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 "opt_inet.h" 80 #include "opt_inet6.h" 81 82 #define EXTERR_CATEGORY EXTERR_CAT_BRIDGE 83 84 #include <sys/param.h> 85 #include <sys/ctype.h> /* string functions */ 86 #include <sys/eventhandler.h> 87 #include <sys/exterrvar.h> 88 #include <sys/jail.h> 89 #include <sys/kernel.h> 90 #include <sys/lock.h> 91 #include <sys/malloc.h> 92 #include <sys/mbuf.h> 93 #include <sys/module.h> 94 #include <sys/mutex.h> 95 #include <sys/priv.h> 96 #include <sys/proc.h> 97 #include <sys/protosw.h> 98 #include <sys/random.h> 99 #include <sys/systm.h> 100 #include <sys/socket.h> /* for net/if.h */ 101 #include <sys/sockio.h> 102 #include <sys/syslog.h> 103 #include <sys/sysctl.h> 104 #include <sys/time.h> 105 106 #include <vm/uma.h> 107 108 #include <net/bpf.h> 109 #include <net/if.h> 110 #include <net/if_clone.h> 111 #include <net/if_dl.h> 112 #include <net/if_types.h> 113 #include <net/if_var.h> 114 #include <net/if_private.h> 115 #include <net/pfil.h> 116 #include <net/vnet.h> 117 118 #include <netinet/in.h> 119 #include <netinet/in_systm.h> 120 #include <netinet/in_var.h> 121 #include <netinet/ip.h> 122 #include <netinet/ip_var.h> 123 #ifdef INET6 124 #include <netinet/ip6.h> 125 #include <netinet6/ip6_var.h> 126 #include <netinet6/in6_ifattach.h> 127 #endif 128 #if defined(INET) || defined(INET6) 129 #include <netinet/ip_carp.h> 130 #endif 131 #include <machine/in_cksum.h> 132 #include <netinet/if_ether.h> 133 #include <net/bridgestp.h> 134 #include <net/if_bridgevar.h> 135 #include <net/if_llc.h> 136 #include <net/if_vlan_var.h> 137 138 #include <net/route.h> 139 140 /* 141 * At various points in the code we need to know if we're hooked into the INET 142 * and/or INET6 pfil. Define some macros to do that based on which IP versions 143 * are enabled in the kernel. This avoids littering the rest of the code with 144 * #ifnet INET6 to avoid referencing V_inet6_pfil_head. 145 */ 146 #ifdef INET6 147 #define PFIL_HOOKED_IN_INET6 PFIL_HOOKED_IN(V_inet6_pfil_head) 148 #define PFIL_HOOKED_OUT_INET6 PFIL_HOOKED_OUT(V_inet6_pfil_head) 149 #else 150 #define PFIL_HOOKED_IN_INET6 false 151 #define PFIL_HOOKED_OUT_INET6 false 152 #endif 153 154 #ifdef INET 155 #define PFIL_HOOKED_IN_INET PFIL_HOOKED_IN(V_inet_pfil_head) 156 #define PFIL_HOOKED_OUT_INET PFIL_HOOKED_OUT(V_inet_pfil_head) 157 #else 158 #define PFIL_HOOKED_IN_INET false 159 #define PFIL_HOOKED_OUT_INET false 160 #endif 161 162 #define PFIL_HOOKED_IN_46 (PFIL_HOOKED_IN_INET6 || PFIL_HOOKED_IN_INET) 163 #define PFIL_HOOKED_OUT_46 (PFIL_HOOKED_OUT_INET6 || PFIL_HOOKED_OUT_INET) 164 165 /* 166 * Size of the route hash table. Must be a power of two. 167 */ 168 #ifndef BRIDGE_RTHASH_SIZE 169 #define BRIDGE_RTHASH_SIZE 1024 170 #endif 171 172 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1) 173 174 /* 175 * Default maximum number of addresses to cache. 176 */ 177 #ifndef BRIDGE_RTABLE_MAX 178 #define BRIDGE_RTABLE_MAX 2000 179 #endif 180 181 /* 182 * Timeout (in seconds) for entries learned dynamically. 183 */ 184 #ifndef BRIDGE_RTABLE_TIMEOUT 185 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */ 186 #endif 187 188 /* 189 * Number of seconds between walks of the route list. 190 */ 191 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD 192 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60) 193 #endif 194 195 /* 196 * List of capabilities to possibly mask on the member interface. 