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 cv_init(&(_sc)->sc_cv, "if_bridge_cv"); \ 193 } while (0) 194 #define BRIDGE_LOCK_DESTROY(_sc) do { \ 195 mtx_destroy(&(_sc)->sc_mtx); \ 196 cv_destroy(&(_sc)->sc_cv); \ 197 } while (0) 198 #define BRIDGE_LOCK(_sc) mtx_lock(&(_sc)->sc_mtx) 199 #define BRIDGE_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_mtx) 200 #define BRIDGE_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_mtx, MA_OWNED) 201 #define BRIDGE_UNLOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_mtx, MA_NOTOWNED) 202 #define BRIDGE_LOCK2REF(_sc, _err) do { \ 203 mtx_assert(&(_sc)->sc_mtx, MA_OWNED); \ 204 if ((_sc)->sc_iflist_xcnt > 0) \ 205 (_err) = EBUSY; \ 206 else \ 207 (_sc)->sc_iflist_ref++; \ 208 mtx_unlock(&(_sc)->sc_mtx); \ 209 } while (0) 210 #define BRIDGE_UNREF(_sc) do { \ 211 mtx_lock(&(_sc)->sc_mtx); \ 212 (_sc)->sc_iflist_ref--; \ 213 if (((_sc)->sc_iflist_xcnt > 0) && ((_sc)->sc_iflist_ref == 0)) \ 214 cv_broadcast(&(_sc)->sc_cv); \ 215 mtx_unlock(&(_sc)->sc_mtx); \ 216 } while (0) 217 #define BRIDGE_XLOCK(_sc) do { \ 218 mtx_assert(&(_sc)->sc_mtx, MA_OWNED); \ 219 (_sc)->sc_iflist_xcnt++; \ 220 while ((_sc)->sc_iflist_ref > 0) \ 221 cv_wait(&(_sc)->sc_cv, &(_sc)->sc_mtx); \ 222 } while (0) 223 #define BRIDGE_XDROP(_sc) do { \ 224 mtx_assert(&(_sc)->sc_mtx, MA_OWNED); \ 225 (_sc)->sc_iflist_xcnt--; \ 226 } while (0) 227 228 /* 229 * Bridge interface list entry. 230 */ 231 struct bridge_iflist { 232 CK_LIST_ENTRY(bridge_iflist) bif_next; 233 struct ifnet *bif_ifp; /* member if */ 234 struct bstp_port bif_stp; /* STP state */ 235 uint32_t bif_flags; /* member if flags */ 236 int bif_savedcaps; /* saved capabilities */ 237 uint32_t bif_addrmax; /* max # of addresses */ 238 uint32_t bif_addrcnt; /* cur. # of addresses */ 239 uint32_t bif_addrexceeded;/* # of address violations */ 240 }; 241 242 /* 243 * Bridge route node. 244 */ 245 struct bridge_rtnode { 246 CK_LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */ 247 CK_LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */ 248 struct bridge_iflist *brt_dst; /* destination if */ 249 unsigned long brt_expire; /* expiration time */ 250 uint8_t brt_flags; /* address flags */ 251 uint8_t brt_addr[ETHER_ADDR_LEN]; 252 uint16_t brt_vlan; /* vlan id */ 253 }; 254 #define brt_ifp brt_dst->bif_ifp 255 256 /* 257 * Software state for each bridge. 258 */ 259 struct bridge_softc { 260 struct ifnet *sc_ifp; /* make this an interface */ 261 LIST_ENTRY(bridge_softc) sc_list; 262 struct mtx sc_mtx; 263 struct cv sc_cv; 264 uint32_t sc_brtmax; /* max # of addresses */ 265 uint32_t sc_brtcnt; /* cur. # of addresses */ 266 uint32_t sc_brttimeout; /* rt timeout in seconds */ 267 struct callout sc_brcallout; /* bridge callout */ 268 uint32_t sc_iflist_ref; /* refcount for sc_iflist */ 269 uint32_t sc_iflist_xcnt; /* refcount for sc_iflist */ 270 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */ 271 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */ 272 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */ 273 uint32_t sc_rthash_key; /* key for hash */ 274 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */ 275 struct bstp_state sc_stp; /* STP state */ 276 uint32_t sc_brtexceeded; /* # of cache drops */ 277 struct ifnet *sc_ifaddr; /* member mac copied from */ 278 struct ether_addr sc_defaddr; /* Default MAC address */ 279 }; 280 281 VNET_DEFINE_STATIC(struct mtx, bridge_list_mtx); 282 #define V_bridge_list_mtx VNET(bridge_list_mtx) 283 static eventhandler_tag bridge_detach_cookie; 284 285 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD; 286 287 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone); 288 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone) 289 290 static int bridge_clone_create(struct if_clone *, int, caddr_t); 291 static void bridge_clone_destroy(struct ifnet *); 292 293 static int bridge_ioctl(struct ifnet *, u_long, caddr_t); 294 static void bridge_mutecaps(struct bridge_softc *); 295 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *, 296 int); 297 static void bridge_ifdetach(void *arg __unused, struct ifnet *); 298 static void bridge_init(void *); 299 static void bridge_dummynet(struct mbuf *, struct ifnet *); 300 static void bridge_stop(struct ifnet *, int); 301 static int bridge_transmit(struct ifnet *, struct mbuf *); 302 static void bridge_qflush(struct ifnet *); 303 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *); 304 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *, 305 struct rtentry *); 306 static int bridge_enqueue(struct bridge_softc *, struct ifnet *, 307 struct mbuf *); 308 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int); 309 310 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *, 311 struct mbuf *m); 312 313 static void bridge_timer(void *); 314 315 static void bridge_broadcast(struct bridge_softc *, struct ifnet *, 316 struct mbuf *, int); 317 static void bridge_span(struct bridge_softc *, struct mbuf *); 318 319 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *, 320 uint16_t, struct bridge_iflist *, int, uint8_t); 321 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *, 322 uint16_t); 323 static void bridge_rttrim(struct bridge_softc *); 324 static void bridge_rtage(struct bridge_softc *); 325 static void bridge_rtflush(struct bridge_softc *, int); 326 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *, 327 uint16_t); 328 329 static void bridge_rtable_init(struct bridge_softc *); 330 static void bridge_rtable_fini(struct bridge_softc *); 331 332 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *); 333 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *, 334 const uint8_t *, uint16_t); 335 static int bridge_rtnode_insert(struct bridge_softc *, 336 struct bridge_rtnode *); 337 static void bridge_rtnode_destroy(struct bridge_softc *, 338 struct bridge_rtnode *); 339 static void bridge_rtable_expire(struct ifnet *, int); 340 static void bridge_state_change(struct ifnet *, int); 341 342 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *, 343 const char *name); 344 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *, 345 struct ifnet *ifp); 346 static void bridge_delete_member(struct bridge_softc *, 347 struct bridge_iflist *, int); 348 static void bridge_delete_span(struct bridge_softc *, 349 struct bridge_iflist *); 350 351 static int bridge_ioctl_add(struct bridge_softc *, void *); 352 static int bridge_ioctl_del(struct bridge_softc *, void *); 353 static int bridge_ioctl_gifflags(struct bridge_softc *, void *); 354 static int bridge_ioctl_sifflags(struct bridge_softc *, void *); 355 static int bridge_ioctl_scache(struct bridge_softc *, void *); 356 static int bridge_ioctl_gcache(struct bridge_softc *, void *); 357 static int bridge_ioctl_gifs(struct bridge_softc *, void *); 358 static int bridge_ioctl_rts(struct bridge_softc *, void *); 359 static int bridge_ioctl_saddr(struct bridge_softc *, void *); 360 static int bridge_ioctl_sto(struct bridge_softc *, void *); 361 static int bridge_ioctl_gto(struct bridge_softc *, void *); 362 static int bridge_ioctl_daddr(struct bridge_softc *, void *); 363 static int bridge_ioctl_flush(struct bridge_softc *, void *); 364 static int bridge_ioctl_gpri(struct bridge_softc *, void *); 365 static int bridge_ioctl_spri(struct bridge_softc *, void *); 366 static int bridge_ioctl_ght(struct bridge_softc *, void *); 367 static int bridge_ioctl_sht(struct bridge_softc *, void *); 368 static int bridge_ioctl_gfd(struct bridge_softc *, void *); 369 static int bridge_ioctl_sfd(struct bridge_softc *, void *); 370 static int bridge_ioctl_gma(struct bridge_softc *, void *); 371 static int bridge_ioctl_sma(struct bridge_softc *, void *); 372 static int bridge_ioctl_sifprio(struct bridge_softc *, void *); 373 static int bridge_ioctl_sifcost(struct bridge_softc *, void *); 374 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *); 375 static int bridge_ioctl_addspan(struct bridge_softc *, void *); 376 static int bridge_ioctl_delspan(struct bridge_softc *, void *); 377 static int bridge_ioctl_gbparam(struct bridge_softc *, void *); 378 static int bridge_ioctl_grte(struct bridge_softc *, void *); 379 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *); 380 static int bridge_ioctl_sproto(struct bridge_softc *, void *); 381 static int bridge_ioctl_stxhc(struct bridge_softc *, void *); 382 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *, 383 int); 384 static int bridge_ip_checkbasic(struct mbuf **mp); 385 #ifdef INET6 386 static int bridge_ip6_checkbasic(struct mbuf **mp); 387 #endif /* INET6 */ 388 static int bridge_fragment(struct ifnet *, struct mbuf **mp, 389 struct ether_header *, int, struct llc *); 390 static void bridge_linkstate(struct ifnet *ifp); 391 static void bridge_linkcheck(struct bridge_softc *sc); 392 393 394 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */ 395 #define VLANTAGOF(_m) \ 396 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1 397 398 static struct bstp_cb_ops bridge_ops = { 399 .bcb_state = bridge_state_change, 400 .bcb_rtage = bridge_rtable_expire 401 }; 402 403 SYSCTL_DECL(_net_link); 404 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, 405 "Bridge"); 406 407 /* only pass IP[46] packets when pfil is enabled */ 408 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1; 409 #define V_pfil_onlyip VNET(pfil_onlyip) 410 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, 411 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0, 412 "Only pass IP packets when pfil is enabled"); 413 414 /* run pfil hooks on the bridge interface */ 415 VNET_DEFINE_STATIC(int, pfil_bridge) = 1; 416 #define V_pfil_bridge VNET(pfil_bridge) 417 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, 418 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0, 419 "Packet filter on the bridge interface"); 420 421 /* layer2 filter with ipfw */ 422 VNET_DEFINE_STATIC(int, pfil_ipfw); 423 #define V_pfil_ipfw VNET(pfil_ipfw) 424 425 /* layer2 ARP filter with ipfw */ 426 VNET_DEFINE_STATIC(int, pfil_ipfw_arp); 427 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp) 428 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, 429 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0, 430 "Filter ARP packets through IPFW layer2"); 431 432 /* run pfil hooks on the member interface */ 433 VNET_DEFINE_STATIC(int, pfil_member) = 1; 434 #define V_pfil_member VNET(pfil_member) 435 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, 436 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0, 437 "Packet filter on the member interface"); 438 439 /* run pfil hooks on the physical interface for locally destined packets */ 440 VNET_DEFINE_STATIC(int, pfil_local_phys); 441 #define V_pfil_local_phys VNET(pfil_local_phys) 442 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, 443 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0, 444 "Packet filter on the physical interface for locally destined packets"); 445 446 /* log STP state changes */ 447 VNET_DEFINE_STATIC(int, log_stp); 448 #define V_log_stp VNET(log_stp) 449 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, 450 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0, 451 "Log STP state changes"); 452 453 /* share MAC with first bridge member */ 454 VNET_DEFINE_STATIC(int, bridge_inherit_mac); 455 #define V_bridge_inherit_mac VNET(bridge_inherit_mac) 456 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, 457 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0, 458 "Inherit MAC address from the first bridge member"); 459 460 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0; 461 #define V_allow_llz_overlap VNET(allow_llz_overlap) 462 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap, 463 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0, 464 "Allow overlap of link-local scope " 465 "zones of a bridge interface and the member interfaces"); 466 467 struct bridge_control { 468 int (*bc_func)(struct bridge_softc *, void *); 469 int bc_argsize; 470 int bc_flags; 471 }; 472 473 #define BC_F_COPYIN 0x01 /* copy arguments in */ 474 #define BC_F_COPYOUT 0x02 /* copy arguments out */ 475 #define BC_F_SUSER 0x04 /* do super-user check */ 476 477 const struct bridge_control bridge_control_table[] = { 478 { bridge_ioctl_add, sizeof(struct ifbreq), 479 BC_F_COPYIN|BC_F_SUSER }, 480 { bridge_ioctl_del, sizeof(struct ifbreq), 481 BC_F_COPYIN|BC_F_SUSER }, 482 483 { bridge_ioctl_gifflags, sizeof(struct ifbreq), 484 BC_F_COPYIN|BC_F_COPYOUT }, 485 { bridge_ioctl_sifflags, sizeof(struct ifbreq), 486 BC_F_COPYIN|BC_F_SUSER }, 487 488 { bridge_ioctl_scache, sizeof(struct ifbrparam), 489 BC_F_COPYIN|BC_F_SUSER }, 490 { bridge_ioctl_gcache, sizeof(struct ifbrparam), 491 BC_F_COPYOUT }, 492 493 { bridge_ioctl_gifs, sizeof(struct ifbifconf), 494 BC_F_COPYIN|BC_F_COPYOUT }, 495 { bridge_ioctl_rts, sizeof(struct ifbaconf), 496 BC_F_COPYIN|BC_F_COPYOUT }, 497 498 { bridge_ioctl_saddr, sizeof(struct ifbareq), 499 BC_F_COPYIN|BC_F_SUSER }, 500 501 { bridge_ioctl_sto, sizeof(struct ifbrparam), 502 BC_F_COPYIN|BC_F_SUSER }, 503 { bridge_ioctl_gto, sizeof(struct ifbrparam), 504 BC_F_COPYOUT }, 505 506 { bridge_ioctl_daddr, sizeof(struct ifbareq), 507 BC_F_COPYIN|BC_F_SUSER }, 508 509 { bridge_ioctl_flush, sizeof(struct ifbreq), 510 BC_F_COPYIN|BC_F_SUSER }, 511 512 { bridge_ioctl_gpri, sizeof(struct ifbrparam), 513 BC_F_COPYOUT }, 514 { bridge_ioctl_spri, sizeof(struct ifbrparam), 515 BC_F_COPYIN|BC_F_SUSER }, 516 517 { bridge_ioctl_ght, sizeof(struct ifbrparam), 518 BC_F_COPYOUT }, 519 { bridge_ioctl_sht, sizeof(struct ifbrparam), 520 BC_F_COPYIN|BC_F_SUSER }, 521 522 { bridge_ioctl_gfd, sizeof(struct ifbrparam), 523 BC_F_COPYOUT }, 524 { bridge_ioctl_sfd, sizeof(struct ifbrparam), 525 BC_F_COPYIN|BC_F_SUSER }, 526 527 { bridge_ioctl_gma, sizeof(struct ifbrparam), 528 BC_F_COPYOUT }, 529 { bridge_ioctl_sma, sizeof(struct ifbrparam), 530 BC_F_COPYIN|BC_F_SUSER }, 531 532 { bridge_ioctl_sifprio, sizeof(struct ifbreq), 533 BC_F_COPYIN|BC_F_SUSER }, 534 535 { bridge_ioctl_sifcost, sizeof(struct ifbreq), 536 BC_F_COPYIN|BC_F_SUSER }, 537 538 { bridge_ioctl_addspan, sizeof(struct ifbreq), 539 BC_F_COPYIN|BC_F_SUSER }, 540 { bridge_ioctl_delspan, sizeof(struct ifbreq), 541 BC_F_COPYIN|BC_F_SUSER }, 542 543 { bridge_ioctl_gbparam, sizeof(struct ifbropreq), 544 BC_F_COPYOUT }, 545 546 { bridge_ioctl_grte, sizeof(struct ifbrparam), 547 BC_F_COPYOUT }, 548 549 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf), 550 BC_F_COPYIN|BC_F_COPYOUT }, 551 552 { bridge_ioctl_sproto, sizeof(struct ifbrparam), 553 BC_F_COPYIN|BC_F_SUSER }, 554 555 { bridge_ioctl_stxhc, sizeof(struct ifbrparam), 556 BC_F_COPYIN|BC_F_SUSER }, 557 558 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq), 559 BC_F_COPYIN|BC_F_SUSER }, 560 561 }; 562 const int bridge_control_table_size = nitems(bridge_control_table); 563 564 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list); 565 #define V_bridge_list VNET(bridge_list) 566 #define BRIDGE_LIST_LOCK_INIT(x) mtx_init(&V_bridge_list_mtx, \ 567 "if_bridge list", NULL, MTX_DEF) 568 #define BRIDGE_LIST_LOCK_DESTROY(x) mtx_destroy(&V_bridge_list_mtx) 569 #define BRIDGE_LIST_LOCK(x) mtx_lock(&V_bridge_list_mtx) 570 #define BRIDGE_LIST_UNLOCK(x) mtx_unlock(&V_bridge_list_mtx) 571 572 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner); 573 #define V_bridge_cloner VNET(bridge_cloner) 574 575 static const char bridge_name[] = "bridge"; 576 577 static void 578 vnet_bridge_init(const void *unused __unused) 579 { 580 581 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode", 582 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL, 583 UMA_ALIGN_PTR, 0); 584 BRIDGE_LIST_LOCK_INIT(); 585 LIST_INIT(&V_bridge_list); 586 V_bridge_cloner = if_clone_simple(bridge_name, 587 bridge_clone_create, bridge_clone_destroy, 0); 588 } 589 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY, 590 vnet_bridge_init, NULL); 591 592 static void 593 vnet_bridge_uninit(const void *unused __unused) 594 { 595 596 if_clone_detach(V_bridge_cloner); 597 V_bridge_cloner = NULL; 598 BRIDGE_LIST_LOCK_DESTROY(); 599 uma_zdestroy(V_bridge_rtnode_zone); 600 } 601 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY, 602 vnet_bridge_uninit, NULL); 603 604 static int 605 bridge_modevent(module_t mod, int type, void *data) 606 { 607 608 switch (type) { 609 case MOD_LOAD: 610 bridge_dn_p = bridge_dummynet; 611 bridge_detach_cookie = EVENTHANDLER_REGISTER( 612 ifnet_departure_event, bridge_ifdetach, NULL, 613 EVENTHANDLER_PRI_ANY); 614 break; 615 case MOD_UNLOAD: 616 EVENTHANDLER_DEREGISTER(ifnet_departure_event, 617 bridge_detach_cookie); 618 bridge_dn_p = NULL; 619 break; 620 default: 621 return (EOPNOTSUPP); 622 } 623 return (0); 624 } 625 626 static moduledata_t bridge_mod = { 627 "if_bridge", 628 bridge_modevent, 629 0 630 }; 631 632 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY); 633 MODULE_VERSION(if_bridge, 1); 634 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1); 635 636 /* 637 * handler for net.link.bridge.ipfw 638 */ 639 static int 640 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS) 641 { 642 int enable = V_pfil_ipfw; 643 int error; 644 645 error = sysctl_handle_int(oidp, &enable, 0, req); 646 enable &= 1; 647 648 if (enable != V_pfil_ipfw) { 649 V_pfil_ipfw = enable; 650 651 /* 652 * Disable pfil so that ipfw doesnt run twice, if the user 653 * really wants both then they can re-enable pfil_bridge and/or 654 * pfil_member. Also allow non-ip packets as ipfw can filter by 655 * layer2 type. 656 */ 657 if (V_pfil_ipfw) { 658 V_pfil_onlyip = 0; 659 V_pfil_bridge = 0; 660 V_pfil_member = 0; 661 } 662 } 663 664 return (error); 665 } 666 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, 667 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT, 668 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I", 669 "Layer2 filter with IPFW"); 670 671 /* 672 * bridge_clone_create: 673 * 674 * Create a new bridge instance. 675 */ 676 static int 677 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params) 678 { 679 struct bridge_softc *sc; 680 struct ifnet *ifp; 681 682 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO); 683 ifp = sc->sc_ifp = if_alloc(IFT_ETHER); 684 if (ifp == NULL) { 685 free(sc, M_DEVBUF); 686 return (ENOSPC); 687 } 688 689 BRIDGE_LOCK_INIT(sc); 690 sc->sc_brtmax = BRIDGE_RTABLE_MAX; 691 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT; 692 693 /* Initialize our routing table. */ 694 bridge_rtable_init(sc); 695 696 callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0); 697 698 CK_LIST_INIT(&sc->sc_iflist); 699 CK_LIST_INIT(&sc->sc_spanlist); 700 701 ifp->if_softc = sc; 702 if_initname(ifp, bridge_name, unit); 703 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; 704 ifp->if_ioctl = bridge_ioctl; 705 ifp->if_transmit = bridge_transmit; 706 ifp->if_qflush = bridge_qflush; 707 ifp->if_init = bridge_init; 708 ifp->if_type = IFT_BRIDGE; 709 710 ether_gen_addr(ifp, &sc->sc_defaddr); 711 712 bstp_attach(&sc->sc_stp, &bridge_ops); 713 ether_ifattach(ifp, sc->sc_defaddr.octet); 714 /* Now undo some of the damage... */ 715 ifp->if_baudrate = 0; 716 ifp->if_type = IFT_BRIDGE; 717 718 BRIDGE_LIST_LOCK(); 719 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list); 720 BRIDGE_LIST_UNLOCK(); 721 722 return (0); 723 } 724 725 /* 726 * bridge_clone_destroy: 727 * 728 * Destroy a bridge instance. 729 */ 730 static void 731 bridge_clone_destroy(struct ifnet *ifp) 732 { 733 struct bridge_softc *sc = ifp->if_softc; 734 struct bridge_iflist *bif; 735 736 BRIDGE_LOCK(sc); 737 738 bridge_stop(ifp, 1); 739 ifp->if_flags &= ~IFF_UP; 740 741 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL) 742 bridge_delete_member(sc, bif, 0); 743 744 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) { 745 bridge_delete_span(sc, bif); 746 } 747 748 /* Tear down the routing table. */ 749 bridge_rtable_fini(sc); 750 751 BRIDGE_UNLOCK(sc); 752 753 callout_drain(&sc->sc_brcallout); 754 755 BRIDGE_LIST_LOCK(); 756 LIST_REMOVE(sc, sc_list); 757 BRIDGE_LIST_UNLOCK(); 758 759 bstp_detach(&sc->sc_stp); 760 ether_ifdetach(ifp); 761 if_free(ifp); 762 763 BRIDGE_LOCK_DESTROY(sc); 764 free(sc, M_DEVBUF); 765 } 766 767 /* 768 * bridge_ioctl: 769 * 770 * Handle a control request from the operator. 771 */ 772 static int 773 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data) 774 { 775 struct bridge_softc *sc = ifp->if_softc; 776 struct ifreq *ifr = (struct ifreq *)data; 777 struct bridge_iflist *bif; 778 struct thread *td = curthread; 779 union { 780 struct ifbreq ifbreq; 781 struct ifbifconf ifbifconf; 782 struct ifbareq ifbareq; 783 struct ifbaconf ifbaconf; 784 struct ifbrparam ifbrparam; 785 struct ifbropreq ifbropreq; 786 } args; 787 struct ifdrv *ifd = (struct ifdrv *) data; 788 const struct bridge_control *bc; 789 int error = 0, oldmtu; 790 791 switch (cmd) { 792 793 case SIOCADDMULTI: 794 case SIOCDELMULTI: 795 break; 796 797 case SIOCGDRVSPEC: 798 case SIOCSDRVSPEC: 799 if (ifd->ifd_cmd >= bridge_control_table_size) { 800 error = EINVAL; 801 break; 802 } 803 bc = &bridge_control_table[ifd->ifd_cmd]; 804 805 if (cmd == SIOCGDRVSPEC && 806 (bc->bc_flags & BC_F_COPYOUT) == 0) { 807 error = EINVAL; 808 break; 809 } 810 else if (cmd == SIOCSDRVSPEC && 811 (bc->bc_flags & BC_F_COPYOUT) != 0) { 812 error = EINVAL; 813 break; 814 } 815 816 if (bc->bc_flags & BC_F_SUSER) { 817 error = priv_check(td, PRIV_NET_BRIDGE); 818 if (error) 819 break; 820 } 821 822 if (ifd->ifd_len != bc->bc_argsize || 823 ifd->ifd_len > sizeof(args)) { 824 error = EINVAL; 825 break; 826 } 827 828 bzero(&args, sizeof(args)); 829 if (bc->bc_flags & BC_F_COPYIN) { 830 error = copyin(ifd->ifd_data, &args, ifd->ifd_len); 831 if (error) 832 break; 833 } 834 835 oldmtu = ifp->if_mtu; 836 BRIDGE_LOCK(sc); 837 error = (*bc->bc_func)(sc, &args); 838 BRIDGE_UNLOCK(sc); 839 if (error) 840 break; 841 842 /* 843 * Bridge MTU may change during addition of the first port. 844 * If it did, do network layer specific procedure. 845 */ 846 if (ifp->if_mtu != oldmtu) { 847 #ifdef INET6 848 nd6_setmtu(ifp); 849 #endif 850 rt_updatemtu(ifp); 851 } 852 853 if (bc->bc_flags & BC_F_COPYOUT) 854 error = copyout(&args, ifd->ifd_data, ifd->ifd_len); 855 856 break; 857 858 case SIOCSIFFLAGS: 859 if (!(ifp->if_flags & IFF_UP) && 860 (ifp->if_drv_flags & IFF_DRV_RUNNING)) { 861 /* 862 * If interface is marked down and it is running, 863 * then stop and disable it. 864 */ 865 BRIDGE_LOCK(sc); 866 bridge_stop(ifp, 1); 867 BRIDGE_UNLOCK(sc); 868 } else if ((ifp->if_flags & IFF_UP) && 869 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) { 870 /* 871 * If interface is marked up and it is stopped, then 872 * start it. 873 */ 874 (*ifp->if_init)(sc); 875 } 876 break; 877 878 case SIOCSIFMTU: 879 if (ifr->ifr_mtu < 576) { 880 error = EINVAL; 881 break; 882 } 883 if (CK_LIST_EMPTY(&sc->sc_iflist)) { 884 sc->sc_ifp->if_mtu = ifr->ifr_mtu; 885 break; 886 } 887 BRIDGE_LOCK(sc); 888 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 889 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) { 890 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)" 891 " != %d\n", sc->sc_ifp->if_xname, 892 bif->bif_ifp->if_mtu, 893 bif->bif_ifp->if_xname, ifr->ifr_mtu); 894 error = EINVAL; 895 break; 896 } 897 } 898 if (!error) 899 sc->sc_ifp->if_mtu = ifr->ifr_mtu; 900 BRIDGE_UNLOCK(sc); 901 break; 902 default: 903 /* 904 * drop the lock as ether_ioctl() will call bridge_start() and 905 * cause the lock to be recursed. 