ptp_link) 1305 req->ifbr_ifsflags |= IFBIF_BSTP_PTP; 1306 if (bp->bp_flags & BSTP_PORT_AUTOPTP) 1307 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP; 1308 if (bp->bp_flags & BSTP_PORT_ADMEDGE) 1309 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE; 1310 if (bp->bp_flags & BSTP_PORT_ADMCOST) 1311 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST; 1312 return (0); 1313 } 1314 1315 static int 1316 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg) 1317 { 1318 struct ifbreq *req = arg; 1319 struct bridge_iflist *bif; 1320 struct bstp_port *bp; 1321 int error; 1322 1323 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1324 if (bif == NULL) 1325 return (ENOENT); 1326 bp = &bif->bif_stp; 1327 1328 if (req->ifbr_ifsflags & IFBIF_SPAN) 1329 /* SPAN is readonly */ 1330 return (EINVAL); 1331 1332 if (req->ifbr_ifsflags & IFBIF_STP) { 1333 if ((bif->bif_flags & IFBIF_STP) == 0) { 1334 error = bstp_enable(&bif->bif_stp); 1335 if (error) 1336 return (error); 1337 } 1338 } else { 1339 if ((bif->bif_flags & IFBIF_STP) != 0) 1340 bstp_disable(&bif->bif_stp); 1341 } 1342 1343 /* Pass on STP flags */ 1344 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0); 1345 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0); 1346 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0); 1347 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0); 1348 1349 /* Save the bits relating to the bridge */ 1350 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK; 1351 1352 return (0); 1353 } 1354 1355 static int 1356 bridge_ioctl_scache(struct bridge_softc *sc, void *arg) 1357 { 1358 struct ifbrparam *param = arg; 1359 1360 sc->sc_brtmax = param->ifbrp_csize; 1361 bridge_rttrim(sc); 1362 1363 return (0); 1364 } 1365 1366 static int 1367 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg) 1368 { 1369 struct ifbrparam *param = arg; 1370 1371 param->ifbrp_csize = sc->sc_brtmax; 1372 1373 return (0); 1374 } 1375 1376 static int 1377 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg) 1378 { 1379 struct ifbifconf *bifc = arg; 1380 struct bridge_iflist *bif; 1381 struct ifbreq breq; 1382 char *buf, *outbuf; 1383 int count, buflen, len, error = 0; 1384 1385 count = 0; 1386 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) 1387 count++; 1388 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1389 count++; 1390 1391 buflen = sizeof(breq) * count; 1392 if (bifc->ifbic_len == 0) { 1393 bifc->ifbic_len = buflen; 1394 return (0); 1395 } 1396 BRIDGE_UNLOCK(sc); 1397 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO); 1398 BRIDGE_LOCK(sc); 1399 1400 count = 0; 1401 buf = outbuf; 1402 len = min(bifc->ifbic_len, buflen); 1403 bzero(&breq, sizeof(breq)); 1404 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1405 if (len < sizeof(breq)) 1406 break; 1407 1408 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname, 1409 sizeof(breq.ifbr_ifsname)); 1410 /* Fill in the ifbreq structure */ 1411 error = bridge_ioctl_gifflags(sc, &breq); 1412 if (error) 1413 break; 1414 memcpy(buf, &breq, sizeof(breq)); 1415 count++; 1416 buf += sizeof(breq); 1417 len -= sizeof(breq); 1418 } 1419 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) { 1420 if (len < sizeof(breq)) 1421 break; 1422 1423 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname, 1424 sizeof(breq.ifbr_ifsname)); 1425 breq.ifbr_ifsflags = bif->bif_flags; 1426 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff; 1427 memcpy(buf, &breq, sizeof(breq)); 1428 count++; 1429 buf += sizeof(breq); 1430 len -= sizeof(breq); 1431 } 1432 1433 BRIDGE_UNLOCK(sc); 1434 bifc->ifbic_len = sizeof(breq) * count; 1435 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len); 1436 BRIDGE_LOCK(sc); 1437 free(outbuf, M_TEMP); 1438 return (error); 1439 } 1440 1441 static int 1442 bridge_ioctl_rts(struct bridge_softc *sc, void *arg) 1443 { 1444 struct ifbaconf *bac = arg; 1445 struct bridge_rtnode *brt; 1446 struct ifbareq bareq; 1447 char *buf, *outbuf; 1448 int count, buflen, len, error = 0; 1449 1450 if (bac->ifbac_len == 0) 1451 return (0); 1452 1453 count = 0; 1454 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) 1455 count++; 1456 buflen = sizeof(bareq) * count; 1457 1458 BRIDGE_UNLOCK(sc); 1459 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO); 1460 BRIDGE_LOCK(sc); 1461 1462 count = 0; 1463 buf = outbuf; 1464 len = min(bac->ifbac_len, buflen); 1465 bzero(&bareq, sizeof(bareq)); 1466 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) { 1467 if (len < sizeof(bareq)) 1468 goto out; 1469 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname, 1470 sizeof(bareq.ifba_ifsname)); 1471 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr)); 1472 bareq.ifba_vlan = brt->brt_vlan; 1473 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 1474 time_uptime < brt->brt_expire) 1475 bareq.ifba_expire = brt->brt_expire - time_uptime; 1476 else 1477 bareq.ifba_expire = 0; 1478 bareq.ifba_flags = brt->brt_flags; 1479 1480 memcpy(buf, &bareq, sizeof(bareq)); 1481 count++; 1482 buf += sizeof(bareq); 1483 len -= sizeof(bareq); 1484 } 1485 out: 1486 BRIDGE_UNLOCK(sc); 1487 bac->ifbac_len = sizeof(bareq) * count; 1488 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len); 1489 BRIDGE_LOCK(sc); 1490 free(outbuf, M_TEMP); 1491 return (error); 1492 } 1493 1494 static int 1495 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg) 1496 { 1497 struct ifbareq *req = arg; 1498 struct bridge_iflist *bif; 1499 int error; 1500 1501 NET_EPOCH_ASSERT(); 1502 1503 bif = bridge_lookup_member(sc, req->ifba_ifsname); 1504 if (bif == NULL) 1505 return (ENOENT); 1506 1507 /* bridge_rtupdate() may acquire the lock. */ 1508 BRIDGE_UNLOCK(sc); 1509 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1, 1510 req->ifba_flags); 1511 BRIDGE_LOCK(sc); 1512 1513 return (error); 1514 } 1515 1516 static int 1517 bridge_ioctl_sto(struct bridge_softc *sc, void *arg) 1518 { 1519 struct ifbrparam *param = arg; 1520 1521 sc->sc_brttimeout = param->ifbrp_ctime; 1522 return (0); 1523 } 1524 1525 static int 1526 bridge_ioctl_gto(struct bridge_softc *sc, void *arg) 1527 { 1528 struct ifbrparam *param = arg; 1529 1530 param->ifbrp_ctime = sc->sc_brttimeout; 1531 return (0); 1532 } 1533 1534 static int 1535 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg) 1536 { 1537 struct ifbareq *req = arg; 1538 1539 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan)); 1540 } 1541 1542 static int 1543 bridge_ioctl_flush(struct bridge_softc *sc, void *arg) 1544 { 1545 struct ifbreq *req = arg; 1546 1547 bridge_rtflush(sc, req->ifbr_ifsflags); 1548 return (0); 1549 } 1550 1551 static int 1552 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg) 1553 { 1554 struct ifbrparam *param = arg; 1555 struct bstp_state *bs = &sc->sc_stp; 1556 1557 param->ifbrp_prio = bs->bs_bridge_priority; 1558 return (0); 1559 } 1560 1561 static int 1562 bridge_ioctl_spri(struct bridge_softc *sc, void *arg) 1563 { 1564 struct ifbrparam *param = arg; 1565 1566 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio)); 1567 } 1568 1569 static int 1570 bridge_ioctl_ght(struct bridge_softc *sc, void *arg) 1571 { 1572 struct ifbrparam *param = arg; 1573 struct bstp_state *bs = &sc->sc_stp; 1574 1575 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8; 1576 return (0); 1577 } 1578 1579 static int 1580 bridge_ioctl_sht(struct bridge_softc *sc, void *arg) 1581 { 1582 struct ifbrparam *param = arg; 1583 1584 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime)); 1585 } 1586 1587 static int 1588 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg) 1589 { 1590 struct ifbrparam *param = arg; 1591 struct bstp_state *bs = &sc->sc_stp; 1592 1593 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8; 1594 return (0); 1595 } 1596 1597 static int 1598 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg) 