197 */ 198 #define BRIDGE_IFCAPS_MASK (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\ 199 IFCAP_TXCSUM_IPV6|IFCAP_MEXTPG) 200 201 /* 202 * List of capabilities to strip 203 */ 204 #define BRIDGE_IFCAPS_STRIP IFCAP_LRO 205 206 /* 207 * Bridge locking 208 * 209 * The bridge relies heavily on the epoch(9) system to protect its data 210 * structures. This means we can safely use CK_LISTs while in NET_EPOCH, but we 211 * must ensure there is only one writer at a time. 212 * 213 * That is: for read accesses we only need to be in NET_EPOCH, but for write 214 * accesses we must hold: 215 * 216 * - BRIDGE_RT_LOCK, for any change to bridge_rtnodes 217 * - BRIDGE_LOCK, for any other change 218 * 219 * The BRIDGE_LOCK is a sleepable lock, because it is held across ioctl() 220 * calls to bridge member interfaces and these ioctl()s can sleep. 221 * The BRIDGE_RT_LOCK is a non-sleepable mutex, because it is sometimes 222 * required while we're in NET_EPOCH and then we're not allowed to sleep. 223 */ 224 #define BRIDGE_LOCK_INIT(_sc) do { \ 225 sx_init(&(_sc)->sc_sx, "if_bridge"); \ 226 mtx_init(&(_sc)->sc_rt_mtx, "if_bridge rt", NULL, MTX_DEF); \ 227 } while (0) 228 #define BRIDGE_LOCK_DESTROY(_sc) do { \ 229 sx_destroy(&(_sc)->sc_sx); \ 230 mtx_destroy(&(_sc)->sc_rt_mtx); \ 231 } while (0) 232 #define BRIDGE_LOCK(_sc) sx_xlock(&(_sc)->sc_sx) 233 #define BRIDGE_UNLOCK(_sc) sx_xunlock(&(_sc)->sc_sx) 234 #define BRIDGE_LOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_XLOCKED) 235 #define BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(_sc) \ 236 MPASS(in_epoch(net_epoch_preempt) || sx_xlocked(&(_sc)->sc_sx)) 237 #define BRIDGE_UNLOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_UNLOCKED) 238 #define BRIDGE_RT_LOCK(_sc) mtx_lock(&(_sc)->sc_rt_mtx) 239 #define BRIDGE_RT_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_rt_mtx) 240 #define BRIDGE_RT_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_rt_mtx, MA_OWNED) 241 #define BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(_sc) \ 242 MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(_sc)->sc_rt_mtx)) 243 244 struct bridge_softc; 245 246 /* 247 * Bridge interface list entry. 248 */ 249 struct bridge_iflist { 250 CK_LIST_ENTRY(bridge_iflist) bif_next; 251 struct ifnet *bif_ifp; /* member if */ 252 struct bridge_softc *bif_sc; /* parent bridge */ 253 struct bstp_port bif_stp; /* STP state */ 254 uint32_t bif_flags; /* member if flags */ 255 int bif_savedcaps; /* saved capabilities */ 256 uint32_t bif_addrmax; /* max # of addresses */ 257 uint32_t bif_addrcnt; /* cur. # of addresses */ 258 uint32_t bif_addrexceeded;/* # of address violations */ 259 struct epoch_context bif_epoch_ctx; 260 ether_vlanid_t bif_pvid; /* port vlan id */ 261 ifbvlan_set_t bif_vlan_set; /* if allowed tagged vlans */ 262 uint16_t bif_vlanproto; /* vlan protocol */ 263 }; 264 265 /* 266 * Bridge route node. 267 */ 268 struct bridge_rtnode { 269 CK_LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */ 270 CK_LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */ 271 struct bridge_iflist *brt_dst; /* destination if */ 272 unsigned long brt_expire; /* expiration time */ 273 uint8_t brt_flags; /* address flags */ 274 uint8_t brt_addr[ETHER_ADDR_LEN]; 275 ether_vlanid_t brt_vlan; /* vlan id */ 276 struct vnet *brt_vnet; 277 struct epoch_context brt_epoch_ctx; 278 }; 279 #define brt_ifp brt_dst->bif_ifp 280 281 /* 282 * Software state for each bridge. 283 */ 284 struct bridge_softc { 285 struct ifnet *sc_ifp; /* make this an interface */ 286 LIST_ENTRY(bridge_softc) sc_list; 287 struct sx sc_sx; 288 struct mtx sc_rt_mtx; 289 uint32_t sc_brtmax; /* max # of addresses */ 290 uint32_t sc_brtcnt; /* cur. # of addresses */ 291 uint32_t sc_brttimeout; /* rt timeout in seconds */ 292 struct callout sc_brcallout; /* bridge callout */ 293 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */ 294 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */ 295 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */ 296 uint32_t sc_rthash_key; /* key for hash */ 297 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */ 298 struct bstp_state