906 */ 907 error = ether_ioctl(ifp, cmd, data); 908 break; 909 } 910 911 return (error); 912 } 913 914 /* 915 * bridge_mutecaps: 916 * 917 * Clear or restore unwanted capabilities on the member interface 918 */ 919 static void 920 bridge_mutecaps(struct bridge_softc *sc) 921 { 922 struct bridge_iflist *bif; 923 int enabled, mask; 924 925 /* Initial bitmask of capabilities to test */ 926 mask = BRIDGE_IFCAPS_MASK; 927 928 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 929 /* Every member must support it or its disabled */ 930 mask &= bif->bif_savedcaps; 931 } 932 933 BRIDGE_XLOCK(sc); 934 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 935 enabled = bif->bif_ifp->if_capenable; 936 enabled &= ~BRIDGE_IFCAPS_STRIP; 937 /* strip off mask bits and enable them again if allowed */ 938 enabled &= ~BRIDGE_IFCAPS_MASK; 939 enabled |= mask; 940 BRIDGE_UNLOCK(sc); 941 bridge_set_ifcap(sc, bif, enabled); 942 BRIDGE_LOCK(sc); 943 } 944 BRIDGE_XDROP(sc); 945 946 } 947 948 static void 949 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set) 950 { 951 struct ifnet *ifp = bif->bif_ifp; 952 struct ifreq ifr; 953 int error, mask, stuck; 954 955 BRIDGE_UNLOCK_ASSERT(sc); 956 957 bzero(&ifr, sizeof(ifr)); 958 ifr.ifr_reqcap = set; 959 960 if (ifp->if_capenable != set) { 961 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr); 962 if (error) 963 if_printf(sc->sc_ifp, 964 "error setting capabilities on %s: %d\n", 965 ifp->if_xname, error); 966 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP; 967 stuck = ifp->if_capenable & mask & ~set; 968 if (stuck != 0) 969 if_printf(sc->sc_ifp, 970 "can't disable some capabilities on %s: 0x%x\n", 971 ifp->if_xname, stuck); 972 } 973 } 974 975 /* 976 * bridge_lookup_member: 977 * 978 * Lookup a bridge member interface. 979 */ 980 static struct bridge_iflist * 981 bridge_lookup_member(struct bridge_softc *sc, const char *name) 982 { 983 struct bridge_iflist *bif; 984 struct ifnet *ifp; 985 986 BRIDGE_LOCK_ASSERT(sc); 987 988 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 989 ifp = bif->bif_ifp; 990 if (strcmp(ifp->if_xname, name) == 0) 991 return (bif); 992 } 993 994 return (NULL); 995 } 996 997 /* 998 * bridge_lookup_member_if: 999 * 1000 * Lookup a bridge member interface by ifnet*. 1001 */ 1002 static struct bridge_iflist * 1003 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp) 1004 { 1005 struct bridge_iflist *bif; 1006 1007 BRIDGE_LOCK_ASSERT(sc); 1008 1009 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1010 if (bif->bif_ifp == member_ifp) 1011 return (bif); 1012 } 1013 1014 return (NULL); 1015 } 1016 1017 /* 1018 * bridge_delete_member: 1019 * 1020 * Delete the specified member interface. 1021 */ 1022 static void 1023 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif, 1024 int gone) 1025 { 1026 struct ifnet *ifs = bif->bif_ifp; 1027 struct ifnet *fif = NULL; 1028 struct bridge_iflist *bifl; 1029 1030 BRIDGE_LOCK_ASSERT(sc); 1031 1032 if (bif->bif_flags & IFBIF_STP) 1033 bstp_disable(&bif->bif_stp); 1034 1035 ifs->if_bridge = NULL; 1036 BRIDGE_XLOCK(sc); 1037 CK_LIST_REMOVE(bif, bif_next); 1038 BRIDGE_XDROP(sc); 1039 1040 /* 1041 * If removing the interface that gave the bridge its mac address, set 1042 * the mac address of the bridge to the address of the next member, or 1043 * to its default address if no members are left. 1044 */ 1045 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) { 1046 if (CK_LIST_EMPTY(&sc->sc_iflist)) { 1047 bcopy(&sc->sc_defaddr, 1048 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1049 sc->sc_ifaddr = NULL; 1050 } else { 1051 bifl = CK_LIST_FIRST(&sc->sc_iflist); 1052 fif = bifl->bif_ifp; 1053 bcopy(IF_LLADDR(fif), 1054 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1055 sc->sc_ifaddr = fif; 1056 } 1057 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp); 1058 } 1059 1060 bridge_linkcheck(sc); 1061 bridge_mutecaps(sc); /* recalcuate now this interface is removed */ 1062 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL); 1063 KASSERT(bif->bif_addrcnt == 0, 1064 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt)); 1065 1066 ifs->if_bridge_output = NULL; 1067 ifs->if_bridge_input = NULL; 1068 ifs->if_bridge_linkstate = NULL; 1069 BRIDGE_UNLOCK(sc); 1070 if (!gone) { 1071 switch (ifs->if_type) { 1072 case IFT_ETHER: 1073 case IFT_L2VLAN: 1074 /* 1075 * Take the interface out of promiscuous mode, but only 1076 * if it was promiscuous in the first place. It might 1077 * not be if we're in the bridge_ioctl_add() error path. 1078 */ 1079 if (ifs->if_flags & IFF_PROMISC) 1080 (void) ifpromisc(ifs, 0); 1081 break; 1082 1083 case IFT_GIF: 1084 break; 1085 1086 default: 1087 #ifdef DIAGNOSTIC 1088 panic("bridge_delete_member: impossible"); 1089 #endif 1090 break; 1091 } 1092 /* reneable any interface capabilities */ 1093 bridge_set_ifcap(sc, bif, bif->bif_savedcaps); 1094 } 1095 bstp_destroy(&bif->bif_stp); /* prepare to free */ 1096 BRIDGE_LOCK(sc); 1097 free(bif, M_DEVBUF); 1098 } 1099 1100 /* 1101 * bridge_delete_span: 1102 * 1103 * Delete the specified span interface. 1104 */ 1105 static void 1106 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif) 1107 { 1108 BRIDGE_LOCK_ASSERT(sc); 1109 1110 KASSERT(bif->bif_ifp->if_bridge == NULL, 1111 ("%s: not a span interface", __func__)); 1112 1113 CK_LIST_REMOVE(bif, bif_next); 1114 free(bif, M_DEVBUF); 1115 } 1116 1117 static int 1118 bridge_ioctl_add(struct bridge_softc *sc, void *arg) 1119 { 1120 struct ifbreq *req = arg; 1121 struct bridge_iflist *bif = NULL; 1122 struct ifnet *ifs; 1123 int error = 0; 1124 1125 ifs = ifunit(req->ifbr_ifsname); 1126 if (ifs == NULL) 1127 return (ENOENT); 1128 if (ifs->if_ioctl == NULL) /* must be supported */ 1129 return (EINVAL); 1130 1131 /* If it's in the span list, it can't be a member. */ 1132 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1133 if (ifs == bif->bif_ifp) 1134 return (EBUSY); 1135 1136 if (ifs->if_bridge == sc) 1137 return (EEXIST); 1138 1139 if (ifs->if_bridge != NULL) 1140 return (EBUSY); 1141 1142 switch (ifs->if_type) { 1143 case IFT_ETHER: 1144 case IFT_L2VLAN: 1145 case IFT_GIF: 1146 /* permitted interface types */ 1147 break; 1148 default: 1149 return (EINVAL); 1150 } 1151 1152 #ifdef INET6 1153 /* 1154 * Two valid inet6 addresses with link-local scope must not be 1155 * on the parent interface and the member interfaces at the 1156 * same time. This restriction is needed to prevent violation 1157 * of link-local scope zone. Attempts to add a member 1158 * interface which has inet6 addresses when the parent has 1159 * inet6 triggers removal of all inet6 addresses on the member 1160 * interface. 1161 */ 1162 1163 /* Check if the parent interface has a link-local scope addr. */ 1164 if (V_allow_llz_overlap == 0 && 1165 in6ifa_llaonifp(sc->sc_ifp) != NULL) { 1166 /* 1167 * If any, remove all inet6 addresses from the member 1168 * interfaces. 1169 */ 1170 BRIDGE_XLOCK(sc); 1171 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1172 if (in6ifa_llaonifp(bif->bif_ifp)) { 1173 BRIDGE_UNLOCK(sc); 1174 in6_ifdetach(bif->bif_ifp); 1175 BRIDGE_LOCK(sc); 1176 if_printf(sc->sc_ifp, 1177 "IPv6 addresses on %s have been removed " 1178 "before adding it as a member to prevent " 1179 "IPv6 address scope violation.\n", 1180 bif->bif_ifp->if_xname); 1181 } 1182 } 1183 BRIDGE_XDROP(sc); 1184 if (in6ifa_llaonifp(ifs)) { 1185 BRIDGE_UNLOCK(sc); 1186 in6_ifdetach(ifs); 1187 BRIDGE_LOCK(sc); 1188 if_printf(sc->sc_ifp, 1189 "IPv6 addresses on %s have been removed " 1190 "before adding it as a member to prevent " 1191 "IPv6 address scope violation.\n", 1192 ifs->if_xname); 1193 } 1194 } 1195 #endif 1196 /* Allow the first Ethernet member to define the MTU */ 1197 if (CK_LIST_EMPTY(&sc->sc_iflist)) 1198 sc->sc_ifp->if_mtu = ifs->if_mtu; 1199 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) { 1200 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n", 1201 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu); 1202 return (EINVAL); 1203 } 1204 1205 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO); 1206 if (bif == NULL) 1207 return (ENOMEM); 1208 1209 bif->bif_ifp = ifs; 1210 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER; 1211 bif->bif_savedcaps = ifs->if_capenable; 1212 1213 /* 1214 * Assign the interface's MAC address to the bridge if it's the first 1215 * member and the MAC address of the bridge has not been changed from 1216 * the default randomly generated one. 1217 */ 1218 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) && 1219 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) { 1220 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1221 sc->sc_ifaddr = ifs; 1222 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp); 1223 } 1224 1225 ifs->if_bridge = sc; 1226 ifs->if_bridge_output = bridge_output; 1227 ifs->if_bridge_input = bridge_input; 1228 ifs->if_bridge_linkstate = bridge_linkstate; 1229 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp); 1230 /* 1231 * XXX: XLOCK HERE!?! 1232 * 1233 * NOTE: insert_***HEAD*** should be safe for the traversals. 1234 */ 1235 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next); 1236 1237 /* Set interface capabilities to the intersection set of all members */ 1238 bridge_mutecaps(sc); 1239 bridge_linkcheck(sc); 1240 1241 /* Place the interface into promiscuous mode */ 1242 switch (ifs->if_type) { 1243 case IFT_ETHER: 1244 case IFT_L2VLAN: 1245 BRIDGE_UNLOCK(sc); 1246 error = ifpromisc(ifs, 1); 1247 BRIDGE_LOCK(sc); 1248 break; 1249 } 1250 1251 if (error) 1252 bridge_delete_member(sc, bif, 0); 1253 return (error); 1254 } 1255 1256 static int 1257 bridge_ioctl_del(struct bridge_softc *sc, void *arg) 1258 { 1259 struct ifbreq *req = arg; 1260 struct bridge_iflist *bif; 1261 1262 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1263 if (bif == NULL) 1264 return (ENOENT); 1265 1266 bridge_delete_member(sc, bif, 0); 1267 1268 return (0); 1269 } 1270 1271 static int 1272 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg) 1273 { 1274 struct ifbreq *req = arg; 1275 struct bridge_iflist *bif; 1276 struct bstp_port *bp; 1277 1278 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1279 if (bif == NULL) 1280 return (ENOENT); 1281 1282 bp = &bif->bif_stp; 1283 req->ifbr_ifsflags = bif->bif_flags; 1284 req->ifbr_state = bp->bp_state; 1285 req->ifbr_priority = bp->bp_priority; 1286 req->ifbr_path_cost = bp->bp_path_cost; 1287 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff; 1288 req->ifbr_proto = bp->bp_protover; 1289 req->ifbr_role = bp->bp_role; 1290 req->ifbr_stpflags = bp->bp_flags; 1291 req->ifbr_addrcnt = bif->bif_addrcnt; 1292 req->ifbr_addrmax = bif->bif_addrmax; 1293 req->ifbr_addrexceeded = bif->bif_addrexceeded; 1294 1295 /* Copy STP state options as flags */ 1296 if (bp->bp_operedge) 1297 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE; 1298 if (bp->bp_flags & BSTP_PORT_AUTOEDGE) 1299 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE; 1300 if (bp->bp_ptp_link) 1301 req->ifbr_ifsflags |= IFBIF_BSTP_PTP; 1302 if (bp->bp_flags & BSTP_PORT_AUTOPTP) 1303 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP; 1304 if (bp->bp_flags & BSTP_PORT_ADMEDGE) 1305 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE; 1306 if (bp->bp_flags & BSTP_PORT_ADMCOST) 1307 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST; 1308 return (0); 1309 } 1310 1311 static int 1312 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg) 1313 { 1314 struct ifbreq *req = arg; 1315 struct bridge_iflist *bif; 1316 struct bstp_port *bp; 1317 int error; 1318 1319 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1320 if (bif == NULL) 1321 return (ENOENT); 1322 bp = &bif->bif_stp; 1323 1324 if (req->ifbr_ifsflags & IFBIF_SPAN) 1325 /* SPAN is readonly */ 1326 return (EINVAL); 1327 1328 if (req->ifbr_ifsflags & IFBIF_STP) { 1329 if ((bif->bif_flags & IFBIF_STP) == 0) { 1330 error = bstp_enable(&bif->bif_stp); 1331 if (error) 1332 return (error); 1333 } 1334 } else { 1335 if ((bif->bif_flags & IFBIF_STP) != 0) 1336 bstp_disable(&bif->bif_stp); 1337 } 1338 1339 /* Pass on STP flags */ 1340 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0); 1341 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0); 1342 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0); 1343 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0); 1344 1345 /* Save the bits relating to the bridge */ 1346 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK; 1347 1348 return (0); 1349 } 1350 1351 static int 1352 bridge_ioctl_scache(struct