1599 { 1600 struct ifbrparam *param = arg; 1601 1602 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay)); 1603 } 1604 1605 static int 1606 bridge_ioctl_gma(struct bridge_softc *sc, void *arg) 1607 { 1608 struct ifbrparam *param = arg; 1609 struct bstp_state *bs = &sc->sc_stp; 1610 1611 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8; 1612 return (0); 1613 } 1614 1615 static int 1616 bridge_ioctl_sma(struct bridge_softc *sc, void *arg) 1617 { 1618 struct ifbrparam *param = arg; 1619 1620 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage)); 1621 } 1622 1623 static int 1624 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg) 1625 { 1626 struct ifbreq *req = arg; 1627 struct bridge_iflist *bif; 1628 1629 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1630 if (bif == NULL) 1631 return (ENOENT); 1632 1633 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority)); 1634 } 1635 1636 static int 1637 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg) 1638 { 1639 struct ifbreq *req = arg; 1640 struct bridge_iflist *bif; 1641 1642 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1643 if (bif == NULL) 1644 return (ENOENT); 1645 1646 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost)); 1647 } 1648 1649 static int 1650 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg) 1651 { 1652 struct ifbreq *req = arg; 1653 struct bridge_iflist *bif; 1654 1655 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1656 if (bif == NULL) 1657 return (ENOENT); 1658 1659 bif->bif_addrmax = req->ifbr_addrmax; 1660 return (0); 1661 } 1662 1663 static int 1664 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg) 1665 { 1666 struct ifbreq *req = arg; 1667 struct bridge_iflist *bif = NULL; 1668 struct ifnet *ifs; 1669 1670 ifs = ifunit(req->ifbr_ifsname); 1671 if (ifs == NULL) 1672 return (ENOENT); 1673 1674 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1675 if (ifs == bif->bif_ifp) 1676 return (EBUSY); 1677 1678 if (ifs->if_bridge != NULL) 1679 return (EBUSY); 1680 1681 switch (ifs->if_type) { 1682 case IFT_ETHER: 1683 case IFT_GIF: 1684 case IFT_L2VLAN: 1685 break; 1686 default: 1687 return (EINVAL); 1688 } 1689 1690 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO); 1691 if (bif == NULL) 1692 return (ENOMEM); 1693 1694 bif->bif_ifp = ifs; 1695 bif->bif_flags = IFBIF_SPAN; 1696 1697 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next); 1698 1699 return (0); 1700 } 1701 1702 static int 1703 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg) 1704 { 1705 struct ifbreq *req = arg; 1706 struct bridge_iflist *bif; 1707 struct ifnet *ifs; 1708 1709 ifs = ifunit(req->ifbr_ifsname); 1710 if (ifs == NULL) 1711 return (ENOENT); 1712 1713 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1714 if (ifs == bif->bif_ifp) 1715 break; 1716 1717 if (bif == NULL) 1718 return (ENOENT); 1719 1720 bridge_delete_span(sc, bif); 1721 1722 return (0); 1723 } 1724 1725 static int 1726 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg) 1727 { 1728 struct ifbropreq *req = arg; 1729 struct bstp_state *bs = &sc->sc_stp; 1730 struct bstp_port *root_port; 1731 1732 req->ifbop_maxage = bs->bs_bridge_max_age >> 8; 1733 req->ifbop_hellotime = bs->bs_bridge_htime >> 8; 1734 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8; 1735 1736 root_port = bs->bs_root_port; 1737 if (root_port == NULL) 1738 req->ifbop_root_port = 0; 1739 else 1740 req->ifbop_root_port = root_port->bp_ifp->if_index; 1741 1742 req->ifbop_holdcount = bs->bs_txholdcount; 1743 req->ifbop_priority = bs->bs_bridge_priority; 1744 req->ifbop_protocol = bs->bs_protover; 1745 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost; 1746 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id; 1747 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id; 1748 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id; 1749 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec; 1750 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec; 1751 1752 return (0); 1753 } 1754 1755 static int 1756 bridge_ioctl_grte(struct bridge_softc *sc, void *arg) 1757 { 1758 struct ifbrparam *param = arg; 1759 1760 param->ifbrp_cexceeded = sc->sc_brtexceeded; 1761 return (0); 1762 } 1763 1764 static int 1765 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg) 1766 { 1767 struct ifbpstpconf *bifstp = arg; 1768 struct bridge_iflist *bif; 1769 struct bstp_port *bp; 1770 struct ifbpstpreq bpreq; 1771 char *buf, *outbuf; 1772 int count, buflen, len, error = 0; 1773 1774 count = 0; 1775 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1776 if ((bif->bif_flags & IFBIF_STP) != 0) 1777 count++; 1778 } 1779 1780 buflen = sizeof(bpreq) * count; 1781 if (bifstp->ifbpstp_len == 0) { 1782 bifstp->ifbpstp_len = buflen; 1783 return (0); 1784 } 1785 1786 BRIDGE_UNLOCK(sc); 1787 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO); 1788 BRIDGE_LOCK(sc); 1789 1790 count = 0; 1791 buf = outbuf; 1792 len = min(bifstp->ifbpstp_len, buflen); 1793 bzero(&bpreq, sizeof(bpreq)); 1794 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1795 if (len < sizeof(bpreq)) 1796 break; 1797 1798 if ((bif->bif_flags & IFBIF_STP) == 0) 1799 continue; 1800 1801 bp = &bif->bif_stp; 1802 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff; 1803 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions; 1804 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost; 1805 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id; 1806 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id; 1807 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id; 1808 1809 memcpy(buf, &bpreq, sizeof(bpreq)); 1810 count++; 1811 buf += sizeof(bpreq); 1812 len -= sizeof(bpreq); 1813 } 1814 1815 BRIDGE_UNLOCK(sc); 1816 bifstp->ifbpstp_len = sizeof(bpreq) * count; 1817 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len); 1818 BRIDGE_LOCK(sc); 1819 free(outbuf, M_TEMP); 1820 return (error); 1821 } 1822 1823 static int 1824 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg) 1825 { 1826 struct ifbrparam *param = arg; 1827 1828 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto)); 1829 } 1830 1831 static int 1832 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg) 1833 { 1834 struct ifbrparam *param = arg; 1835 1836 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc)); 1837 } 1838 1839 /* 1840 * bridge_ifdetach: 1841 * 1842 * Detach an interface from a bridge. Called when a member 1843 * interface is detaching. 1844 */ 1845 static void 1846 bridge_ifdetach(void *arg __unused, struct ifnet *ifp) 1847 { 1848 struct bridge_softc *sc = ifp->if_bridge; 1849 struct bridge_iflist *bif; 1850 struct epoch_tracker et; 1851 1852 if (ifp->if_flags & IFF_RENAMING) 1853 return; 1854 if (V_bridge_cloner == NULL) { 1855 /* 1856 * This detach handler can be called after 1857 * vnet_bridge_uninit(). Just return in that case. 1858 */ 1859 return; 1860 } 1861 NET_EPOCH_ENTER(et); 1862 /* Check if the interface is a bridge member */ 1863 if (sc != NULL) { 1864 BRIDGE_LOCK(sc); 1865 1866 bif = bridge_lookup_member_if(sc, ifp); 1867 if (bif != NULL) 1868 bridge_delete_member(sc, bif, 1); 1869 1870 BRIDGE_UNLOCK(sc); 1871 NET_EPOCH_EXIT(et); 1872 return; 1873 } 1874 1875 /* Check if the interface is a span port */ 1876 BRIDGE_LIST_LOCK(); 1877 LIST_FOREACH(sc, &V_bridge_list, sc_list) { 1878 BRIDGE_LOCK(sc); 1879 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1880 if (ifp == bif->bif_ifp) { 1881 bridge_delete_span(sc, bif); 1882 break; 1883 } 1884 1885 BRIDGE_UNLOCK(sc); 1886 } 1887 BRIDGE_LIST_UNLOCK(); 1888 NET_EPOCH_EXIT(et); 1889 } 1890 1891 /* 1892 * bridge_init: 1893 * 1894 * Initialize a bridge interface. 