sc_stp; /* STP state */ 299 uint32_t sc_brtexceeded; /* # of cache drops */ 300 struct ifnet *sc_ifaddr; /* member mac copied from */ 301 struct ether_addr sc_defaddr; /* Default MAC address */ 302 if_input_fn_t sc_if_input; /* Saved copy of if_input */ 303 struct epoch_context sc_epoch_ctx; 304 ifbr_flags_t sc_flags; /* bridge flags */ 305 ether_vlanid_t sc_defpvid; /* default PVID */ 306 }; 307 308 VNET_DEFINE_STATIC(struct sx, bridge_list_sx); 309 #define V_bridge_list_sx VNET(bridge_list_sx) 310 static eventhandler_tag bridge_detach_cookie; 311 312 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD; 313 314 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone); 315 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone) 316 317 static int bridge_clone_create(struct if_clone *, char *, size_t, 318 struct ifc_data *, struct ifnet **); 319 static int bridge_clone_destroy(struct if_clone *, struct ifnet *, uint32_t); 320 321 static int bridge_ioctl(struct ifnet *, u_long, caddr_t); 322 static void bridge_mutecaps(struct bridge_softc *); 323 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *, 324 int); 325 static void bridge_ifdetach(void *arg __unused, struct ifnet *); 326 static void bridge_init(void *); 327 static void bridge_dummynet(struct mbuf *, struct ifnet *); 328 static bool bridge_same(const void *, const void *); 329 static void *bridge_get_softc(struct ifnet *); 330 static void bridge_stop(struct ifnet *); 331 static int bridge_transmit(struct ifnet *, struct mbuf *); 332 #ifdef ALTQ 333 static void bridge_altq_start(if_t); 334 static int bridge_altq_transmit(if_t, struct mbuf *); 335 #endif 336 static void bridge_qflush(struct ifnet *); 337 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *); 338 static void bridge_inject(struct ifnet *, struct mbuf *); 339 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *, 340 struct rtentry *); 341 static int bridge_enqueue(struct bridge_softc *, struct ifnet *, 342 struct mbuf *, struct bridge_iflist *); 343 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int); 344 345 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *, 346 struct mbuf *m); 347 static bool bridge_member_ifaddrs(void); 348 static void bridge_timer(void *); 349 350 static void bridge_broadcast(struct bridge_softc *, struct ifnet *, 351 struct mbuf *, int); 352 static void bridge_span(struct bridge_softc *, struct mbuf *); 353 354 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *, 355 ether_vlanid_t, struct bridge_iflist *, int, uint8_t); 356 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *, 357 ether_vlanid_t); 358 static void bridge_rttrim(struct bridge_softc *); 359 static void bridge_rtage(struct bridge_softc *); 360 static void bridge_rtflush(struct bridge_softc *, int); 361 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *, 362 ether_vlanid_t); 363 static bool bridge_vfilter_in(const struct bridge_iflist *, struct mbuf *); 364 static bool bridge_vfilter_out(const struct bridge_iflist *, 365 const struct mbuf *); 366 367 static void bridge_rtable_init(struct bridge_softc *); 368 static void bridge_rtable_fini(struct bridge_softc *); 369 370 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *); 371 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *, 372 const uint8_t *, ether_vlanid_t); 373 static int bridge_rtnode_insert(struct bridge_softc *, 374 struct bridge_rtnode *); 375 static void bridge_rtnode_destroy(struct bridge_softc *, 376 struct bridge_rtnode *); 377 static void bridge_rtable_expire(struct ifnet *, int); 378 static void bridge_state_change(struct ifnet *, int); 379 380 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *, 381 const char *name); 382 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *, 383 struct ifnet *ifp); 384 static