bridge_softc *sc, void *arg) 1353 { 1354 struct ifbrparam *param = arg; 1355 1356 sc->sc_brtmax = param->ifbrp_csize; 1357 bridge_rttrim(sc); 1358 1359 return (0); 1360 } 1361 1362 static int 1363 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg) 1364 { 1365 struct ifbrparam *param = arg; 1366 1367 param->ifbrp_csize = sc->sc_brtmax; 1368 1369 return (0); 1370 } 1371 1372 static int 1373 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg) 1374 { 1375 struct ifbifconf *bifc = arg; 1376 struct bridge_iflist *bif; 1377 struct ifbreq breq; 1378 char *buf, *outbuf; 1379 int count, buflen, len, error = 0; 1380 1381 count = 0; 1382 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) 1383 count++; 1384 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1385 count++; 1386 1387 buflen = sizeof(breq) * count; 1388 if (bifc->ifbic_len == 0) { 1389 bifc->ifbic_len = buflen; 1390 return (0); 1391 } 1392 BRIDGE_UNLOCK(sc); 1393 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO); 1394 BRIDGE_LOCK(sc); 1395 1396 count = 0; 1397 buf = outbuf; 1398 len = min(bifc->ifbic_len, buflen); 1399 bzero(&breq, sizeof(breq)); 1400 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1401 if (len < sizeof(breq)) 1402 break; 1403 1404 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname, 1405 sizeof(breq.ifbr_ifsname)); 1406 /* Fill in the ifbreq structure */ 1407 error = bridge_ioctl_gifflags(sc, &breq); 1408 if (error) 1409 break; 1410 memcpy(buf, &breq, sizeof(breq)); 1411 count++; 1412 buf += sizeof(breq); 1413 len -= sizeof(breq); 1414 } 1415 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) { 1416 if (len < sizeof(breq)) 1417 break; 1418 1419 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname, 1420 sizeof(breq.ifbr_ifsname)); 1421 breq.ifbr_ifsflags = bif->bif_flags; 1422 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff; 1423 memcpy(buf, &breq, sizeof(breq)); 1424 count++; 1425 buf += sizeof(breq); 1426 len -= sizeof(breq); 1427 } 1428 1429 BRIDGE_UNLOCK(sc); 1430 bifc->ifbic_len = sizeof(breq) * count; 1431 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len); 1432 BRIDGE_LOCK(sc); 1433 free(outbuf, M_TEMP); 1434 return (error); 1435 } 1436 1437 static int 1438 bridge_ioctl_rts(struct bridge_softc *sc, void *arg) 1439 { 1440 struct ifbaconf *bac = arg; 1441 struct bridge_rtnode *brt; 1442 struct ifbareq bareq; 1443 char *buf, *outbuf; 1444 int count, buflen, len, error = 0; 1445 1446 if (bac->ifbac_len == 0) 1447 return (0); 1448 1449 count = 0; 1450 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) 1451 count++; 1452 buflen = sizeof(bareq) * count; 1453 1454 BRIDGE_UNLOCK(sc); 1455 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO); 1456 BRIDGE_LOCK(sc); 1457 1458 count = 0; 1459 buf = outbuf; 1460 len = min(bac->ifbac_len, buflen); 1461 bzero(&bareq, sizeof(bareq)); 1462 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) { 1463 if (len < sizeof(bareq)) 1464 goto out; 1465 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname, 1466 sizeof(bareq.ifba_ifsname)); 1467 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr)); 1468 bareq.ifba_vlan = brt->brt_vlan; 1469 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 1470 time_uptime < brt->brt_expire) 1471 bareq.ifba_expire = brt->brt_expire - time_uptime; 1472 else 1473 bareq.ifba_expire = 0; 1474 bareq.ifba_flags = brt->brt_flags; 1475 1476 memcpy(buf, &bareq, sizeof(bareq)); 1477 count++; 1478 buf += sizeof(bareq); 1479 len -= sizeof(bareq); 1480 } 1481 out: 1482 BRIDGE_UNLOCK(sc); 1483 bac->ifbac_len = sizeof(bareq) * count; 1484 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len); 1485 BRIDGE_LOCK(sc); 1486 free(outbuf, M_TEMP); 1487 return (error); 1488 } 1489 1490 static int 1491 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg) 1492 { 1493 struct ifbareq *req = arg; 1494 struct bridge_iflist *bif; 1495 int error; 1496 1497 bif = bridge_lookup_member(sc, req->ifba_ifsname); 1498 if (bif == NULL) 1499 return (ENOENT); 1500 1501 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1, 1502 req->ifba_flags); 1503 1504 return (error); 1505 } 1506 1507 static int 1508 bridge_ioctl_sto(struct bridge_softc *sc, void *arg) 1509 { 1510 struct ifbrparam *param = arg; 1511 1512 sc->sc_brttimeout = param->ifbrp_ctime; 1513 return (0); 1514 } 1515 1516 static int 1517 bridge_ioctl_gto(struct bridge_softc *sc, void *arg) 1518 { 1519 struct ifbrparam *param = arg; 1520 1521 param->ifbrp_ctime = sc->sc_brttimeout; 1522 return (0); 1523 } 1524 1525 static int 1526 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg) 1527 { 1528 struct ifbareq *req = arg; 1529 1530 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan)); 1531 } 1532 1533 static int 1534 bridge_ioctl_flush(struct bridge_softc *sc, void *arg) 1535 { 1536 struct ifbreq *req = arg; 1537 1538 bridge_rtflush(sc, req->ifbr_ifsflags); 1539 return (0); 1540 } 1541 1542 static int 1543 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg) 1544 { 1545 struct ifbrparam *param = arg; 1546 struct bstp_state *bs = &sc->sc_stp; 1547 1548 param->ifbrp_prio = bs->bs_bridge_priority; 1549 return (0); 1550 } 1551 1552 static int 1553 bridge_ioctl_spri(struct bridge_softc *sc, void *arg) 1554 { 1555 struct ifbrparam *param = arg; 1556 1557 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio)); 1558 } 1559 1560 static int 1561 bridge_ioctl_ght(struct bridge_softc *sc, void *arg) 1562 { 1563 struct ifbrparam *param = arg; 1564 struct bstp_state *bs = &sc->sc_stp; 1565 1566 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8; 1567 return (0); 1568 } 1569 1570 static int 1571 bridge_ioctl_sht(struct bridge_softc *sc, void *arg) 1572 { 1573 struct ifbrparam *param = arg; 1574 1575 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime)); 1576 } 1577 1578 static int 1579 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg) 1580 { 1581 struct ifbrparam *param = arg; 1582 struct bstp_state *bs = &sc->sc_stp; 1583 1584 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8; 1585 return (0); 1586 } 1587 1588 static int 1589 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg) 1590 { 1591 struct ifbrparam *param = arg; 1592 1593 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay)); 1594 } 1595 1596 static int 1597 bridge_ioctl_gma(struct bridge_softc *sc, void *arg) 1598 { 1599 struct ifbrparam *param = arg; 1600 struct bstp_state *bs = &sc->sc_stp; 1601 1602 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8; 1603 return (0); 1604 } 1605 1606 static int 1607 bridge_ioctl_sma(struct bridge_softc *sc, void *arg) 1608 { 1609 struct ifbrparam *param = arg; 1610 1611 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage)); 1612 } 1613 1614 static int 1615 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg) 1616 { 1617 struct ifbreq *req = arg; 1618 struct bridge_iflist *bif; 1619 1620 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1621 if (bif == NULL) 1622 return (ENOENT); 1623 1624 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority)); 1625 } 1626 1627 static int 1628 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg) 1629 { 1630 struct ifbreq *req = arg; 1631 struct bridge_iflist *bif; 1632 1633 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1634 if (bif == NULL) 1635 return (ENOENT); 1636 1637 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost)); 1638 } 1639 1640 static int 1641 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg) 1642 { 1643 struct ifbreq *req = arg; 1644 struct bridge_iflist *bif; 1645 1646 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1647 if (bif == NULL) 1648 return (ENOENT); 1649 1650 bif->bif_addrmax = req->ifbr_addrmax; 1651 return (0); 1652 } 1653 1654 static int 1655 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg) 1656 { 1657 struct ifbreq *req = arg; 1658 struct bridge_iflist *bif = NULL; 1659 struct ifnet *ifs; 1660 1661 ifs = ifunit(req->ifbr_ifsname); 1662 if (ifs == NULL) 1663 return (ENOENT); 1664 1665 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1666 if (ifs == bif->bif_ifp) 1667 return (EBUSY); 1668 1669 if (ifs->if_bridge != NULL) 1670 return (EBUSY); 1671 1672 switch (ifs->if_type) { 1673 case IFT_ETHER: 1674 case IFT_GIF: 1675 case IFT_L2VLAN: 1676 break; 1677 default: 1678 return (EINVAL); 1679 } 1680 1681 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO); 1682 if (bif == NULL) 1683 return (ENOMEM); 1684 1685 bif->bif_ifp = ifs; 1686 bif->bif_flags = IFBIF_SPAN; 1687 1688 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next); 1689 1690 return (0); 1691 } 1692 1693 static int 1694 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg) 1695 { 1696 struct ifbreq *req = arg; 1697 struct bridge_iflist *bif; 1698 struct ifnet *ifs; 1699 1700 ifs = ifunit(req->ifbr_ifsname); 1701 if (ifs == NULL) 1702 return (ENOENT); 1703 1704 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1705 if (ifs == bif->bif_ifp) 1706 break; 1707 1708 if (bif == NULL) 1709 return (ENOENT); 1710 1711 bridge_delete_span(sc, bif); 1712 1713 return (0); 1714 } 1715 1716 static int 1717 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg) 1718 { 1719 struct ifbropreq *req = arg; 1720 struct bstp_state *bs = &sc->sc_stp; 1721 struct bstp_port *root_port; 1722 1723 req->ifbop_maxage = bs->bs_bridge_max_age >> 8; 1724 req->ifbop_hellotime = bs->bs_bridge_htime >> 8; 1725 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8; 1726 1727 root_port = bs->bs_root_port; 1728 if (root_port == NULL) 1729 req->ifbop_root_port = 0; 1730 else 1731 req->ifbop_root_port = root_port->bp_ifp->if_index; 1732 1733 req->ifbop_holdcount = bs->bs_txholdcount; 1734 req->ifbop_priority = bs->bs_bridge_priority; 1735 req->ifbop_protocol = bs->bs_protover; 1736 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost; 1737 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id; 1738 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id; 1739 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id; 1740 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec; 1741 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec; 1742 1743 return (0); 1744 } 1745 1746 static int 1747 bridge_ioctl_grte(struct bridge_softc *sc, void *arg) 1748 { 1749 struct ifbrparam *param = arg; 1750 1751 param->ifbrp_cexceeded = sc->sc_brtexceeded; 1752 return (0); 1753 } 1754 1755 static int 1756 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg) 1757 { 1758 struct ifbpstpconf *bifstp = arg; 1759 struct bridge_iflist *bif; 1760 struct bstp_port *bp; 1761 struct ifbpstpreq bpreq; 1762 char *buf, *outbuf; 1763 int count, buflen, len, error = 0; 1764 1765 count = 0; 1766 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1767 if ((bif->bif_flags & IFBIF_STP) != 0) 1768 count++; 1769 } 1770 1771 buflen = sizeof(bpreq) * count; 1772 if (bifstp->ifbpstp_len == 0) { 1773 bifstp->ifbpstp_len = buflen; 1774 return (0); 1775 } 1776 1777 BRIDGE_UNLOCK(sc); 1778 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO); 1779 BRIDGE_LOCK(sc); 1780 1781 count = 0; 1782 buf = outbuf; 1783 len = min(bifstp->ifbpstp_len, buflen); 1784 bzero(&bpreq, sizeof(bpreq)); 1785 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1786 if (len < sizeof(bpreq)) 1787 break; 1788 1789 if ((bif->bif_flags & IFBIF_STP) == 0) 1790 continue; 1791 1792 bp = &bif->bif_stp; 1793 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff; 1794 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions; 1795 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost; 1796 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id; 1797 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id; 1798 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id; 1799 1800 memcpy(buf, &bpreq, sizeof(bpreq)); 1801 count++; 1802 buf += sizeof(bpreq); 1803 len -= sizeof(bpreq); 1804 } 1805 1806 BRIDGE_UNLOCK(sc); 1807 bifstp->ifbpstp_len = sizeof(bpreq) * count; 1808 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len); 1809 BRIDGE_LOCK(sc); 1810 free(outbuf, M_TEMP); 1811 return (error); 1812 } 1813 1814 static int 1815 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg) 1816 { 1817 struct ifbrparam *param = arg; 1818 1819 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto)); 1820 } 1821 1822 static int 1823 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg) 1824 { 1825 struct ifbrparam *param = arg; 1826 1827 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc)); 1828 } 1829 1830 /* 1831 * bridge_ifdetach: 1832 * 1833 * Detach an interface from a bridge. Called when a member 1834 * interface is detaching. 