1895 */ 1896 static void 1897 bridge_init(void *xsc) 1898 { 1899 struct bridge_softc *sc = (struct bridge_softc *)xsc; 1900 struct ifnet *ifp = sc->sc_ifp; 1901 1902 if (ifp->if_drv_flags & IFF_DRV_RUNNING) 1903 return; 1904 1905 BRIDGE_LOCK(sc); 1906 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz, 1907 bridge_timer, sc); 1908 1909 ifp->if_drv_flags |= IFF_DRV_RUNNING; 1910 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */ 1911 1912 BRIDGE_UNLOCK(sc); 1913 } 1914 1915 /* 1916 * bridge_stop: 1917 * 1918 * Stop the bridge interface. 1919 */ 1920 static void 1921 bridge_stop(struct ifnet *ifp, int disable) 1922 { 1923 struct bridge_softc *sc = ifp->if_softc; 1924 1925 NET_EPOCH_ASSERT(); 1926 BRIDGE_LOCK_ASSERT(sc); 1927 1928 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) 1929 return; 1930 1931 callout_stop(&sc->sc_brcallout); 1932 bstp_stop(&sc->sc_stp); 1933 1934 bridge_rtflush(sc, IFBF_FLUSHDYN); 1935 1936 ifp->if_drv_flags &= ~IFF_DRV_RUNNING; 1937 } 1938 1939 /* 1940 * bridge_enqueue: 1941 * 1942 * Enqueue a packet on a bridge member interface. 1943 * 1944 */ 1945 static int 1946 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m) 1947 { 1948 int len, err = 0; 1949 short mflags; 1950 struct mbuf *m0; 1951 1952 /* We may be sending a fragment so traverse the mbuf */ 1953 for (; m; m = m0) { 1954 m0 = m->m_nextpkt; 1955 m->m_nextpkt = NULL; 1956 len = m->m_pkthdr.len; 1957 mflags = m->m_flags; 1958 1959 /* 1960 * If underlying interface can not do VLAN tag insertion itself 1961 * then attach a packet tag that holds it. 1962 */ 1963 if ((m->m_flags & M_VLANTAG) && 1964 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) { 1965 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag); 1966 if (m == NULL) { 1967 if_printf(dst_ifp, 1968 "unable to prepend VLAN header\n"); 1969 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1); 1970 continue; 1971 } 1972 m->m_flags &= ~M_VLANTAG; 1973 } 1974 1975 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */ 1976 if ((err = dst_ifp->if_transmit(dst_ifp, m))) { 1977 m_freem(m0); 1978 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 1979 break; 1980 } 1981 1982 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1); 1983 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len); 1984 if (mflags & M_MCAST) 1985 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1); 1986 } 1987 1988 return (err); 1989 } 1990 1991 /* 1992 * bridge_dummynet: 1993 * 1994 * Receive a queued packet from dummynet and pass it on to the output 1995 * interface. 1996 * 1997 * The mbuf has the Ethernet header already attached. 1998 */ 1999 static void 2000 bridge_dummynet(struct mbuf *m, struct ifnet *ifp) 2001 { 2002 struct bridge_softc *sc; 2003 2004 sc = ifp->if_bridge; 2005 2006 /* 2007 * The packet didnt originate from a member interface. This should only 2008 * ever happen if a member interface is removed while packets are 2009 * queued for it. 2010 */ 2011 if (sc == NULL) { 2012 m_freem(m); 2013 return; 2014 } 2015 2016 if (PFIL_HOOKED_OUT(V_inet_pfil_head) 2017 #ifdef INET6 2018 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2019 #endif 2020 ) { 2021 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0) 2022 return; 2023 if (m == NULL) 2024 return; 2025 } 2026 2027 bridge_enqueue(sc, ifp, m); 2028 } 2029 2030 /* 2031 * bridge_output: 2032 * 2033 * Send output from a bridge member interface. This 2034 * performs the bridging function for locally originated 2035 * packets. 2036 * 2037 * The mbuf has the Ethernet header already attached. We must 2038 * enqueue or free the mbuf before returning. 2039 */ 2040 static int 2041 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa, 2042 struct rtentry *rt) 2043 { 2044 struct ether_header *eh; 2045 struct ifnet *bifp, *dst_if; 2046 struct bridge_softc *sc; 2047 uint16_t vlan; 2048 2049 NET_EPOCH_ASSERT(); 2050 2051 if (m->m_len < ETHER_HDR_LEN) { 2052 m = m_pullup(m, ETHER_HDR_LEN); 2053 if (m == NULL) 2054 return (0); 2055 } 2056 2057 eh = mtod(m, struct ether_header *); 2058 sc = ifp->if_bridge; 2059 vlan = VLANTAGOF(m); 2060 2061 bifp = sc->sc_ifp; 2062 2063 /* 2064 * If bridge is down, but the original output interface is up, 2065 * go ahead and send out that interface. Otherwise, the packet 2066 * is dropped below. 2067 */ 2068 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { 2069 dst_if = ifp; 2070 goto sendunicast; 2071 } 2072 2073 /* 2074 * If the packet is a multicast, or we don't know a better way to 2075 * get there, send to all interfaces. 2076 */ 2077 if (ETHER_IS_MULTICAST(eh->ether_dhost)) 2078 dst_if = NULL; 2079 else 2080 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan); 2081 /* Tap any traffic not passing back out the originating interface */ 2082 if (dst_if != ifp) 2083 ETHER_BPF_MTAP(bifp, m); 2084 if (dst_if == NULL) { 2085 struct bridge_iflist *bif; 2086 struct mbuf *mc; 2087 int used = 0; 2088 2089 bridge_span(sc, m); 2090 2091 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 2092 dst_if = bif->bif_ifp; 2093 2094 if (dst_if->if_type == IFT_GIF) 2095 continue; 2096 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2097 continue; 2098 2099 /* 2100 * If this is not the original output interface, 2101 * and the interface is participating in spanning 2102 * tree, make sure the port is in a state that 2103 * allows forwarding. 2104 */ 2105 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) && 2106 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2107 continue; 2108 2109 if (CK_LIST_NEXT(bif, bif_next) == NULL) { 2110 used = 1; 2111 mc = m; 2112 } else { 2113 mc = m_copypacket(m, M_NOWAIT); 2114 if (mc == NULL) { 2115 if_inc_counter(bifp, IFCOUNTER_OERRORS, 1); 2116 continue; 2117 } 2118 } 2119 2120 bridge_enqueue(sc, dst_if, mc); 2121 } 2122 if (used == 0) 2123 m_freem(m); 2124 return (0); 2125 } 2126 2127 sendunicast: 2128 /* 2129 * XXX Spanning tree consideration here? 2130 */ 2131 2132 bridge_span(sc, m); 2133 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) { 2134 m_freem(m); 2135 return (0); 2136 } 2137 2138 bridge_enqueue(sc, dst_if, m); 2139 return (0); 2140 } 2141 2142 /* 2143 * bridge_transmit: 2144 * 2145 * Do output on a bridge. 2146 * 2147 */ 2148 static int 2149 bridge_transmit(struct ifnet *ifp, struct mbuf *m) 2150 { 2151 struct bridge_softc *sc; 2152 struct ether_header *eh; 2153 struct ifnet *dst_if; 2154 int error = 0; 2155 2156 sc = ifp->if_softc; 2157 2158 ETHER_BPF_MTAP(ifp, m); 2159 2160 eh = mtod(m, struct ether_header *); 2161 2162 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) && 2163 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) { 2164 error = bridge_enqueue(sc, dst_if, m); 2165 } else 2166 bridge_broadcast(sc, ifp, m, 0); 2167 2168 return (error); 2169 } 2170 2171 /* 2172 * The ifp->if_qflush entry point for if_bridge(4) is no-op. 2173 */ 2174 static void 2175 bridge_qflush(struct ifnet *ifp __unused) 2176 { 2177 } 2178 2179 /* 2180 * bridge_forward: 2181 * 2182 * The forwarding function of the bridge. 2183 * 2184 * NOTE: Releases the lock on return. 2185 */ 2186 static void 2187 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif, 2188 struct mbuf *m) 2189 { 2190 struct bridge_iflist *dbif; 2191 struct ifnet *src_if, *dst_if, *ifp; 2192 struct ether_header *eh; 2193 uint16_t vlan; 2194 uint8_t *dst; 2195 int error; 2196 2197 NET_EPOCH_ASSERT(); 2198 2199 src_if = m->m_pkthdr.rcvif; 2200 ifp = sc->sc_ifp; 2201 2202 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1); 2203 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); 2204 vlan = VLANTAGOF(m); 2205 2206 if ((sbif->bif_flags & IFBIF_STP) && 2207 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2208 goto drop; 2209 2210 eh = mtod(m, struct ether_header *); 2211 dst = eh->ether_dhost; 2212 2213 /* If the interface is learning, record the address. */ 2214 if (sbif->bif_flags & IFBIF_LEARNING) { 2215 error = bridge_rtupdate(sc, eh->ether_shost, vlan, 2216 sbif, 0, IFBAF_DYNAMIC); 2217 /* 2218 * If the interface has addresses limits then deny any source 2219 * that is not in the cache. 