void bridge_delete_member(struct bridge_softc *, 385 struct bridge_iflist *, int); 386 static void bridge_delete_span(struct bridge_softc *, 387 struct bridge_iflist *); 388 389 static int bridge_ioctl_add(struct bridge_softc *, void *); 390 static int bridge_ioctl_del(struct bridge_softc *, void *); 391 static int bridge_ioctl_gifflags(struct bridge_softc *, void *); 392 static int bridge_ioctl_sifflags(struct bridge_softc *, void *); 393 static int bridge_ioctl_scache(struct bridge_softc *, void *); 394 static int bridge_ioctl_gcache(struct bridge_softc *, void *); 395 static int bridge_ioctl_gifs(struct bridge_softc *, void *); 396 static int bridge_ioctl_rts(struct bridge_softc *, void *); 397 static int bridge_ioctl_saddr(struct bridge_softc *, void *); 398 static int bridge_ioctl_sto(struct bridge_softc *, void *); 399 static int bridge_ioctl_gto(struct bridge_softc *, void *); 400 static int bridge_ioctl_daddr(struct bridge_softc *, void *); 401 static int bridge_ioctl_flush(struct bridge_softc *, void *); 402 static int bridge_ioctl_gpri(struct bridge_softc *, void *); 403 static int bridge_ioctl_spri(struct bridge_softc *, void *); 404 static int bridge_ioctl_ght(struct bridge_softc *, void *); 405 static int bridge_ioctl_sht(struct bridge_softc *, void *); 406 static int bridge_ioctl_gfd(struct bridge_softc *, void *); 407 static int bridge_ioctl_sfd(struct bridge_softc *, void *); 408 static int bridge_ioctl_gma(struct bridge_softc *, void *); 409 static int bridge_ioctl_sma(struct bridge_softc *, void *); 410 static int bridge_ioctl_sifprio(struct bridge_softc *, void *); 411 static int bridge_ioctl_sifcost(struct bridge_softc *, void *); 412 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *); 413 static int bridge_ioctl_sifpvid(struct bridge_softc *, void *); 414 static int bridge_ioctl_sifvlanset(struct bridge_softc *, void *); 415 static int bridge_ioctl_gifvlanset(struct bridge_softc *, void *); 416 static int bridge_ioctl_addspan(struct bridge_softc *, void *); 417 static int bridge_ioctl_delspan(struct bridge_softc *, void *); 418 static int bridge_ioctl_gbparam(struct bridge_softc *, void *); 419 static int bridge_ioctl_grte(struct bridge_softc *, void *); 420 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *); 421 static int bridge_ioctl_sproto(struct bridge_softc *, void *); 422 static int bridge_ioctl_stxhc(struct bridge_softc *, void *); 423 static int bridge_ioctl_gflags(struct bridge_softc *, void *); 424 static int bridge_ioctl_sflags(struct bridge_softc *, void *); 425 static int bridge_ioctl_gdefpvid(struct bridge_softc *, void *); 426 static int bridge_ioctl_sdefpvid(struct bridge_softc *, void *); 427 static int bridge_ioctl_svlanproto(struct bridge_softc *, void *); 428 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *, 429 int); 430 #ifdef INET 431 static int bridge_ip_checkbasic(struct mbuf **mp); 432 static int bridge_fragment(struct ifnet *, struct mbuf **mp, 433 struct ether_header *, int, struct llc *); 434 #endif /* INET */ 435 #ifdef INET6 436 static int bridge_ip6_checkbasic(struct mbuf **mp); 437 #endif /* INET6 */ 438 static void bridge_linkstate(struct ifnet *ifp); 439 static void bridge_linkcheck(struct bridge_softc *sc); 440 441 /* 442 * Use the "null" value from IEEE 802.1Q-2014 Table 9-2 443 * to indicate untagged frames. 