1835 */ 1836 static void 1837 bridge_ifdetach(void *arg __unused, struct ifnet *ifp) 1838 { 1839 struct bridge_softc *sc = ifp->if_bridge; 1840 struct bridge_iflist *bif; 1841 1842 if (ifp->if_flags & IFF_RENAMING) 1843 return; 1844 if (V_bridge_cloner == NULL) { 1845 /* 1846 * This detach handler can be called after 1847 * vnet_bridge_uninit(). Just return in that case. 1848 */ 1849 return; 1850 } 1851 /* Check if the interface is a bridge member */ 1852 if (sc != NULL) { 1853 BRIDGE_LOCK(sc); 1854 1855 bif = bridge_lookup_member_if(sc, ifp); 1856 if (bif != NULL) 1857 bridge_delete_member(sc, bif, 1); 1858 1859 BRIDGE_UNLOCK(sc); 1860 return; 1861 } 1862 1863 /* Check if the interface is a span port */ 1864 BRIDGE_LIST_LOCK(); 1865 LIST_FOREACH(sc, &V_bridge_list, sc_list) { 1866 BRIDGE_LOCK(sc); 1867 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1868 if (ifp == bif->bif_ifp) { 1869 bridge_delete_span(sc, bif); 1870 break; 1871 } 1872 1873 BRIDGE_UNLOCK(sc); 1874 } 1875 BRIDGE_LIST_UNLOCK(); 1876 } 1877 1878 /* 1879 * bridge_init: 1880 * 1881 * Initialize a bridge interface. 1882 */ 1883 static void 1884 bridge_init(void *xsc) 1885 { 1886 struct bridge_softc *sc = (struct bridge_softc *)xsc; 1887 struct ifnet *ifp = sc->sc_ifp; 1888 1889 if (ifp->if_drv_flags & IFF_DRV_RUNNING) 1890 return; 1891 1892 BRIDGE_LOCK(sc); 1893 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz, 1894 bridge_timer, sc); 1895 1896 ifp->if_drv_flags |= IFF_DRV_RUNNING; 1897 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */ 1898 1899 BRIDGE_UNLOCK(sc); 1900 } 1901 1902 /* 1903 * bridge_stop: 1904 * 1905 * Stop the bridge interface. 1906 */ 1907 static void 1908 bridge_stop(struct ifnet *ifp, int disable) 1909 { 1910 struct bridge_softc *sc = ifp->if_softc; 1911 1912 BRIDGE_LOCK_ASSERT(sc); 1913 1914 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) 1915 return; 1916 1917 callout_stop(&sc->sc_brcallout); 1918 bstp_stop(&sc->sc_stp); 1919 1920 bridge_rtflush(sc, IFBF_FLUSHDYN); 1921 1922 ifp->if_drv_flags &= ~IFF_DRV_RUNNING; 1923 } 1924 1925 /* 1926 * bridge_enqueue: 1927 * 1928 * Enqueue a packet on a bridge member interface. 1929 * 1930 */ 1931 static int 1932 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m) 1933 { 1934 int len, err = 0; 1935 short mflags; 1936 struct mbuf *m0; 1937 1938 /* We may be sending a fragment so traverse the mbuf */ 1939 for (; m; m = m0) { 1940 m0 = m->m_nextpkt; 1941 m->m_nextpkt = NULL; 1942 len = m->m_pkthdr.len; 1943 mflags = m->m_flags; 1944 1945 /* 1946 * If underlying interface can not do VLAN tag insertion itself 1947 * then attach a packet tag that holds it. 1948 */ 1949 if ((m->m_flags & M_VLANTAG) && 1950 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) { 1951 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag); 1952 if (m == NULL) { 1953 if_printf(dst_ifp, 1954 "unable to prepend VLAN header\n"); 1955 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1); 1956 continue; 1957 } 1958 m->m_flags &= ~M_VLANTAG; 1959 } 1960 1961 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */ 1962 if ((err = dst_ifp->if_transmit(dst_ifp, m))) { 1963 m_freem(m0); 1964 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 1965 break; 1966 } 1967 1968 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1); 1969 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len); 1970 if (mflags & M_MCAST) 1971 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1); 1972 } 1973 1974 return (err); 1975 } 1976 1977 /* 1978 * bridge_dummynet: 1979 * 1980 * Receive a queued packet from dummynet and pass it on to the output 1981 * interface. 1982 * 1983 * The mbuf has the Ethernet header already attached. 1984 */ 1985 static void 1986 bridge_dummynet(struct mbuf *m, struct ifnet *ifp) 1987 { 1988 struct bridge_softc *sc; 1989 1990 sc = ifp->if_bridge; 1991 1992 /* 1993 * The packet didnt originate from a member interface. This should only 1994 * ever happen if a member interface is removed while packets are 1995 * queued for it. 1996 */ 1997 if (sc == NULL) { 1998 m_freem(m); 1999 return; 2000 } 2001 2002 if (PFIL_HOOKED_OUT(V_inet_pfil_head) 2003 #ifdef INET6 2004 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2005 #endif 2006 ) { 2007 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0) 2008 return; 2009 if (m == NULL) 2010 return; 2011 } 2012 2013 bridge_enqueue(sc, ifp, m); 2014 } 2015 2016 /* 2017 * bridge_output: 2018 * 2019 * Send output from a bridge member interface. This 2020 * performs the bridging function for locally originated 2021 * packets. 2022 * 2023 * The mbuf has the Ethernet header already attached. We must 2024 * enqueue or free the mbuf before returning. 2025 */ 2026 static int 2027 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa, 2028 struct rtentry *rt) 2029 { 2030 struct ether_header *eh; 2031 struct ifnet *bifp, *dst_if; 2032 struct bridge_softc *sc; 2033 uint16_t vlan; 2034 2035 if (m->m_len < ETHER_HDR_LEN) { 2036 m = m_pullup(m, ETHER_HDR_LEN); 2037 if (m == NULL) 2038 return (0); 2039 } 2040 2041 eh = mtod(m, struct ether_header *); 2042 sc = ifp->if_bridge; 2043 vlan = VLANTAGOF(m); 2044 2045 BRIDGE_LOCK(sc); 2046 bifp = sc->sc_ifp; 2047 2048 /* 2049 * If bridge is down, but the original output interface is up, 2050 * go ahead and send out that interface. Otherwise, the packet 2051 * is dropped below. 2052 */ 2053 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { 2054 dst_if = ifp; 2055 goto sendunicast; 2056 } 2057 2058 /* 2059 * If the packet is a multicast, or we don't know a better way to 2060 * get there, send to all interfaces. 2061 */ 2062 if (ETHER_IS_MULTICAST(eh->ether_dhost)) 2063 dst_if = NULL; 2064 else 2065 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan); 2066 /* Tap any traffic not passing back out the originating interface */ 2067 if (dst_if != ifp) 2068 ETHER_BPF_MTAP(bifp, m); 2069 if (dst_if == NULL) { 2070 struct bridge_iflist *bif; 2071 struct mbuf *mc; 2072 int error = 0, used = 0; 2073 2074 bridge_span(sc, m); 2075 2076 BRIDGE_LOCK2REF(sc, error); 2077 if (error) { 2078 m_freem(m); 2079 return (0); 2080 } 2081 2082 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 2083 dst_if = bif->bif_ifp; 2084 2085 if (dst_if->if_type == IFT_GIF) 2086 continue; 2087 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2088 continue; 2089 2090 /* 2091 * If this is not the original output interface, 2092 * and the interface is participating in spanning 2093 * tree, make sure the port is in a state that 2094 * allows forwarding. 2095 */ 2096 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) && 2097 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2098 continue; 2099 2100 if (CK_LIST_NEXT(bif, bif_next) == NULL) { 2101 used = 1; 2102 mc = m; 2103 } else { 2104 mc = m_copypacket(m, M_NOWAIT); 2105 if (mc == NULL) { 2106 if_inc_counter(bifp, IFCOUNTER_OERRORS, 1); 2107 continue; 2108 } 2109 } 2110 2111 bridge_enqueue(sc, dst_if, mc); 2112 } 2113 if (used == 0) 2114 m_freem(m); 2115 BRIDGE_UNREF(sc); 2116 return (0); 2117 } 2118 2119 sendunicast: 2120 /* 2121 * XXX Spanning tree consideration here? 2122 */ 2123 2124 bridge_span(sc, m); 2125 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) { 2126 m_freem(m); 2127 BRIDGE_UNLOCK(sc); 2128 return (0); 2129 } 2130 2131 BRIDGE_UNLOCK(sc); 2132 bridge_enqueue(sc, dst_if, m); 2133 return (0); 2134 } 2135 2136 /* 2137 * bridge_transmit: 2138 * 2139 * Do output on a bridge. 2140 * 2141 */ 2142 static int 2143 bridge_transmit(struct ifnet *ifp, struct mbuf *m) 2144 { 2145 struct bridge_softc *sc; 2146 struct ether_header *eh; 2147 struct ifnet *dst_if; 2148 int error = 0; 2149 2150 sc = ifp->if_softc; 2151 2152 ETHER_BPF_MTAP(ifp, m); 2153 2154 eh = mtod(m, struct ether_header *); 2155 2156 BRIDGE_LOCK(sc); 2157 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) && 2158 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) { 2159 BRIDGE_UNLOCK(sc); 2160 error = bridge_enqueue(sc, dst_if, m); 2161 } else 2162 bridge_broadcast(sc, ifp, m, 0); 2163 2164 return (error); 2165 } 2166 2167 /* 2168 * The ifp->if_qflush entry point for if_bridge(4) is no-op. 2169 */ 2170 static void 2171 bridge_qflush(struct ifnet *ifp __unused) 2172 { 2173 } 2174 2175 /* 2176 * bridge_forward: 2177 * 2178 * The forwarding function of the bridge. 2179 * 2180 * NOTE: Releases the lock on return. 2181 */ 2182 static void 2183 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif, 2184 struct mbuf *m) 2185 { 2186 struct bridge_iflist *dbif; 2187 struct ifnet *src_if, *dst_if, *ifp; 2188 struct ether_header *eh; 2189 uint16_t vlan; 2190 uint8_t *dst; 2191 int error; 2192 2193 src_if = m->m_pkthdr.rcvif; 2194 ifp = sc->sc_ifp; 2195 2196 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1); 2197 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); 2198 vlan = VLANTAGOF(m); 2199 2200 if ((sbif->bif_flags & IFBIF_STP) && 2201 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2202 goto drop; 2203 2204 eh = mtod(m, struct ether_header *); 2205 dst = eh->ether_dhost; 2206 2207 /* If the interface is learning, record the address. */ 2208 if (sbif->bif_flags & IFBIF_LEARNING) { 2209 error = bridge_rtupdate(sc, eh->ether_shost, vlan, 2210 sbif, 0, IFBAF_DYNAMIC); 2211 /* 2212 * If the interface has addresses limits then deny any source 2213 * that is not in the cache. 2214 */ 2215 if (error && sbif->bif_addrmax) 2216 goto drop; 2217 } 2218 2219 if ((sbif->bif_flags & IFBIF_STP) != 0 && 2220 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING) 2221 goto drop; 2222 2223 /* 2224 * At this point, the port either doesn't participate 2225 * in spanning tree or it is in the forwarding state. 2226 */ 2227 2228 /* 2229 * If the packet is unicast, destined for someone on 2230 * "this" side of the bridge, drop it. 2231 */ 2232 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) { 2233 dst_if = bridge_rtlookup(sc, dst, vlan); 2234 if (src_if == dst_if) 2235 goto drop; 2236 } else { 2237 /* 2238 * Check if its a reserved multicast address, any address 2239 * listed in 802.1D section 7.12.6 may not be forwarded by the 2240 * bridge. 2241 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F 2242 */ 2243 if (dst[0] == 0x01 && dst[1] == 0x80 && 2244 dst[2] == 0xc2 && dst[3] == 0x00 && 2245 dst[4] == 0x00 && dst[5] <= 0x0f) 2246 goto drop; 2247 2248 /* ...forward it to all interfaces. */ 2249 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1); 2250 dst_if = NULL; 2251 } 2252 2253 /* 2254 * If we have a destination interface which is a member of our bridge, 2255 * OR this is a unicast packet, push it through the bpf(4) machinery. 2256 * For broadcast or multicast packets, don't bother because it will 2257 * be reinjected into ether_input. We do this before we pass the packets 2258 * through the pfil(9) framework, as it is possible that pfil(9) will 2259 * drop the packet, or possibly modify it, making it difficult to debug 2260 * firewall issues on the bridge. 2261 */ 2262 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0) 2263 ETHER_BPF_MTAP(ifp, m); 2264 2265 /* run the packet filter */ 2266 if (PFIL_HOOKED_IN(V_inet_pfil_head) 2267 #ifdef INET6 2268 || PFIL_HOOKED_IN(V_inet6_pfil_head) 2269 #endif 2270 ) { 2271 BRIDGE_UNLOCK(sc); 2272 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0) 2273 return; 2274 if (m == NULL) 2275 return; 2276 BRIDGE_LOCK(sc); 2277 } 2278 2279 if (dst_if == NULL) { 2280 bridge_broadcast(sc, src_if, m, 1); 2281 return; 2282 } 2283 2284 /* 2285 * At this point, we're dealing with a unicast frame 2286 * going to a different interface. 