2220 */ 2221 if (error && sbif->bif_addrmax) 2222 goto drop; 2223 } 2224 2225 if ((sbif->bif_flags & IFBIF_STP) != 0 && 2226 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING) 2227 goto drop; 2228 2229 /* 2230 * At this point, the port either doesn't participate 2231 * in spanning tree or it is in the forwarding state. 2232 */ 2233 2234 /* 2235 * If the packet is unicast, destined for someone on 2236 * "this" side of the bridge, drop it. 2237 */ 2238 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) { 2239 dst_if = bridge_rtlookup(sc, dst, vlan); 2240 if (src_if == dst_if) 2241 goto drop; 2242 } else { 2243 /* 2244 * Check if its a reserved multicast address, any address 2245 * listed in 802.1D section 7.12.6 may not be forwarded by the 2246 * bridge. 2247 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F 2248 */ 2249 if (dst[0] == 0x01 && dst[1] == 0x80 && 2250 dst[2] == 0xc2 && dst[3] == 0x00 && 2251 dst[4] == 0x00 && dst[5] <= 0x0f) 2252 goto drop; 2253 2254 /* ...forward it to all interfaces. */ 2255 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1); 2256 dst_if = NULL; 2257 } 2258 2259 /* 2260 * If we have a destination interface which is a member of our bridge, 2261 * OR this is a unicast packet, push it through the bpf(4) machinery. 2262 * For broadcast or multicast packets, don't bother because it will 2263 * be reinjected into ether_input. We do this before we pass the packets 2264 * through the pfil(9) framework, as it is possible that pfil(9) will 2265 * drop the packet, or possibly modify it, making it difficult to debug 2266 * firewall issues on the bridge. 2267 */ 2268 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0) 2269 ETHER_BPF_MTAP(ifp, m); 2270 2271 /* run the packet filter */ 2272 if (PFIL_HOOKED_IN(V_inet_pfil_head) 2273 #ifdef INET6 2274 || PFIL_HOOKED_IN(V_inet6_pfil_head) 2275 #endif 2276 ) { 2277 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0) 2278 return; 2279 if (m == NULL) 2280 return; 2281 } 2282 2283 if (dst_if == NULL) { 2284 bridge_broadcast(sc, src_if, m, 1); 2285 return; 2286 } 2287 2288 /* 2289 * At this point, we're dealing with a unicast frame 2290 * going to a different interface. 2291 */ 2292 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2293 goto drop; 2294 2295 dbif = bridge_lookup_member_if(sc, dst_if); 2296 if (dbif == NULL) 2297 /* Not a member of the bridge (anymore?) */ 2298 goto drop; 2299 2300 /* Private segments can not talk to each other */ 2301 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE) 2302 goto drop; 2303 2304 if ((dbif->bif_flags & IFBIF_STP) && 2305 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2306 goto drop; 2307 2308 if (PFIL_HOOKED_OUT(V_inet_pfil_head) 2309 #ifdef INET6 2310 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2311 #endif 2312 ) { 2313 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0) 2314 return; 2315 if (m == NULL) 2316 return; 2317 } 2318 2319 bridge_enqueue(sc, dst_if, m); 2320 return; 2321 2322 drop: 2323 m_freem(m); 2324 } 2325 2326 /* 2327 * bridge_input: 2328 * 2329 * Receive input from a member interface. Queue the packet for 2330 * bridging if it is not for us. 2331 */ 2332 static struct mbuf * 2333 bridge_input(struct ifnet *ifp, struct mbuf *m) 2334 { 2335 struct bridge_softc *sc = ifp->if_bridge; 2336 struct bridge_iflist *bif, *bif2; 2337 struct ifnet *bifp; 2338 struct ether_header *eh; 2339 struct mbuf *mc, *mc2; 2340 uint16_t vlan; 2341 int error; 2342 2343 NET_EPOCH_ASSERT(); 2344 2345 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) 2346 return (m); 2347 2348 bifp = sc->sc_ifp; 2349 vlan = VLANTAGOF(m); 2350 2351 /* 2352 * Implement support for bridge monitoring. If this flag has been 2353 * set on this interface, discard the packet once we push it through 2354 * the bpf(4) machinery, but before we do, increment the byte and 2355 * packet counters associated with this interface. 2356 */ 2357 if ((bifp->if_flags & IFF_MONITOR) != 0) { 2358 m->m_pkthdr.rcvif = bifp; 2359 ETHER_BPF_MTAP(bifp, m); 2360 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); 2361 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); 2362 m_freem(m); 2363 return (NULL); 2364 } 2365 bif = bridge_lookup_member_if(sc, ifp); 2366 if (bif == NULL) { 2367 return (m); 2368 } 2369 2370 eh = mtod(m, struct ether_header *); 2371 2372 bridge_span(sc, m); 2373 2374 if (m->m_flags & (M_BCAST|M_MCAST)) { 2375 /* Tap off 802.1D packets; they do not get forwarded. */ 2376 if (memcmp(eh->ether_dhost, bstp_etheraddr, 2377 ETHER_ADDR_LEN) == 0) { 2378 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */ 2379 return (NULL); 2380 } 2381 2382 if ((bif->bif_flags & IFBIF_STP) && 2383 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) { 2384 return (m); 2385 } 2386 2387 /* 2388 * Make a deep copy of the packet and enqueue the copy 2389 * for bridge processing; return the original packet for 2390 * local processing. 2391 */ 2392 mc = m_dup(m, M_NOWAIT); 2393 if (mc == NULL) { 2394 return (m); 2395 } 2396 2397 /* Perform the bridge forwarding function with the copy. */ 2398 bridge_forward(sc, bif, mc); 2399 2400 /* 2401 * Reinject the mbuf as arriving on the bridge so we have a 2402 * chance at claiming multicast packets. We can not loop back 2403 * here from ether_input as a bridge is never a member of a 2404 * bridge. 2405 */ 2406 KASSERT(bifp->if_bridge == NULL, 2407 ("loop created in bridge_input")); 2408 mc2 = m_dup(m, M_NOWAIT); 2409 if (mc2 != NULL) { 2410 /* Keep the layer3 header aligned */ 2411 int i = min(mc2->m_pkthdr.len, max_protohdr); 2412 mc2 = m_copyup(mc2, i, ETHER_ALIGN); 2413 } 2414 if (mc2 != NULL) { 2415 mc2->m_pkthdr.rcvif = bifp; 2416 (*bifp->if_input)(bifp, mc2); 2417 } 2418 2419 /* Return the original packet for local processing. */ 2420 return (m); 2421 } 2422 2423 if ((bif->bif_flags & IFBIF_STP) && 2424 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) { 2425 return (m); 2426 } 2427 2428 #if (defined(INET) || defined(INET6)) 2429 # define OR_CARP_CHECK_WE_ARE_DST(iface) \ 2430 || ((iface)->if_carp \ 2431 && (*carp_forus_p)((iface), eh->ether_dhost)) 2432 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \ 2433 || ((iface)->if_carp \ 2434 && (*carp_forus_p)((iface), eh->ether_shost)) 2435 #else 2436 # define OR_CARP_CHECK_WE_ARE_DST(iface) 2437 # define OR_CARP_CHECK_WE_ARE_SRC(iface) 2438 #endif 2439 2440 #ifdef INET6 2441 # define OR_PFIL_HOOKED_INET6 \ 2442 || PFIL_HOOKED_IN(V_inet6_pfil_head) 2443 #else 2444 # define OR_PFIL_HOOKED_INET6 2445 #endif 2446 2447 #define GRAB_OUR_PACKETS(iface) \ 2448 if ((iface)->if_type == IFT_GIF) \ 2449 continue; \ 2450 /* It is destined for us. */ \ 2451 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \ 2452 OR_CARP_CHECK_WE_ARE_DST((iface)) \ 2453 ) { \ 2454 if (bif->bif_flags & IFBIF_LEARNING) { \ 2455 error = bridge_rtupdate(sc, eh->ether_shost, \ 2456 vlan, bif, 0, IFBAF_DYNAMIC); \ 2457 if (error && bif->bif_addrmax) { \ 2458 m_freem(m); \ 2459 return (NULL); \ 2460 } \ 2461 } \ 2462 m->m_pkthdr.rcvif = iface; \ 2463 if ((iface) == ifp) { \ 2464 /* Skip bridge processing... src == dest */ \ 2465 return (m); \ 2466 } \ 2467 /* It's passing over or to the bridge, locally. */ \ 2468 ETHER_BPF_MTAP(bifp, m); \ 2469 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \ 2470 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \ 2471 /* Filter on the physical interface. */ \ 2472 if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \ 2473 OR_PFIL_HOOKED_INET6)) { \ 2474 if (bridge_pfil(&m, NULL, ifp, \ 2475 PFIL_IN) != 0 || m == NULL) { \ 2476 return (NULL); \ 2477 } \ 2478 } \ 2479 if ((iface) != bifp) \ 2480 ETHER_BPF_MTAP(iface, m); \ 2481 return (m); \ 2482 } \ 2483 \ 2484 /* We just received a packet that we sent out. */ \ 2485 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \ 2486 OR_CARP_CHECK_WE_ARE_SRC((iface)) \ 2487 ) { \ 2488 m_freem(m); \ 2489 return (NULL); \ 2490 } 2491 2492 /* 2493 * Unicast. Make sure it's not for the bridge. 