444 */ 445 #define VLANTAGOF(_m) \ 446 ((_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : DOT1Q_VID_NULL) 447 448 static struct bstp_cb_ops bridge_ops = { 449 .bcb_state = bridge_state_change, 450 .bcb_rtage = bridge_rtable_expire 451 }; 452 453 SYSCTL_DECL(_net_link); 454 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, 455 "Bridge"); 456 457 /* only pass IP[46] packets when pfil is enabled */ 458 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1; 459 #define V_pfil_onlyip VNET(pfil_onlyip) 460 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, 461 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0, 462 "Only pass IP packets when pfil is enabled"); 463 464 /* run pfil hooks on the bridge interface */ 465 VNET_DEFINE_STATIC(int, pfil_bridge) = 0; 466 #define V_pfil_bridge VNET(pfil_bridge) 467 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, 468 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0, 469 "Packet filter on the bridge interface"); 470 471 /* layer2 filter with ipfw */ 472 VNET_DEFINE_STATIC(int, pfil_ipfw); 473 #define V_pfil_ipfw VNET(pfil_ipfw) 474 475 /* layer2 ARP filter with ipfw */ 476 VNET_DEFINE_STATIC(int, pfil_ipfw_arp); 477 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp) 478 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, 479 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0, 480 "Filter ARP packets through IPFW layer2"); 481 482 /* run pfil hooks on the member interface */ 483 VNET_DEFINE_STATIC(int, pfil_member) = 0; 484 #define V_pfil_member VNET(pfil_member) 485 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, 486 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0, 487 "Packet filter on the member interface"); 488 489 /* run pfil hooks on the physical interface for locally destined packets */ 490 VNET_DEFINE_STATIC(int, pfil_local_phys); 491 #define V_pfil_local_phys VNET(pfil_local_phys) 492 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, 493 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0, 494 "Packet filter on the physical interface for locally destined packets"); 495 496 /* log STP state changes */ 497 VNET_DEFINE_STATIC(int, log_stp); 498 #define V_log_stp VNET(log_stp) 499 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, 500 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0, 501 "Log STP state changes"); 502 503 /* share MAC with first bridge member */ 504 VNET_DEFINE_STATIC(int, bridge_inherit_mac); 505 #define V_bridge_inherit_mac VNET(bridge_inherit_mac) 506 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, 507 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0, 508 "Inherit MAC address from the first bridge member"); 509 510 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0; 511 #define V_allow_llz_overlap VNET(allow_llz_overlap) 512 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap, 513 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0, 514 "Allow overlap of link-local scope " 515 "zones of a bridge interface and the member interfaces"); 516 517 /* log MAC address port flapping */ 518 VNET_DEFINE_STATIC(bool, log_mac_flap) = true; 519 #define V_log_mac_flap VNET(log_mac_flap) 520 SYSCTL_BOOL(_net_link_bridge, OID_AUTO, log_mac_flap, 521 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(log_mac_flap), true, 522 "Log MAC address port flapping"); 523 524 /* allow IP addresses on bridge members */ 525 VNET_DEFINE_STATIC(bool, member_ifaddrs) = true; 526 #define V_member_ifaddrs VNET(member_ifaddrs) 527 SYSCTL_BOOL(_net_link_bridge, OID_AUTO, member_ifaddrs, 528 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(member_ifaddrs), false, 529 "Allow layer 3 addresses on bridge members (deprecated)"); 530 531 static bool 532 bridge_member_ifaddrs(void) 533 { 534 return (V_member_ifaddrs); 535 } 536 537 VNET_DEFINE_STATIC(int, log_interval) = 5; 538 VNET_DEFINE_STATIC(int, log_count) = 0; 539 VNET_DEFINE_STATIC(struct timeval, log_last) = { 0 }; 540 541 #define V_log_interval VNET(log_interval) 542 #define V_log_count VNET(log_count) 543 #define V_log_last VNET(log_last) 544 545 struct bridge_control { 546 int (*bc_func)(struct bridge_softc *, void *); 547 int bc_argsize; 548 int bc_flags; 549 }; 550 551 #define BC_F_COPYIN 