2287 */ 2288 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2289 goto drop; 2290 2291 dbif = bridge_lookup_member_if(sc, dst_if); 2292 if (dbif == NULL) 2293 /* Not a member of the bridge (anymore?) */ 2294 goto drop; 2295 2296 /* Private segments can not talk to each other */ 2297 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE) 2298 goto drop; 2299 2300 if ((dbif->bif_flags & IFBIF_STP) && 2301 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2302 goto drop; 2303 2304 BRIDGE_UNLOCK(sc); 2305 2306 if (PFIL_HOOKED_OUT(V_inet_pfil_head) 2307 #ifdef INET6 2308 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2309 #endif 2310 ) { 2311 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0) 2312 return; 2313 if (m == NULL) 2314 return; 2315 } 2316 2317 bridge_enqueue(sc, dst_if, m); 2318 return; 2319 2320 drop: 2321 BRIDGE_UNLOCK(sc); 2322 m_freem(m); 2323 } 2324 2325 /* 2326 * bridge_input: 2327 * 2328 * Receive input from a member interface. Queue the packet for 2329 * bridging if it is not for us. 2330 */ 2331 static struct mbuf * 2332 bridge_input(struct ifnet *ifp, struct mbuf *m) 2333 { 2334 struct bridge_softc *sc = ifp->if_bridge; 2335 struct bridge_iflist *bif, *bif2; 2336 struct ifnet *bifp; 2337 struct ether_header *eh; 2338 struct mbuf *mc, *mc2; 2339 uint16_t vlan; 2340 int error; 2341 2342 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) 2343 return (m); 2344 2345 bifp = sc->sc_ifp; 2346 vlan = VLANTAGOF(m); 2347 2348 /* 2349 * Implement support for bridge monitoring. If this flag has been 2350 * set on this interface, discard the packet once we push it through 2351 * the bpf(4) machinery, but before we do, increment the byte and 2352 * packet counters associated with this interface. 2353 */ 2354 if ((bifp->if_flags & IFF_MONITOR) != 0) { 2355 m->m_pkthdr.rcvif = bifp; 2356 ETHER_BPF_MTAP(bifp, m); 2357 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); 2358 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); 2359 m_freem(m); 2360 return (NULL); 2361 } 2362 BRIDGE_LOCK(sc); 2363 bif = bridge_lookup_member_if(sc, ifp); 2364 if (bif == NULL) { 2365 BRIDGE_UNLOCK(sc); 2366 return (m); 2367 } 2368 2369 eh = mtod(m, struct ether_header *); 2370 2371 bridge_span(sc, m); 2372 2373 if (m->m_flags & (M_BCAST|M_MCAST)) { 2374 /* Tap off 802.1D packets; they do not get forwarded. */ 2375 if (memcmp(eh->ether_dhost, bstp_etheraddr, 2376 ETHER_ADDR_LEN) == 0) { 2377 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */ 2378 BRIDGE_UNLOCK(sc); 2379 return (NULL); 2380 } 2381 2382 if ((bif->bif_flags & IFBIF_STP) && 2383 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) { 2384 BRIDGE_UNLOCK(sc); 2385 return (m); 2386 } 2387 2388 /* 2389 * Make a deep copy of the packet and enqueue the copy 2390 * for bridge processing; return the original packet for 2391 * local processing. 2392 */ 2393 mc = m_dup(m, M_NOWAIT); 2394 if (mc == NULL) { 2395 BRIDGE_UNLOCK(sc); 2396 return (m); 2397 } 2398 2399 /* Perform the bridge forwarding function with the copy. */ 2400 bridge_forward(sc, bif, mc); 2401 2402 /* 2403 * Reinject the mbuf as arriving on the bridge so we have a 2404 * chance at claiming multicast packets. We can not loop back 2405 * here from ether_input as a bridge is never a member of a 2406 * bridge. 2407 */ 2408 KASSERT(bifp->if_bridge == NULL, 2409 ("loop created in bridge_input")); 2410 mc2 = m_dup(m, M_NOWAIT); 2411 if (mc2 != NULL) { 2412 /* Keep the layer3 header aligned */ 2413 int i = min(mc2->m_pkthdr.len, max_protohdr); 2414 mc2 = m_copyup(mc2, i, ETHER_ALIGN); 2415 } 2416 if (mc2 != NULL) { 2417 mc2->m_pkthdr.rcvif = bifp; 2418 (*bifp->if_input)(bifp, mc2); 2419 } 2420 2421 /* Return the original packet for local processing. */ 2422 return (m); 2423 } 2424 2425 if ((bif->bif_flags & IFBIF_STP) && 2426 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) { 2427 BRIDGE_UNLOCK(sc); 2428 return (m); 2429 } 2430 2431 #if (defined(INET) || defined(INET6)) 2432 # define OR_CARP_CHECK_WE_ARE_DST(iface) \ 2433 || ((iface)->if_carp \ 2434 && (*carp_forus_p)((iface), eh->ether_dhost)) 2435 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \ 2436 || ((iface)->if_carp \ 2437 && (*carp_forus_p)((iface), eh->ether_shost)) 2438 #else 2439 # define OR_CARP_CHECK_WE_ARE_DST(iface) 2440 # define OR_CARP_CHECK_WE_ARE_SRC(iface) 2441 #endif 2442 2443 #ifdef INET6 2444 # define OR_PFIL_HOOKED_INET6 \ 2445 || PFIL_HOOKED_IN(V_inet6_pfil_head) 2446 #else 2447 # define OR_PFIL_HOOKED_INET6 2448 #endif 2449 2450 #define GRAB_OUR_PACKETS(iface) \ 2451 if ((iface)->if_type == IFT_GIF) \ 2452 continue; \ 2453 /* It is destined for us. */ \ 2454 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \ 2455 OR_CARP_CHECK_WE_ARE_DST((iface)) \ 2456 ) { \ 2457 if (bif->bif_flags & IFBIF_LEARNING) { \ 2458 error = bridge_rtupdate(sc, eh->ether_shost, \ 2459 vlan, bif, 0, IFBAF_DYNAMIC); \ 2460 if (error && bif->bif_addrmax) { \ 2461 BRIDGE_UNLOCK(sc); \ 2462 m_freem(m); \ 2463 return (NULL); \ 2464 } \ 2465 } \ 2466 m->m_pkthdr.rcvif = iface; \ 2467 if ((iface) == ifp) { \ 2468 /* Skip bridge processing... src == dest */ \ 2469 BRIDGE_UNLOCK(sc); \ 2470 return (m); \ 2471 } \ 2472 /* It's passing over or to the bridge, locally. */ \ 2473 ETHER_BPF_MTAP(bifp, m); \ 2474 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \ 2475 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \ 2476 /* Filter on the physical interface. */ \ 2477 if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \ 2478 OR_PFIL_HOOKED_INET6)) { \ 2479 if (bridge_pfil(&m, NULL, ifp, \ 2480 PFIL_IN) != 0 || m == NULL) { \ 2481 BRIDGE_UNLOCK(sc); \ 2482 return (NULL); \ 2483 } \ 2484 } \ 2485 if ((iface) != bifp) \ 2486 ETHER_BPF_MTAP(iface, m); \ 2487 BRIDGE_UNLOCK(sc); \ 2488 return (m); \ 2489 } \ 2490 \ 2491 /* We just received a packet that we sent out. */ \ 2492 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \ 2493 OR_CARP_CHECK_WE_ARE_SRC((iface)) \ 2494 ) { \ 2495 BRIDGE_UNLOCK(sc); \ 2496 m_freem(m); \ 2497 return (NULL); \ 2498 } 2499 2500 /* 2501 * Unicast. Make sure it's not for the bridge. 2502 */ 2503 do { GRAB_OUR_PACKETS(bifp) } while (0); 2504 2505 /* 2506 * Give a chance for ifp at first priority. This will help when the 2507 * packet comes through the interface like VLAN's with the same MACs 2508 * on several interfaces from the same bridge. This also will save 2509 * some CPU cycles in case the destination interface and the input 2510 * interface (eq ifp) are the same. 2511 */ 2512 do { GRAB_OUR_PACKETS(ifp) } while (0); 2513 2514 /* Now check the all bridge members. */ 2515 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) { 2516 GRAB_OUR_PACKETS(bif2->bif_ifp) 2517 } 2518 2519 #undef OR_CARP_CHECK_WE_ARE_DST 2520 #undef OR_CARP_CHECK_WE_ARE_SRC 2521 #undef OR_PFIL_HOOKED_INET6 2522 #undef GRAB_OUR_PACKETS 2523 2524 /* Perform the bridge forwarding function. */ 2525 bridge_forward(sc, bif, m); 2526 2527 return (NULL); 2528 } 2529 2530 /* 2531 * bridge_broadcast: 2532 * 2533 * Send a frame to all interfaces that are members of 2534 * the bridge, except for the one on which the packet 2535 * arrived. 2536 * 2537 * NOTE: Releases the lock on return. 2538 */ 2539 static void 2540 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if, 2541 struct mbuf *m, int runfilt) 2542 { 2543 struct bridge_iflist *dbif, *sbif; 2544 struct mbuf *mc; 2545 struct ifnet *dst_if; 2546 int error = 0, used = 0, i; 2547 2548 sbif = bridge_lookup_member_if(sc, src_if); 2549 2550 BRIDGE_LOCK2REF(sc, error); 2551 if (error) { 2552 m_freem(m); 2553 return; 2554 } 2555 2556 /* Filter on the bridge interface before broadcasting */ 2557 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head) 2558 #ifdef INET6 2559 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2560 #endif 2561 )) { 2562 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0) 2563 goto out; 2564 if (m == NULL) 2565 goto out; 2566 } 2567 2568 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) { 2569 dst_if = dbif->bif_ifp; 2570 if (dst_if == src_if) 2571 continue; 2572 2573 /* Private segments can not talk to each other */ 2574 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)) 2575 continue; 2576 2577 if ((dbif->bif_flags & IFBIF_STP) && 2578 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2579 continue; 2580 2581 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 && 2582 (m->m_flags & (M_BCAST|M_MCAST)) == 0) 2583 continue; 2584 2585 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2586 continue; 2587 2588 if (CK_LIST_NEXT(dbif, bif_next) == NULL) { 2589 mc = m; 2590 used = 1; 2591 } else { 2592 mc = m_dup(m, M_NOWAIT); 2593 if (mc == NULL) { 2594 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 2595 continue; 2596 } 2597 } 2598 2599 /* 2600 * Filter on the output interface. Pass a NULL bridge interface 2601 * pointer so we do not redundantly filter on the bridge for 2602 * each interface we broadcast on. 2603 */ 2604 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head) 2605 #ifdef INET6 2606 || PFIL_HOOKED_OUT(V_inet6_pfil_head) 2607 #endif 2608 )) { 2609 if (used == 0) { 2610 /* Keep the layer3 header aligned */ 2611 i = min(mc->m_pkthdr.len, max_protohdr); 2612 mc = m_copyup(mc, i, ETHER_ALIGN); 2613 if (mc == NULL) { 2614 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 2615 continue; 2616 } 2617 } 2618 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0) 2619 continue; 2620 if (mc == NULL) 2621 continue; 2622 } 2623 2624 bridge_enqueue(sc, dst_if, mc); 2625 } 2626 if (used == 0) 2627 m_freem(m); 2628 2629 out: 2630 BRIDGE_UNREF(sc); 2631 } 2632 2633 /* 2634 * bridge_span: 2635 * 2636 * Duplicate a packet out one or more interfaces that are in span mode, 2637 * the original mbuf is unmodified. 2638 */ 2639 static void 2640 bridge_span(struct bridge_softc *sc, struct mbuf *m) 2641 { 2642 struct bridge_iflist *bif; 2643 struct ifnet *dst_if; 2644 struct mbuf *mc; 2645 2646 if (CK_LIST_EMPTY(&sc->sc_spanlist)) 2647 return; 2648 2649 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) { 2650 dst_if = bif->bif_ifp; 2651 2652 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2653 continue; 2654 2655 mc = m_copypacket(m, M_NOWAIT); 2656 if (mc == NULL) { 2657 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 2658 continue; 2659 } 2660 2661 bridge_enqueue(sc, dst_if, mc); 2662 } 2663 } 2664 2665 /* 2666 * bridge_rtupdate: 2667 * 2668 * Add a bridge routing entry. 2669 */ 2670 static int 2671 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan, 2672 struct bridge_iflist *bif, int setflags, uint8_t flags) 2673 { 2674 struct bridge_rtnode *brt; 2675 int error; 2676 2677 BRIDGE_LOCK_ASSERT(sc); 2678 2679 /* Check the source address is valid and not multicast. */ 2680 if (ETHER_IS_MULTICAST(dst) || 2681 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 && 2682 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0) 2683 return (EINVAL); 2684 2685 /* 802.1p frames map to vlan 1 */ 2686 if (vlan == 0) 2687 vlan = 1; 2688 2689 /* 2690 * A route for this destination might already exist. If so, 2691 * update it, otherwise create a new one. 2692 */ 2693 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) { 2694 if (sc->sc_brtcnt >= sc->sc_brtmax) { 2695 sc->sc_brtexceeded++; 2696 return (ENOSPC); 2697 } 2698 /* Check per interface address limits (if enabled) */ 2699 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) { 2700 bif->bif_addrexceeded++; 2701 return (ENOSPC); 2702 } 2703 2704 /* 2705 * Allocate a new bridge forwarding node, and 2706 * initialize the expiration time and Ethernet 2707 * address. 