2494 */ 2495 do { GRAB_OUR_PACKETS(bifp) } while (0); 2496 2497 /* 2498 * Give a chance for ifp at first priority. This will help when the 2499 * packet comes through the interface like VLAN's with the same MACs 2500 * on several interfaces from the same bridge. This also will save 2501 * some CPU cycles in case the destination interface and the input 2502 * interface (eq ifp) are the same. 2503 */ 2504 do { GRAB_OUR_PACKETS(ifp) } while (0); 2505 2506 /* Now check the all bridge members. */ 2507 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) { 2508 GRAB_OUR_PACKETS(bif2->bif_ifp) 2509 } 2510 2511 #undef OR_CARP_CHECK_WE_ARE_DST 2512 #undef OR_CARP_CHECK_WE_ARE_SRC 2513 #undef OR_PFIL_HOOKED_INET6 2514 #undef GRAB_OUR_PACKETS 2515 2516 /* Perform the bridge forwarding function. */ 2517 bridge_forward(sc, bif, m); 2518 2519 return (NULL); 2520 } 2521 2522 /* 2523 * bridge_broadcast: 2524 * 2525 * Send a frame to all interfaces that are members of 2526 * the bridge, except for the one on which the packet 2527 * arrived. 2528 * 2529 * NOTE: Releases the lock on return. 2530 */ 2531 static void 2532 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if, 2533 struct mbuf *m, int runfilt) 2534 { 2535 struct bridge_iflist *dbif, *sbif; 2536 struct mbuf *mc; 2537 struct ifnet *dst_if; 2538 int used = 0, i; 2539 2540 NET_EPOCH_ASSERT(); 2541 2542 sbif = bridge_lookup_member_if(sc, src_if); 2543 2544 /* Filter on the bridge interface before broadcasting */ 2545 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head) 2546 #ifdef INET6 2547 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2548 #endif 2549 )) { 2550 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0) 2551 return; 2552 if (m == NULL) 2553 return; 2554 } 2555 2556 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) { 2557 dst_if = dbif->bif_ifp; 2558 if (dst_if == src_if) 2559 continue; 2560 2561 /* Private segments can not talk to each other */ 2562 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)) 2563 continue; 2564 2565 if ((dbif->bif_flags & IFBIF_STP) && 2566 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2567 continue; 2568 2569 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 && 2570 (m->m_flags & (M_BCAST|M_MCAST)) == 0) 2571 continue; 2572 2573 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2574 continue; 2575 2576 if (CK_LIST_NEXT(dbif, bif_next) == NULL) { 2577 mc = m; 2578 used = 1; 2579 } else { 2580 mc = m_dup(m, M_NOWAIT); 2581 if (mc == NULL) { 2582 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 2583 continue; 2584 } 2585 } 2586 2587 /* 2588 * Filter on the output interface. Pass a NULL bridge interface 2589 * pointer so we do not redundantly filter on the bridge for 2590 * each interface we broadcast on. 2591 */ 2592 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head) 2593 #ifdef INET6 2594 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2595 #endif 2596 )) { 2597 if (used == 0) { 2598 /* Keep the layer3 header aligned */ 2599 i = min(mc->m_pkthdr.len, max_protohdr); 2600 mc = m_copyup(mc, i, ETHER_ALIGN); 2601 if (mc == NULL) { 2602 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 2603 continue; 2604 } 2605 } 2606 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0) 2607 continue; 2608 if (mc == NULL) 2609 continue; 2610 } 2611 2612 bridge_enqueue(sc, dst_if, mc); 2613 } 2614 if (used == 0) 2615 m_freem(m); 2616 } 2617 2618 /* 2619 * bridge_span: 2620 * 2621 * Duplicate a packet out one or more interfaces that are in span mode, 2622 * the original mbuf is unmodified. 2623 */ 2624 static void 2625 bridge_span(struct bridge_softc *sc, struct mbuf *m) 2626 { 2627 struct bridge_iflist *bif; 2628 struct ifnet *dst_if; 2629 struct mbuf *mc; 2630 2631 NET_EPOCH_ASSERT(); 2632 2633 if (CK_LIST_EMPTY(&sc->sc_spanlist)) 2634 return; 2635 2636 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) { 2637 dst_if = bif->bif_ifp; 2638 2639 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2640 continue; 2641 2642 mc = m_copypacket(m, M_NOWAIT); 2643 if (mc == NULL) { 2644 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 2645 continue; 2646 } 2647 2648 bridge_enqueue(sc, dst_if, mc); 2649 } 2650 } 2651 2652 /* 2653 * bridge_rtupdate: 2654 * 2655 * Add a bridge routing entry. 2656 */ 2657 static int 2658 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan, 2659 struct bridge_iflist *bif, int setflags, uint8_t flags) 2660 { 2661 struct bridge_rtnode *brt; 2662 int error; 2663 2664 NET_EPOCH_ASSERT(); 2665 BRIDGE_UNLOCK_ASSERT(sc); 2666 2667 /* Check the source address is valid and not multicast. */ 2668 if (ETHER_IS_MULTICAST(dst) || 2669 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 && 2670 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0) 2671 return (EINVAL); 2672 2673 /* 802.1p frames map to vlan 1 */ 2674 if (vlan == 0) 2675 vlan = 1; 2676 2677 /* 2678 * A route for this destination might already exist. If so, 2679 * update it, otherwise create a new one. 2680 */ 2681 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) { 2682 BRIDGE_LOCK(sc); 2683 2684 /* Check again, now that we have the lock. There could have 2685 * been a race and we only want to insert this once. */ 2686 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) { 2687 BRIDGE_UNLOCK(sc); 2688 return (0); 2689 } 2690 2691 if (sc->sc_brtcnt >= sc->sc_brtmax) { 2692 sc->sc_brtexceeded++; 2693 BRIDGE_UNLOCK(sc); 2694 return (ENOSPC); 2695 } 2696 /* Check per interface address limits (if enabled) */ 2697 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) { 2698 bif->bif_addrexceeded++; 2699 BRIDGE_UNLOCK(sc); 2700 return (ENOSPC); 2701 } 2702 2703 /* 2704 * Allocate a new bridge forwarding node, and 2705 * initialize the expiration time and Ethernet 2706 * address. 2707 */ 2708 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO); 2709 if (brt == NULL) { 2710 BRIDGE_UNLOCK(sc); 2711 return (ENOMEM); 2712 } 2713 brt->brt_vnet = curvnet; 2714 2715 if (bif->bif_flags & IFBIF_STICKY) 2716 brt->brt_flags = IFBAF_STICKY; 2717 else 2718 brt->brt_flags = IFBAF_DYNAMIC; 2719 2720 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN); 2721 brt->brt_vlan = vlan; 2722 2723 if ((error = bridge_rtnode_insert(sc, brt)) != 0) { 2724 uma_zfree(V_bridge_rtnode_zone, brt); 2725 BRIDGE_UNLOCK(sc); 2726 return (error); 2727 } 2728 brt->brt_dst = bif; 2729 bif->bif_addrcnt++; 2730 2731 BRIDGE_UNLOCK(sc); 2732 } 2733 2734 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 2735 brt->brt_dst != bif) { 2736 BRIDGE_LOCK(sc); 2737 brt->brt_dst->bif_addrcnt--; 2738 brt->brt_dst = bif; 2739 brt->brt_dst->bif_addrcnt++; 2740 BRIDGE_UNLOCK(sc); 2741 } 2742 2743 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 2744 brt->brt_expire = time_uptime + sc->sc_brttimeout; 2745 if (setflags) 2746 brt->brt_flags = flags; 2747 2748 return (0); 2749 } 2750 2751 /* 2752 * bridge_rtlookup: 2753 * 2754 * Lookup the destination interface for an address. 2755 */ 2756 static struct ifnet * 2757 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) 2758 { 2759 struct bridge_rtnode *brt; 2760 2761 NET_EPOCH_ASSERT(); 2762 2763 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL) 2764 return (NULL); 2765 2766 return (brt->brt_ifp); 2767 } 2768 2769 /* 2770 * bridge_rttrim: 2771 * 2772 * Trim the routine table so that we have a number 2773 * of routing entries less than or equal to the 2774 * maximum number. 2775 */ 2776 static void 2777 bridge_rttrim(struct bridge_softc *sc) 2778 { 2779 struct bridge_rtnode *brt, *nbrt; 2780 2781 NET_EPOCH_ASSERT(); 2782 BRIDGE_LOCK_ASSERT(sc); 2783 2784 /* Make sure we actually need to do this. */ 2785 if (sc->sc_brtcnt <= sc->sc_brtmax) 2786 return; 2787 2788 /* Force an aging cycle; this might trim enough addresses. */ 2789 bridge_rtage(sc); 2790 if (sc->sc_brtcnt <= sc->sc_brtmax) 2791 return; 2792 2793 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 2794 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { 2795 bridge_rtnode_destroy(sc, brt); 2796 if (sc->sc_brtcnt <= sc->sc_brtmax) 2797 return; 2798 } 2799 } 2800 } 2801 2802 /* 2803 * bridge_timer: 2804 * 2805 * Aging timer for the bridge. 2806 */ 2807 static void 2808 bridge_timer(void *arg) 2809 { 2810 struct bridge_softc *sc = arg; 2811 struct epoch_tracker et; 2812 2813 NET_EPOCH_ENTER(et); 2814 BRIDGE_LOCK_ASSERT(sc); 2815 2816 /* Destruction of rtnodes requires a proper vnet context */ 2817 CURVNET_SET(sc->sc_ifp->if_vnet); 2818 bridge_rtage(sc); 2819 2820 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) 2821 callout_reset(&sc->sc_brcallout, 2822 bridge_rtable_prune_period * hz, bridge_timer, sc); 2823 CURVNET_RESTORE(); 2824 NET_EPOCH_EXIT(et); 2825 } 2826 2827 /* 2828 * bridge_rtage: 2829 * 2830 * Perform an aging cycle. 2831 */ 2832 static void 2833 bridge_rtage(struct bridge_softc *sc) 2834 { 2835 struct bridge_rtnode *brt, *nbrt; 2836 2837 NET_EPOCH_ASSERT(); 2838 BRIDGE_LOCK_ASSERT(sc); 2839 2840 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 2841 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { 2842 if (time_uptime >= brt->brt_expire) 2843 bridge_rtnode_destroy(sc, brt); 2844 } 2845 } 2846 } 2847 2848 /* 2849 * bridge_rtflush: 2850 * 2851 * Remove all dynamic addresses from the bridge. 2852 */ 2853 static void 2854 bridge_rtflush(struct bridge_softc *sc, int full) 2855 { 2856 struct bridge_rtnode *brt, *nbrt; 2857 2858 NET_EPOCH_ASSERT(); 2859 BRIDGE_LOCK_ASSERT(sc); 2860 2861 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 2862 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 2863 bridge_rtnode_destroy(sc, brt); 2864 } 2865 } 2866 2867 /* 2868 * bridge_rtdaddr: 2869 * 2870 * Remove an address from the table. 2871 */ 2872 static int 2873 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) 2874 { 2875 struct bridge_rtnode *brt; 2876 int found = 0; 2877 2878 NET_EPOCH_ASSERT(); 2879 BRIDGE_LOCK_ASSERT(sc); 2880 2881 /* 2882 * If vlan is zero then we want to delete for all vlans so the lookup 2883 * may return more than one. 2884 */ 2885 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) { 2886 bridge_rtnode_destroy(sc, brt); 2887 found = 1; 2888 } 2889 2890 return (found ? 0 : ENOENT); 2891 } 2892 2893 /* 2894 * bridge_rtdelete: 2895 * 2896 * Delete routes to a speicifc member interface. 2897 */ 2898 static void 2899 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full) 2900 { 2901 struct bridge_rtnode *brt, *nbrt; 2902 2903 NET_EPOCH_ASSERT(); 2904 BRIDGE_LOCK_ASSERT(sc); 2905 2906 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 2907 if (brt->brt_ifp == ifp && (full || 2908 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)) 2909 bridge_rtnode_destroy(sc, brt); 2910 } 2911 } 2912 2913 /* 2914 * bridge_rtable_init: 2915 * 2916 * Initialize the route table for this bridge. 2917 */ 2918 static void 2919 bridge_rtable_init(struct bridge_softc *sc) 2920 { 2921 int i; 2922 2923 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE, 2924 M_DEVBUF, M_WAITOK); 2925 2926 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++) 2927 CK_LIST_INIT(&sc->sc_rthash[i]); 2928 2929 sc->sc_rthash_key = arc4random(); 2930 CK_LIST_INIT(&sc->sc_rtlist); 2931 } 2932 2933 /* 2934 * bridge_rtable_fini: 2935 * 2936 * Deconstruct the route table for this bridge. 2937 */ 2938 static void 2939 bridge_rtable_fini(struct bridge_softc *sc) 2940 { 2941 2942 KASSERT(sc->sc_brtcnt == 0, 2943 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt)); 2944 free(sc->sc_rthash, M_DEVBUF); 2945 } 2946 2947 /* 2948 * The following hash function is adapted from "Hash Functions" by Bob Jenkins 2949 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997). 2950 */ 2951 #define mix(a, b, c) \ 2952 do { \ 2953 a -= b; a -= c; a ^= (c >> 13); \ 2954 b -= c; b -= a; b ^= (a << 8); \ 2955 c -= a; c -= b; c ^= (b >> 13); \ 2956 a -= b; a -= c; a ^= (c >> 12); \ 2957 b -= c; b -= a; b ^= (a << 16); \ 2958 c -= a; c -= b; c ^= (b >> 5); \ 2959 a -= b; a -= c; a ^= (c >> 3); \ 2960 b -= c; b -= a; b ^= (a << 10); \ 2961 c -= a; c -= b; c ^= (b >> 15); \ 2962 } while (/*CONSTCOND*/0) 2963 2964 static __inline uint32_t 2965 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr) 2966 { 2967 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key; 2968 2969 b += addr[5] << 8; 2970 b += addr[4]; 2971 a += addr[3] << 24; 2972 a += addr[2] << 16; 2973 a += addr[1] << 8; 2974 a += addr[0]; 2975 2976 mix(a, b, c); 2977 2978 return (c & BRIDGE_RTHASH_MASK); 2979 } 2980 2981 #undef mix 2982 2983 static int 2984 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b) 2985 { 2986 int i, d; 2987 2988 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) { 2989 d = ((int)a[i]) - ((int)b[i]); 2990 } 2991 2992 return (d); 2993 } 2994 2995 /* 2996 * bridge_rtnode_lookup: 2997 * 2998 * Look up a bridge route node for the specified destination. Compare the 2999 * vlan id or if zero then just return the first match. 3000 */ 3001 static struct bridge_rtnode * 3002 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) 3003 { 3004 struct bridge_rtnode *brt; 3005 uint32_t hash; 3006 int dir; 3007 3008 NET_EPOCH_ASSERT(); 3009 3010 hash = bridge_rthash(sc, addr); 3011 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) { 3012 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr); 3013 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0)) 3014 return (brt); 3015 if (dir > 0) 3016 return (NULL); 3017 } 3018 3019 return (NULL); 3020 } 3021 3022 /* 3023 * bridge_rtnode_insert: 3024 * 3025 * Insert the specified bridge node into the route table. We 3026 * assume the entry is not already in the table. 3027 */ 3028 static int 3029 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt) 3030 { 3031 struct bridge_rtnode *lbrt; 3032 uint32_t hash; 3033 int dir; 3034 3035 BRIDGE_LOCK_ASSERT(sc); 3036 3037 hash = bridge_rthash(sc, brt->brt_addr); 3038 3039 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]); 3040 if (lbrt == NULL) { 3041 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash); 3042 goto out; 3043 } 3044 3045 do { 3046 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr); 3047 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan) 3048 return (EEXIST); 3049 if (dir > 0) { 3050 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash); 3051 goto out; 3052 } 3053 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) { 3054 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash); 3055 goto out; 3056 } 3057 lbrt = CK_LIST_NEXT(lbrt, brt_hash); 3058 } while (lbrt != NULL); 3059 3060 #ifdef DIAGNOSTIC 3061 panic("bridge_rtnode_insert: impossible"); 3062 #endif 3063 3064 out: 3065 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list); 3066 sc->sc_brtcnt++; 3067 3068 return (0); 3069 } 3070 3071 static void 3072 bridge_rtnode_destroy_cb(struct epoch_context *ctx) 3073 { 3074 struct bridge_rtnode *brt; 3075 3076 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx); 3077 3078 CURVNET_SET(brt->brt_vnet); 3079 uma_zfree(V_bridge_rtnode_zone, brt); 3080 CURVNET_RESTORE(); 3081 } 3082 3083 /* 3084 * bridge_rtnode_destroy: 3085 * 3086 * Destroy a bridge rtnode. 