2708 */ 2709 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO); 2710 if (brt == NULL) 2711 return (ENOMEM); 2712 2713 if (bif->bif_flags & IFBIF_STICKY) 2714 brt->brt_flags = IFBAF_STICKY; 2715 else 2716 brt->brt_flags = IFBAF_DYNAMIC; 2717 2718 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN); 2719 brt->brt_vlan = vlan; 2720 2721 if ((error = bridge_rtnode_insert(sc, brt)) != 0) { 2722 uma_zfree(V_bridge_rtnode_zone, brt); 2723 return (error); 2724 } 2725 brt->brt_dst = bif; 2726 bif->bif_addrcnt++; 2727 } 2728 2729 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 2730 brt->brt_dst != bif) { 2731 brt->brt_dst->bif_addrcnt--; 2732 brt->brt_dst = bif; 2733 brt->brt_dst->bif_addrcnt++; 2734 } 2735 2736 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 2737 brt->brt_expire = time_uptime + sc->sc_brttimeout; 2738 if (setflags) 2739 brt->brt_flags = flags; 2740 2741 return (0); 2742 } 2743 2744 /* 2745 * bridge_rtlookup: 2746 * 2747 * Lookup the destination interface for an address. 2748 */ 2749 static struct ifnet * 2750 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) 2751 { 2752 struct bridge_rtnode *brt; 2753 2754 BRIDGE_LOCK_ASSERT(sc); 2755 2756 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL) 2757 return (NULL); 2758 2759 return (brt->brt_ifp); 2760 } 2761 2762 /* 2763 * bridge_rttrim: 2764 * 2765 * Trim the routine table so that we have a number 2766 * of routing entries less than or equal to the 2767 * maximum number. 2768 */ 2769 static void 2770 bridge_rttrim(struct bridge_softc *sc) 2771 { 2772 struct bridge_rtnode *brt, *nbrt; 2773 2774 BRIDGE_LOCK_ASSERT(sc); 2775 2776 /* Make sure we actually need to do this. */ 2777 if (sc->sc_brtcnt <= sc->sc_brtmax) 2778 return; 2779 2780 /* Force an aging cycle; this might trim enough addresses. */ 2781 bridge_rtage(sc); 2782 if (sc->sc_brtcnt <= sc->sc_brtmax) 2783 return; 2784 2785 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 2786 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { 2787 bridge_rtnode_destroy(sc, brt); 2788 if (sc->sc_brtcnt <= sc->sc_brtmax) 2789 return; 2790 } 2791 } 2792 } 2793 2794 /* 2795 * bridge_timer: 2796 * 2797 * Aging timer for the bridge. 2798 */ 2799 static void 2800 bridge_timer(void *arg) 2801 { 2802 struct bridge_softc *sc = arg; 2803 2804 BRIDGE_LOCK_ASSERT(sc); 2805 2806 /* Destruction of rtnodes requires a proper vnet context */ 2807 CURVNET_SET(sc->sc_ifp->if_vnet); 2808 bridge_rtage(sc); 2809 2810 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) 2811 callout_reset(&sc->sc_brcallout, 2812 bridge_rtable_prune_period * hz, bridge_timer, sc); 2813 CURVNET_RESTORE(); 2814 } 2815 2816 /* 2817 * bridge_rtage: 2818 * 2819 * Perform an aging cycle. 2820 */ 2821 static void 2822 bridge_rtage(struct bridge_softc *sc) 2823 { 2824 struct bridge_rtnode *brt, *nbrt; 2825 2826 BRIDGE_LOCK_ASSERT(sc); 2827 2828 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 2829 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { 2830 if (time_uptime >= brt->brt_expire) 2831 bridge_rtnode_destroy(sc, brt); 2832 } 2833 } 2834 } 2835 2836 /* 2837 * bridge_rtflush: 2838 * 2839 * Remove all dynamic addresses from the bridge. 2840 */ 2841 static void 2842 bridge_rtflush(struct bridge_softc *sc, int full) 2843 { 2844 struct bridge_rtnode *brt, *nbrt; 2845 2846 BRIDGE_LOCK_ASSERT(sc); 2847 2848 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 2849 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 2850 bridge_rtnode_destroy(sc, brt); 2851 } 2852 } 2853 2854 /* 2855 * bridge_rtdaddr: 2856 * 2857 * Remove an address from the table. 2858 */ 2859 static int 2860 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) 2861 { 2862 struct bridge_rtnode *brt; 2863 int found = 0; 2864 2865 BRIDGE_LOCK_ASSERT(sc); 2866 2867 /* 2868 * If vlan is zero then we want to delete for all vlans so the lookup 2869 * may return more than one. 2870 */ 2871 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) { 2872 bridge_rtnode_destroy(sc, brt); 2873 found = 1; 2874 } 2875 2876 return (found ? 0 : ENOENT); 2877 } 2878 2879 /* 2880 * bridge_rtdelete: 2881 * 2882 * Delete routes to a speicifc member interface. 2883 */ 2884 static void 2885 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full) 2886 { 2887 struct bridge_rtnode *brt, *nbrt; 2888 2889 BRIDGE_LOCK_ASSERT(sc); 2890 2891 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 2892 if (brt->brt_ifp == ifp && (full || 2893 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)) 2894 bridge_rtnode_destroy(sc, brt); 2895 } 2896 } 2897 2898 /* 2899 * bridge_rtable_init: 2900 * 2901 * Initialize the route table for this bridge. 2902 */ 2903 static void 2904 bridge_rtable_init(struct bridge_softc *sc) 2905 { 2906 int i; 2907 2908 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE, 2909 M_DEVBUF, M_WAITOK); 2910 2911 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++) 2912 CK_LIST_INIT(&sc->sc_rthash[i]); 2913 2914 sc->sc_rthash_key = arc4random(); 2915 CK_LIST_INIT(&sc->sc_rtlist); 2916 } 2917 2918 /* 2919 * bridge_rtable_fini: 2920 * 2921 * Deconstruct the route table for this bridge. 2922 */ 2923 static void 2924 bridge_rtable_fini(struct bridge_softc *sc) 2925 { 2926 2927 KASSERT(sc->sc_brtcnt == 0, 2928 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt)); 2929 free(sc->sc_rthash, M_DEVBUF); 2930 } 2931 2932 /* 2933 * The following hash function is adapted from "Hash Functions" by Bob Jenkins 2934 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997). 2935 */ 2936 #define mix(a, b, c) \ 2937 do { \ 2938 a -= b; a -= c; a ^= (c >> 13); \ 2939 b -= c; b -= a; b ^= (a << 8); \ 2940 c -= a; c -= b; c ^= (b >> 13); \ 2941 a -= b; a -= c; a ^= (c >> 12); \ 2942 b -= c; b -= a; b ^= (a << 16); \ 2943 c -= a; c -= b; c ^= (b >> 5); \ 2944 a -= b; a -= c; a ^= (c >> 3); \ 2945 b -= c; b -= a; b ^= (a << 10); \ 2946 c -= a; c -= b; c ^= (b >> 15); \ 2947 } while (/*CONSTCOND*/0) 2948 2949 static __inline uint32_t 2950 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr) 2951 { 2952 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key; 2953 2954 b += addr[5] << 8; 2955 b += addr[4]; 2956 a += addr[3] << 24; 2957 a += addr[2] << 16; 2958 a += addr[1] << 8; 2959 a += addr[0]; 2960 2961 mix(a, b, c); 2962 2963 return (c & BRIDGE_RTHASH_MASK); 2964 } 2965 2966 #undef mix 2967 2968 static int 2969 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b) 2970 { 2971 int i, d; 2972 2973 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) { 2974 d = ((int)a[i]) - ((int)b[i]); 2975 } 2976 2977 return (d); 2978 } 2979 2980 /* 2981 * bridge_rtnode_lookup: 2982 * 2983 * Look up a bridge route node for the specified destination. Compare the 2984 * vlan id or if zero then just return the first match. 2985 */ 2986 static struct bridge_rtnode * 2987 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) 2988 { 2989 struct bridge_rtnode *brt; 2990 uint32_t hash; 2991 int dir; 2992 2993 BRIDGE_LOCK_ASSERT(sc); 2994 2995 hash = bridge_rthash(sc, addr); 2996 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) { 2997 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr); 2998 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0)) 2999 return (brt); 3000 if (dir > 0) 3001 return (NULL); 3002 } 3003 3004 return (NULL); 3005 } 3006 3007 /* 3008 * bridge_rtnode_insert: 3009 * 3010 * Insert the specified bridge node into the route table. We 3011 * assume the entry is not already in the table. 3012 */ 3013 static int 3014 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt) 3015 { 3016 struct bridge_rtnode *lbrt; 3017 uint32_t hash; 3018 int dir; 3019 3020 BRIDGE_LOCK_ASSERT(sc); 3021 3022 hash = bridge_rthash(sc, brt->brt_addr); 3023 3024 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]); 3025 if (lbrt == NULL) { 3026 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash); 3027 goto out; 3028 } 3029 3030 do { 3031 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr); 3032 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan) 3033 return (EEXIST); 3034 if (dir > 0) { 3035 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash); 3036 goto out; 3037 } 3038 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) { 3039 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash); 3040 goto out; 3041 } 3042 lbrt = CK_LIST_NEXT(lbrt, brt_hash); 3043 } while (lbrt != NULL); 3044 3045 #ifdef DIAGNOSTIC 3046 panic("bridge_rtnode_insert: impossible"); 3047 #endif 3048 3049 out: 3050 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list); 3051 sc->sc_brtcnt++; 3052 3053 return (0); 3054 } 3055 3056 /* 3057 * bridge_rtnode_destroy: 3058 * 3059 * Destroy a bridge rtnode. 3060 */ 3061 static void 3062 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt) 3063 { 3064 BRIDGE_LOCK_ASSERT(sc); 3065 3066 CK_LIST_REMOVE(brt, brt_hash); 3067 3068 CK_LIST_REMOVE(brt, brt_list); 3069 sc->sc_brtcnt--; 3070 brt->brt_dst->bif_addrcnt--; 3071 uma_zfree(V_bridge_rtnode_zone, brt); 3072 } 3073 3074 /* 3075 * bridge_rtable_expire: 3076 * 3077 * Set the expiry time for all routes on an interface. 3078 */ 3079 static void 3080 bridge_rtable_expire(struct ifnet *ifp, int age) 3081 { 3082 struct bridge_softc *sc = ifp->if_bridge; 3083 struct bridge_rtnode *brt; 3084 3085 CURVNET_SET(ifp->if_vnet); 3086 BRIDGE_LOCK(sc); 3087 3088 /* 3089 * If the age is zero then flush, otherwise set all the expiry times to 3090 * age for the interface 3091 */ 3092 if (age == 0) 3093 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN); 3094 else { 3095 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) { 3096 /* Cap the expiry time to 'age' */ 3097 if (brt->brt_ifp == ifp && 3098 brt->brt_expire > time_uptime + age && 3099 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 3100 brt->brt_expire = time_uptime + age; 3101 } 3102 } 3103 BRIDGE_UNLOCK(sc); 3104 CURVNET_RESTORE(); 3105 } 3106 3107 /* 3108 * bridge_state_change: 3109 * 3110 * Callback from the bridgestp code when a port changes states. 3111 */ 3112 static void 3113 bridge_state_change(struct ifnet *ifp, int state) 3114 { 3115 struct bridge_softc *sc = ifp->if_bridge; 3116 static const char *stpstates[] = { 3117 "disabled", 3118 "listening", 3119 "learning", 3120 "forwarding", 3121 "blocking", 3122 "discarding" 3123 }; 3124 3125 CURVNET_SET(ifp->if_vnet); 3126 if (V_log_stp) 3127 log(LOG_NOTICE, "%s: state changed to %s on %s\n", 3128 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname); 3129 CURVNET_RESTORE(); 3130 } 3131 3132 /* 3133 * Send bridge packets through pfil if they are one of the types pfil can deal 3134 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without 3135 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for 3136 * that interface. 3137 */ 3138 static int 3139 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir) 3140 { 3141 int snap, error, i, hlen; 3142 struct ether_header *eh1, eh2; 3143 struct ip *ip; 3144 struct llc llc1; 3145 u_int16_t ether_type; 3146 pfil_return_t rv; 3147 3148 snap = 0; 3149 error = -1; /* Default error if not error == 0 */ 3150 3151 #if 0 3152 /* we may return with the IP fields swapped, ensure its not shared */ 3153 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__)); 3154 #endif 3155 3156 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0) 3157 return (0); /* filtering is disabled */ 3158 3159 i = min((*mp)->m_pkthdr.len, max_protohdr); 3160 if ((*mp)->m_len < i) { 3161 *mp = m_pullup(*mp, i); 3162 if (*mp == NULL) { 3163 printf("%s: m_pullup failed\n", __func__); 3164 return (-1); 3165 } 3166 } 3167 3168 eh1 = mtod(*mp, struct ether_header *); 3169 ether_type = ntohs(eh1->ether_type); 3170 3171 /* 3172 * Check for SNAP/LLC. 3173 */ 3174 if (ether_type < ETHERMTU) { 3175 struct llc *llc2 = (struct llc *)(eh1 + 1); 3176 3177 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 && 3178 llc2->llc_dsap == LLC_SNAP_LSAP && 3179 llc2->llc_ssap == LLC_SNAP_LSAP && 3180 llc2->llc_control == LLC_UI) { 3181 ether_type = htons(llc2->llc_un.type_snap.ether_type); 3182 snap = 1; 3183 } 3184 } 3185 3186 /* 3187 * If we're trying to filter bridge traffic, don't look at anything 3188 * other than IP and ARP traffic. If the filter doesn't understand 3189 * IPv6, don't allow IPv6 through the bridge either. This is lame 3190 * since if we really wanted, say, an AppleTalk filter, we are hosed, 3191 * but of course we don't have an AppleTalk filter to begin with. 3192 * (Note that since pfil doesn't understand ARP it will pass *ALL* 3193 * ARP traffic.) 3194 */ 3195 switch (ether_type) { 3196 case ETHERTYPE_ARP: 3197 case ETHERTYPE_REVARP: 3198 if (V_pfil_ipfw_arp == 0) 3199 return (0); /* Automatically pass */ 3200 break; 3201 3202 case ETHERTYPE_IP: 3203 #ifdef INET6 3204 case ETHERTYPE_IPV6: 3205 #endif /* INET6 */ 3206 break; 3207 default: 3208 /* 3209 * Check to see if the user wants to pass non-ip 3210 * packets, these will not be checked by pfil(9) and 3211 * passed unconditionally so the default is to drop. 3212 */ 3213 if (V_pfil_onlyip) 3214 goto bad; 3215 } 3216 3217 /* Run the packet through pfil before stripping link headers */ 3218 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 && 3219 dir == PFIL_OUT && ifp != NULL) { 3220 switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) { 3221 case PFIL_DROPPED: 3222 return (EACCES); 3223 case PFIL_CONSUMED: 3224 return (0); 3225 } 3226 } 3227 3228 /* Strip off the Ethernet header and keep a copy. */ 3229 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2); 3230 m_adj(*mp, ETHER_HDR_LEN); 3231 3232 /* Strip off snap header, if present */ 3233 if (snap) { 3234 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1); 3235 m_adj(*mp, sizeof(struct llc)); 3236 } 3237 3238 /* 3239 * Check the IP header for alignment and errors 3240 */ 3241 if (dir == PFIL_IN) { 3242 switch (ether_type) { 3243 case ETHERTYPE_IP: 3244 error = bridge_ip_checkbasic(mp); 3245 break; 3246 #ifdef INET6 3247 case ETHERTYPE_IPV6: 3248 error = bridge_ip6_checkbasic(mp); 3249 break; 3250 #endif /* INET6 */ 3251 default: 3252 error = 0; 3253 } 3254 if (error) 3255 goto bad; 3256 } 3257 3258 error = 0; 3259 3260 /* 3261 * Run the packet through pfil 3262 */ 3263 rv = PFIL_PASS; 3264 switch (ether_type) { 3265 case ETHERTYPE_IP: 3266 /* 3267 * Run pfil on the member interface and the bridge, both can 3268 * be skipped by clearing pfil_member or pfil_bridge. 3269 * 3270 * Keep the order: 3271 * in_if -> bridge_if -> out_if 3272 */ 3273 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv = 3274 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) != 3275 PFIL_PASS) 3276 break; 3277 3278 if (V_pfil_member && ifp != NULL && (rv = 3279 pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) != 3280 PFIL_PASS) 3281 break; 3282 3283 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv = 3284 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) != 3285 PFIL_PASS) 3286 break; 3287 3288 /* check if we need to fragment the packet */ 3289 /* bridge_fragment generates a mbuf chain of packets */ 3290 /* that already include eth headers */ 3291 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) { 3292 i = (*mp)->m_pkthdr.len; 3293 if (i > ifp->if_mtu) { 3294 error = bridge_fragment(ifp, mp, &eh2, snap, 3295 &llc1); 3296 return (error); 3297 } 3298 } 3299 3300 /* Recalculate the ip checksum. */ 3301 ip = mtod(*mp, struct ip *); 3302 hlen = ip->ip_hl << 2; 3303 if (hlen < sizeof(struct ip)) 3304 goto bad; 3305 if (hlen > (*mp)->m_len) { 3306 if ((*mp = m_pullup(*mp, hlen)) == NULL) 3307 goto bad; 3308 ip = mtod(*mp, struct ip *); 3309 if (ip == NULL) 3310 goto bad; 3311 } 3312 ip->ip_sum = 0; 3313 if (hlen == sizeof(struct ip)) 3314 ip->ip_sum = in_cksum_hdr(ip); 3315 else 3316 ip->ip_sum = in_cksum(*mp, hlen); 3317 3318 break; 3319 #ifdef INET6 3320 case ETHERTYPE_IPV6: 3321 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv = 3322 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) != 3323 PFIL_PASS) 3324 break; 3325 3326 if (V_pfil_member && ifp != NULL && (rv = 3327 pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) != 3328 PFIL_PASS) 3329 break; 3330 3331 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv = 3332 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) != 3333 PFIL_PASS) 3334 break; 3335 break; 3336 #endif 3337 } 3338 3339 switch (rv) { 3340 case PFIL_CONSUMED: 3341 return (0); 3342 case PFIL_DROPPED: 3343 return (EACCES); 3344 default: 3345 break; 3346 } 3347 3348 error = -1; 3349 3350 /* 3351 * Finally, put everything back the way it was and return 3352 */ 3353 if (snap) { 3354 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT); 3355 if (*mp == NULL) 3356 return (error); 3357 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc)); 3358 } 3359 3360 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT); 3361 if (*mp == NULL) 3362 return (error); 3363 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN); 3364 3365 return (0); 3366 3367 bad: 3368 m_freem(*mp); 3369 *mp = NULL; 3370 return (error); 3371 } 3372 3373 /* 3374 * Perform basic checks on header size since 3375 * pfil assumes ip_input has already processed 3376 * it for it. Cut-and-pasted from ip_input.c. 3377 * Given how simple the IPv6 version is, 3378 * does the IPv4 version really need to be 3379 * this complicated? 3380 * 3381 * XXX Should we update ipstat here, or not? 3382 * XXX Right now we update ipstat but not 3383 * XXX csum_counter. 3384 */ 3385 static int 3386 bridge_ip_checkbasic(struct mbuf **mp) 3387 { 3388 struct mbuf *m = *mp; 3389 struct ip *ip; 3390 int len, hlen; 3391 u_short sum; 3392 3393 if (*mp == NULL) 3394 return (-1); 3395 3396 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 3397 if ((m = m_copyup(m, sizeof(struct ip), 3398 (max_linkhdr + 3) & ~3)) == NULL) { 3399 /* XXXJRT new stat, please */ 3400 KMOD_IPSTAT_INC(ips_toosmall); 3401 goto bad; 3402 } 3403 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 3404 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 3405 KMOD_IPSTAT_INC(ips_toosmall); 3406 goto bad; 3407 } 3408 } 3409 ip = mtod(m, struct ip *); 3410 if (ip == NULL) goto bad; 3411 3412 if (ip->ip_v != IPVERSION) { 3413 KMOD_IPSTAT_INC(ips_badvers); 3414 goto bad; 3415 } 3416 hlen = ip->ip_hl << 2; 3417 if (hlen < sizeof(struct ip)) { /* minimum header length */ 3418 KMOD_IPSTAT_INC(ips_badhlen); 3419 goto bad; 3420 } 3421 if (hlen > m->m_len) { 3422 if ((m = m_pullup(m, hlen)) == NULL) { 3423 KMOD_IPSTAT_INC(ips_badhlen); 3424 goto bad; 3425 } 3426 ip = mtod(m, struct ip *); 3427 if (ip == NULL) goto bad; 3428 } 3429 3430 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) { 3431 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID); 3432 } else { 3433 if (hlen == sizeof(struct ip)) { 3434 sum = in_cksum_hdr(ip); 3435 } else { 3436 sum = in_cksum(m, hlen); 3437 } 3438 } 3439 if (sum) { 3440 KMOD_IPSTAT_INC(ips_badsum); 3441 goto bad; 3442 } 3443 3444 /* Retrieve the packet length. */ 3445 len = ntohs(ip->ip_len); 3446 3447 /* 3448 * Check for additional length bogosity 3449 */ 3450 if (len < hlen) { 3451 KMOD_IPSTAT_INC(ips_badlen); 3452 goto bad; 3453 } 3454 3455 /* 3456 * Check that the amount of data in the buffers 3457 * is as at least much as the IP header would have us expect. 3458 * Drop packet if shorter than we expect. 3459 */ 3460 if (m->m_pkthdr.len < len) { 3461 KMOD_IPSTAT_INC(ips_tooshort); 3462 goto bad; 3463 } 3464 3465 /* Checks out, proceed */ 3466 *mp = m; 3467 return (0); 3468 3469 bad: 3470 *mp = m; 3471 return (-1); 3472 } 3473 3474 #ifdef INET6 3475 /* 3476 * Same as above, but for IPv6. 3477 * Cut-and-pasted from ip6_input.c. 3478 * XXX Should we update ip6stat, or not? 3479 */ 3480 static int 3481 bridge_ip6_checkbasic(struct mbuf **mp) 3482 { 3483 struct mbuf *m = *mp; 3484 struct ip6_hdr *ip6; 3485 3486 /* 3487 * If the IPv6 header is not aligned, slurp it up into a new 3488 * mbuf with space for link headers, in the event we forward 3489 * it. Otherwise, if it is aligned, make sure the entire base 3490 * IPv6 header is in the first mbuf of the chain. 3491 */ 3492 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 3493 struct ifnet *inifp = m->m_pkthdr.rcvif; 3494 if ((m = m_copyup(m, sizeof(struct ip6_hdr), 3495 (max_linkhdr + 3) & ~3)) == NULL) { 3496 /* XXXJRT new stat, please */ 3497 IP6STAT_INC(ip6s_toosmall); 3498 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 3499 goto bad; 3500 } 3501 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) { 3502 struct ifnet *inifp = m->m_pkthdr.rcvif; 3503 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 3504 IP6STAT_INC(ip6s_toosmall); 3505 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 3506 goto bad; 3507 } 3508 } 3509 3510 ip6 = mtod(m, struct ip6_hdr *); 3511 3512 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 3513 IP6STAT_INC(ip6s_badvers); 3514 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 3515 goto bad; 3516 } 3517 3518 /* Checks out, proceed */ 3519 *mp = m; 3520 return (0); 3521 3522 bad: 3523 *mp = m; 3524 return (-1); 3525 } 3526 #endif /* INET6 */ 3527 3528 /* 3529 * bridge_fragment: 3530 * 3531 * Fragment mbuf chain in multiple packets and prepend ethernet header. 3532 */ 3533 static int 3534 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh, 3535 int snap, struct llc *llc) 3536 { 3537 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL; 3538 struct ip *ip; 3539 int error = -1; 3540 3541 if (m->m_len < sizeof(struct ip) && 3542 (m = m_pullup(m, sizeof(struct ip))) == NULL) 3543 goto dropit; 3544 ip = mtod(m, struct ip *); 3545 3546 m->m_pkthdr.csum_flags |= CSUM_IP; 3547 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist); 3548 if (error) 3549 goto dropit; 3550 3551 /* 3552 * Walk the chain and re-add the Ethernet header for 3553 * each mbuf packet. 3554 */ 3555 for (mcur = m; mcur; mcur = mcur->m_nextpkt) { 3556 nextpkt = mcur->m_nextpkt; 3557 mcur->m_nextpkt = NULL; 3558 if (snap) { 3559 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT); 3560 if (mcur == NULL) { 3561 error = ENOBUFS; 3562 if (mprev != NULL) 3563 mprev->m_nextpkt = nextpkt; 3564 goto dropit; 3565 } 3566 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc)); 3567 } 3568 3569 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT); 3570 if (mcur == NULL) { 3571 error = ENOBUFS; 3572 if (mprev != NULL) 3573 mprev->m_nextpkt = nextpkt; 3574 goto dropit; 3575 } 3576 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN); 3577 3578 /* 3579 * The previous two M_PREPEND could have inserted one or two 3580 * mbufs in front so we have to update the previous packet's 3581 * m_nextpkt. 3582 */ 3583 mcur->m_nextpkt = nextpkt; 3584 if (mprev != NULL) 3585 mprev->m_nextpkt = mcur; 3586 else { 3587 /* The first mbuf in the original chain needs to be 3588 * updated. */ 3589 *mp = mcur; 3590 } 3591 mprev = mcur; 3592 } 3593 3594 KMOD_IPSTAT_INC(ips_fragmented); 3595 return (error); 3596 3597 dropit: 3598 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */ 3599 m = mcur->m_nextpkt; 3600 m_freem(mcur); 3601 } 3602 return (error); 3603 } 3604 3605 static void 3606 bridge_linkstate(struct ifnet *ifp) 3607 { 3608 struct bridge_softc *sc = ifp->if_bridge; 3609 struct bridge_iflist *bif; 3610 3611 BRIDGE_LOCK(sc); 3612 bif = bridge_lookup_member_if(sc, ifp); 3613 if (bif == NULL) { 3614 BRIDGE_UNLOCK(sc); 3615 return; 3616 } 3617 bridge_linkcheck(sc); 3618 BRIDGE_UNLOCK(sc); 3619 3620 bstp_linkstate(&bif->bif_stp); 3621 } 3622 3623 static void 3624 bridge_linkcheck(struct bridge_softc *sc) 3625 { 3626 struct bridge_iflist *bif; 3627 int new_link, hasls; 3628 3629 BRIDGE_LOCK_ASSERT(sc); 3630 new_link = LINK_STATE_DOWN; 3631 hasls = 0; 3632 /* Our link is considered up if at least one of our ports is active */ 3633 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 3634 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE) 3635 hasls++; 3636 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) { 3637 new_link = LINK_STATE_UP; 3638 break; 3639 } 3640 } 3641 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) { 3642 /* If no interfaces support link-state then we default to up */ 3643 new_link = LINK_STATE_UP; 3644 } 3645 if_link_state_change(sc->sc_ifp, new_link); 3646 } 3647