3087 */ 3088 static void 3089 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt) 3090 { 3091 NET_EPOCH_ASSERT(); 3092 BRIDGE_LOCK_ASSERT(sc); 3093 3094 CK_LIST_REMOVE(brt, brt_hash); 3095 3096 CK_LIST_REMOVE(brt, brt_list); 3097 sc->sc_brtcnt--; 3098 brt->brt_dst->bif_addrcnt--; 3099 3100 NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx); 3101 } 3102 3103 /* 3104 * bridge_rtable_expire: 3105 * 3106 * Set the expiry time for all routes on an interface. 3107 */ 3108 static void 3109 bridge_rtable_expire(struct ifnet *ifp, int age) 3110 { 3111 struct bridge_softc *sc = ifp->if_bridge; 3112 struct bridge_rtnode *brt; 3113 struct epoch_tracker et; 3114 3115 NET_EPOCH_ENTER(et); 3116 CURVNET_SET(ifp->if_vnet); 3117 BRIDGE_LOCK(sc); 3118 3119 /* 3120 * If the age is zero then flush, otherwise set all the expiry times to 3121 * age for the interface 3122 */ 3123 if (age == 0) 3124 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN); 3125 else { 3126 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) { 3127 /* Cap the expiry time to 'age' */ 3128 if (brt->brt_ifp == ifp && 3129 brt->brt_expire > time_uptime + age && 3130 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 3131 brt->brt_expire = time_uptime + age; 3132 } 3133 } 3134 BRIDGE_UNLOCK(sc); 3135 CURVNET_RESTORE(); 3136 NET_EPOCH_EXIT(et); 3137 } 3138 3139 /* 3140 * bridge_state_change: 3141 * 3142 * Callback from the bridgestp code when a port changes states. 3143 */ 3144 static void 3145 bridge_state_change(struct ifnet *ifp, int state) 3146 { 3147 struct bridge_softc *sc = ifp->if_bridge; 3148 static const char *stpstates[] = { 3149 "disabled", 3150 "listening", 3151 "learning", 3152 "forwarding", 3153 "blocking", 3154 "discarding" 3155 }; 3156 3157 CURVNET_SET(ifp->if_vnet); 3158 if (V_log_stp) 3159 log(LOG_NOTICE, "%s: state changed to %s on %s\n", 3160 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname); 3161 CURVNET_RESTORE(); 3162 } 3163 3164 /* 3165 * Send bridge packets through pfil if they are one of the types pfil can deal 3166 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without 3167 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for 3168 * that interface. 3169 */ 3170 static int 3171 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir) 3172 { 3173 int snap, error, i, hlen; 3174 struct ether_header *eh1, eh2; 3175 struct ip *ip; 3176 struct llc llc1; 3177 u_int16_t ether_type; 3178 pfil_return_t rv; 3179 3180 snap = 0; 3181 error = -1; /* Default error if not error == 0 */ 3182 3183 #if 0 3184 /* we may return with the IP fields swapped, ensure its not shared */ 3185 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__)); 3186 #endif 3187 3188 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0) 3189 return (0); /* filtering is disabled */ 3190 3191 i = min((*mp)->m_pkthdr.len, max_protohdr); 3192 if ((*mp)->m_len < i) { 3193 *mp = m_pullup(*mp, i); 3194 if (*mp == NULL) { 3195 printf("%s: m_pullup failed\n", __func__); 3196 return (-1); 3197 } 3198 } 3199 3200 eh1 = mtod(*mp, struct ether_header *); 3201 ether_type = ntohs(eh1->ether_type); 3202 3203 /* 3204 * Check for SNAP/LLC. 3205 */ 3206 if (ether_type < ETHERMTU) { 3207 struct llc *llc2 = (struct llc *)(eh1 + 1); 3208 3209 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 && 3210 llc2->llc_dsap == LLC_SNAP_LSAP && 3211 llc2->llc_ssap == LLC_SNAP_LSAP && 3212 llc2->llc_control == LLC_UI) { 3213 ether_type = htons(llc2->llc_un.type_snap.ether_type); 3214 snap = 1; 3215 } 3216 } 3217 3218 /* 3219 * If we're trying to filter bridge traffic, don't look at anything 3220 * other than IP and ARP traffic. If the filter doesn't understand 3221 * IPv6, don't allow IPv6 through the bridge either. This is lame 3222 * since if we really wanted, say, an AppleTalk filter, we are hosed, 3223 * but of course we don't have an AppleTalk filter to begin with. 3224 * (Note that since pfil doesn't understand ARP it will pass *ALL* 3225 * ARP traffic.) 3226 */ 3227 switch (ether_type) { 3228 case ETHERTYPE_ARP: 3229 case ETHERTYPE_REVARP: 3230 if (V_pfil_ipfw_arp == 0) 3231 return (0); /* Automatically pass */ 3232 break; 3233 3234 case ETHERTYPE_IP: 3235 #ifdef INET6 3236 case ETHERTYPE_IPV6: 3237 #endif /* INET6 */ 3238 break; 3239 default: 3240 /* 3241 * Check to see if the user wants to pass non-ip 3242 * packets, these will not be checked by pfil(9) and 3243 * passed unconditionally so the default is to drop. 3244 */ 3245 if (V_pfil_onlyip) 3246 goto bad; 3247 } 3248 3249 /* Run the packet through pfil before stripping link headers */ 3250 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 && 3251 dir == PFIL_OUT && ifp != NULL) { 3252 switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) { 3253 case PFIL_DROPPED: 3254 return (EACCES); 3255 case PFIL_CONSUMED: 3256 return (0); 3257 } 3258 } 3259 3260 /* Strip off the Ethernet header and keep a copy. */ 3261 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2); 3262 m_adj(*mp, ETHER_HDR_LEN); 3263 3264 /* Strip off snap header, if present */ 3265 if (snap) { 3266 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1); 3267 m_adj(*mp, sizeof(struct llc)); 3268 } 3269 3270 /* 3271 * Check the IP header for alignment and errors 3272 */ 3273 if (dir == PFIL_IN) { 3274 switch (ether_type) { 3275 case ETHERTYPE_IP: 3276 error = bridge_ip_checkbasic(mp); 3277 break; 3278 #ifdef INET6 3279 case ETHERTYPE_IPV6: 3280 error = bridge_ip6_checkbasic(mp); 3281 break; 3282 #endif /* INET6 */ 3283 default: 3284 error = 0; 3285 } 3286 if (error) 3287 goto bad; 3288 } 3289 3290 error = 0; 3291 3292 /* 3293 * Run the packet through pfil 3294 */ 3295 rv = PFIL_PASS; 3296 switch (ether_type) { 3297 case ETHERTYPE_IP: 3298 /* 3299 * Run pfil on the member interface and the bridge, both can 3300 * be skipped by clearing pfil_member or pfil_bridge. 3301 * 3302 * Keep the order: 3303 * in_if -> bridge_if -> out_if 3304 */ 3305 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv = 3306 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) != 3307 PFIL_PASS) 3308 break; 3309 3310 if (V_pfil_member && ifp != NULL && (rv = 3311 pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) != 3312 PFIL_PASS) 3313 break; 3314 3315 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv = 3316 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) != 3317 PFIL_PASS) 3318 break; 3319 3320 /* check if we need to fragment the packet */ 3321 /* bridge_fragment generates a mbuf chain of packets */ 3322 /* that already include eth headers */ 3323 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) { 3324 i = (*mp)->m_pkthdr.len; 3325 if (i > ifp->if_mtu) { 3326 error = bridge_fragment(ifp, mp, &eh2, snap, 3327 &llc1); 3328 return (error); 3329 } 3330 } 3331 3332 /* Recalculate the ip checksum. */ 3333 ip = mtod(*mp, struct ip *); 3334 hlen = ip->ip_hl << 2; 3335 if (hlen < sizeof(struct ip)) 3336 goto bad; 3337 if (hlen > (*mp)->m_len) { 3338 if ((*mp = m_pullup(*mp, hlen)) == NULL) 3339 goto bad; 3340 ip = mtod(*mp, struct ip *); 3341 if (ip == NULL) 3342 goto bad; 3343 } 3344 ip->ip_sum = 0; 3345 if (hlen == sizeof(struct ip)) 3346 ip->ip_sum = in_cksum_hdr(ip); 3347 else 3348 ip->ip_sum = in_cksum(*mp, hlen); 3349 3350 break; 3351 #ifdef INET6 3352 case ETHERTYPE_IPV6: 3353 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv = 3354 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) != 3355 PFIL_PASS) 3356 break; 3357 3358 if (V_pfil_member && ifp != NULL && (rv = 3359 pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) != 3360 PFIL_PASS) 3361 break; 3362 3363 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv = 3364 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) != 3365 PFIL_PASS) 3366 break; 3367 break; 3368 #endif 3369 } 3370 3371 switch (rv) { 3372 case PFIL_CONSUMED: 3373 return (0); 3374 case PFIL_DROPPED: 3375 return (EACCES); 3376 default: 3377 break; 3378 } 3379 3380 error = -1; 3381 3382 /* 3383 * Finally, put everything back the way it was and return 3384 */ 3385 if (snap) { 3386 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT); 3387 if (*mp == NULL) 3388 return (error); 3389 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc)); 3390 } 3391 3392 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT); 3393 if (*mp == NULL) 3394 return (error); 3395 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN); 3396 3397 return (0); 3398 3399 bad: 3400 m_freem(*mp); 3401 *mp = NULL; 3402 return (error); 3403 } 3404 3405 /* 3406 * Perform basic checks on header size since 3407 * pfil assumes ip_input has already processed 3408 * it for it. Cut-and-pasted from ip_input.c. 3409 * Given how simple the IPv6 version is, 3410 * does the IPv4 version really need to be 3411 * this complicated? 3412 * 3413 * XXX Should we update ipstat here, or not? 3414 * XXX Right now we update ipstat but not 3415 * XXX csum_counter. 3416 */ 3417 static int 3418 bridge_ip_checkbasic(struct mbuf **mp) 3419 { 3420 struct mbuf *m = *mp; 3421 struct ip *ip; 3422 int len, hlen; 3423 u_short sum; 3424 3425 if (*mp == NULL) 3426 return (-1); 3427 3428 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 3429 if ((m = m_copyup(m, sizeof(struct ip), 3430 (max_linkhdr + 3) & ~3)) == NULL) { 3431 /* XXXJRT new stat, please */ 3432 KMOD_IPSTAT_INC(ips_toosmall); 3433 goto bad; 3434 } 3435 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 3436 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 3437 KMOD_IPSTAT_INC(ips_toosmall); 3438 goto bad; 3439 } 3440 } 3441 ip = mtod(m, struct ip *); 3442 if (ip == NULL) goto bad; 3443 3444 if (ip->ip_v != IPVERSION) { 3445 KMOD_IPSTAT_INC(ips_badvers); 3446 goto bad; 3447 } 3448 hlen = ip->ip_hl << 2; 3449 if (hlen < sizeof(struct ip)) { /* minimum header length */ 3450 KMOD_IPSTAT_INC(ips_badhlen); 3451 goto bad; 3452 } 3453 if (hlen > m->m_len) { 3454 if ((m = m_pullup(m, hlen)) == NULL) { 3455 KMOD_IPSTAT_INC(ips_badhlen); 3456 goto bad; 3457 } 3458 ip = mtod(m, struct ip *); 3459 if (ip == NULL) goto bad; 3460 } 3461 3462 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) { 3463 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID); 3464 } else { 3465 if (hlen == sizeof(struct ip)) { 3466 sum = in_cksum_hdr(ip); 3467 } else { 3468 sum = in_cksum(m, hlen); 3469 } 3470 } 3471 if (sum) { 3472 KMOD_IPSTAT_INC(ips_badsum); 3473 goto bad; 3474 } 3475 3476 /* Retrieve the packet length. */ 3477 len = ntohs(ip->ip_len); 3478 3479 /* 3480 * Check for additional length bogosity 3481 */ 3482 if (len < hlen) { 3483 KMOD_IPSTAT_INC(ips_badlen); 3484 goto bad; 3485 } 3486 3487 /* 3488 * Check that the amount of data in the buffers 3489 * is as at least much as the IP header would have us expect. 3490 * Drop packet if shorter than we expect. 3491 */ 3492 if (m->m_pkthdr.len < len) { 3493 KMOD_IPSTAT_INC(ips_tooshort); 3494 goto bad; 3495 } 3496 3497 /* Checks out, proceed */ 3498 *mp = m; 3499 return (0); 3500 3501 bad: 3502 *mp = m; 3503 return (-1); 3504 } 3505 3506 #ifdef INET6 3507 /* 3508 * Same as above, but for IPv6. 3509 * Cut-and-pasted from ip6_input.c. 3510 * XXX Should we update ip6stat, or not? 3511 */ 3512 static int 3513 bridge_ip6_checkbasic(struct mbuf **mp) 3514 { 3515 struct mbuf *m = *mp; 3516 struct ip6_hdr *ip6; 3517 3518 /* 3519 * If the IPv6 header is not aligned, slurp it up into a new 3520 * mbuf with space for link headers, in the event we forward 3521 * it. Otherwise, if it is aligned, make sure the entire base 3522 * IPv6 header is in the first mbuf of the chain. 3523 */ 3524 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 3525 struct ifnet *inifp = m->m_pkthdr.rcvif; 3526 if ((m = m_copyup(m, sizeof(struct ip6_hdr), 3527 (max_linkhdr + 3) & ~3)) == NULL) { 3528 /* XXXJRT new stat, please */ 3529 IP6STAT_INC(ip6s_toosmall); 3530 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 3531 goto bad; 3532 } 3533 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) { 3534 struct ifnet *inifp = m->m_pkthdr.rcvif; 3535 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 3536 IP6STAT_INC(ip6s_toosmall); 3537 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 3538 goto bad; 3539 } 3540 } 3541 3542 ip6 = mtod(m, struct ip6_hdr *); 3543 3544 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 3545 IP6STAT_INC(ip6s_badvers); 3546 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 3547 goto bad; 3548 } 3549 3550 /* Checks out, proceed */ 3551 *mp = m; 3552 return (0); 3553 3554 bad: 3555 *mp = m; 3556 return (-1); 3557 } 3558 #endif /* INET6 */ 3559 3560 /* 3561 * bridge_fragment: 3562 * 3563 * Fragment mbuf chain in multiple packets and prepend ethernet header. 3564 */ 3565 static int 3566 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh, 3567 int snap, struct llc *llc) 3568 { 3569 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL; 3570 struct ip *ip; 3571 int error = -1; 3572 3573 if (m->m_len < sizeof(struct ip) && 3574 (m = m_pullup(m, sizeof(struct ip))) == NULL) 3575 goto dropit; 3576 ip = mtod(m, struct ip *); 3577 3578 m->m_pkthdr.csum_flags |= CSUM_IP; 3579 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist); 3580 if (error) 3581 goto dropit; 3582 3583 /* 3584 * Walk the chain and re-add the Ethernet header for 3585 * each mbuf packet. 3586 */ 3587 for (mcur = m; mcur; mcur = mcur->m_nextpkt) { 3588 nextpkt = mcur->m_nextpkt; 3589 mcur->m_nextpkt = NULL; 3590 if (snap) { 3591 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT); 3592 if (mcur == NULL) { 3593 error = ENOBUFS; 3594 if (mprev != NULL) 3595 mprev->m_nextpkt = nextpkt; 3596 goto dropit; 3597 } 3598 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc)); 3599 } 3600 3601 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT); 3602 if (mcur == NULL) { 3603 error = ENOBUFS; 3604 if (mprev != NULL) 3605 mprev->m_nextpkt = nextpkt; 3606 goto dropit; 3607 } 3608 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN); 3609 3610 /* 3611 * The previous two M_PREPEND could have inserted one or two 3612 * mbufs in front so we have to update the previous packet's 3613 * m_nextpkt. 3614 */ 3615 mcur->m_nextpkt = nextpkt; 3616 if (mprev != NULL) 3617 mprev->m_nextpkt = mcur; 3618 else { 3619 /* The first mbuf in the original chain needs to be 3620 * updated. */ 3621 *mp = mcur; 3622 } 3623 mprev = mcur; 3624 } 3625 3626 KMOD_IPSTAT_INC(ips_fragmented); 3627 return (error); 3628 3629 dropit: 3630 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */ 3631 m = mcur->m_nextpkt; 3632 m_freem(mcur); 3633 } 3634 return (error); 3635 } 3636 3637 static void 3638 bridge_linkstate(struct ifnet *ifp) 3639 { 3640 struct bridge_softc *sc = ifp->if_bridge; 3641 struct bridge_iflist *bif; 3642 struct epoch_tracker et; 3643 3644 NET_EPOCH_ENTER(et); 3645 3646 bif = bridge_lookup_member_if(sc, ifp); 3647 if (bif == NULL) { 3648 NET_EPOCH_EXIT(et); 3649 return; 3650 } 3651 bridge_linkcheck(sc); 3652 3653 bstp_linkstate(&bif->bif_stp); 3654 3655 NET_EPOCH_EXIT(et); 3656 } 3657 3658 static void 3659 bridge_linkcheck(struct bridge_softc *sc) 3660 { 3661 struct bridge_iflist *bif; 3662 int new_link, hasls; 3663 3664 NET_EPOCH_ASSERT(); 3665 3666 new_link = LINK_STATE_DOWN; 3667 hasls = 0; 3668 /* Our link is considered up if at least one of our ports is active */ 3669 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 3670 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE) 3671 hasls++; 3672 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) { 3673 new_link = LINK_STATE_UP; 3674 break; 3675 } 3676 } 3677 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) { 3678 /* If no interfaces support link-state then we default to up */ 3679 new_link = LINK_STATE_UP; 3680 } 3681 if_link_state_change(sc->sc_ifp, new_link); 3682 } 3683