0x01 /* copy arguments in */ 552 #define BC_F_COPYOUT 0x02 /* copy arguments out */ 553 #define BC_F_SUSER 0x04 /* do super-user check */ 554 555 static const struct bridge_control bridge_control_table[] = { 556 { bridge_ioctl_add, sizeof(struct ifbreq), 557 BC_F_COPYIN|BC_F_SUSER }, 558 { bridge_ioctl_del, sizeof(struct ifbreq), 559 BC_F_COPYIN|BC_F_SUSER }, 560 561 { bridge_ioctl_gifflags, sizeof(struct ifbreq), 562 BC_F_COPYIN|BC_F_COPYOUT }, 563 { bridge_ioctl_sifflags, sizeof(struct ifbreq), 564 BC_F_COPYIN|BC_F_SUSER }, 565 566 { bridge_ioctl_scache, sizeof(struct ifbrparam), 567 BC_F_COPYIN|BC_F_SUSER }, 568 { bridge_ioctl_gcache, sizeof(struct ifbrparam), 569 BC_F_COPYOUT }, 570 571 { bridge_ioctl_gifs, sizeof(struct ifbifconf), 572 BC_F_COPYIN|BC_F_COPYOUT }, 573 { bridge_ioctl_rts, sizeof(struct ifbaconf), 574 BC_F_COPYIN|BC_F_COPYOUT }, 575 576 { bridge_ioctl_saddr, sizeof(struct ifbareq), 577 BC_F_COPYIN|BC_F_SUSER }, 578 579 { bridge_ioctl_sto, sizeof(struct ifbrparam), 580 BC_F_COPYIN|BC_F_SUSER }, 581 { bridge_ioctl_gto, sizeof(struct ifbrparam), 582 BC_F_COPYOUT }, 583 584 { bridge_ioctl_daddr, sizeof(struct ifbareq), 585 BC_F_COPYIN|BC_F_SUSER }, 586 587 { bridge_ioctl_flush, sizeof(struct ifbreq), 588 BC_F_COPYIN|BC_F_SUSER }, 589 590 { bridge_ioctl_gpri, sizeof(struct ifbrparam), 591 BC_F_COPYOUT }, 592 { bridge_ioctl_spri, sizeof(struct ifbrparam), 593 BC_F_COPYIN|BC_F_SUSER }, 594 595 { bridge_ioctl_ght, sizeof(struct ifbrparam), 596 BC_F_COPYOUT }, 597 { bridge_ioctl_sht, sizeof(struct ifbrparam), 598 BC_F_COPYIN|BC_F_SUSER }, 599 600 { bridge_ioctl_gfd, sizeof(struct ifbrparam), 601 BC_F_COPYOUT }, 602 { bridge_ioctl_sfd, sizeof(struct ifbrparam), 603 BC_F_COPYIN|BC_F_SUSER }, 604 605 { bridge_ioctl_gma, sizeof(struct ifbrparam), 606 BC_F_COPYOUT }, 607 { bridge_ioctl_sma, sizeof(struct ifbrparam), 608 BC_F_COPYIN|BC_F_SUSER }, 609 610 { bridge_ioctl_sifprio, sizeof(struct ifbreq), 611 BC_F_COPYIN|BC_F_SUSER }, 612 613 { bridge_ioctl_sifcost, sizeof(struct ifbreq), 614 BC_F_COPYIN|BC_F_SUSER }, 615 616 { bridge_ioctl_addspan, sizeof(struct ifbreq), 617 BC_F_COPYIN|BC_F_SUSER }, 618 { bridge_ioctl_delspan, sizeof(struct ifbreq), 619 BC_F_COPYIN|BC_F_SUSER }, 620 621 { bridge_ioctl_gbparam, sizeof(struct ifbropreq), 622 BC_F_COPYOUT }, 623 624 { bridge_ioctl_grte, sizeof(struct ifbrparam), 625 BC_F_COPYOUT }, 626 627 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf), 628 BC_F_COPYIN|BC_F_COPYOUT }, 629 630 { bridge_ioctl_sproto, sizeof(struct ifbrparam), 631 BC_F_COPYIN|BC_F_SUSER }, 632 633 { bridge_ioctl_stxhc, sizeof(struct ifbrparam), 634 BC_F_COPYIN|BC_F_SUSER }, 635 636 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq), 637 BC_F_COPYIN|BC_F_SUSER }, 638 639 { bridge_ioctl_sifpvid, sizeof(struct ifbreq), 640 BC_F_COPYIN|BC_F_SUSER }, 641 642 { bridge_ioctl_sifvlanset, sizeof(struct ifbif_vlan_req), 643 BC_F_COPYIN|BC_F_SUSER }, 644 645 { bridge_ioctl_gifvlanset, sizeof(struct ifbif_vlan_req), 646 BC_F_COPYIN|BC_F_COPYOUT }, 647 648 { bridge_ioctl_gflags, sizeof(struct ifbrparam), 649 BC_F_COPYOUT }, 650 651 { bridge_ioctl_sflags, sizeof(struct ifbrparam), 652 BC_F_COPYIN|BC_F_SUSER }, 653 654 { bridge_ioctl_gdefpvid, sizeof(struct ifbrparam), 655 BC_F_COPYOUT }, 656 657 { bridge_ioctl_sdefpvid, sizeof(struct ifbrparam), 658 BC_F_COPYIN|BC_F_SUSER }, 659 660 { bridge_ioctl_svlanproto, sizeof(struct ifbreq), 661 BC_F_COPYIN|BC_F_SUSER }, 662 }; 663 static const int bridge_control_table_size = nitems(bridge_control_table); 664 665 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list) = 666 LIST_HEAD_INITIALIZER(); 667 #define V_bridge_list VNET(bridge_list) 668 #define BRIDGE_LIST_LOCK_INIT(x) sx_init(&V_bridge_list_sx, \ 669 "if_bridge list") 670 #define BRIDGE_LIST_LOCK_DESTROY(x) sx_destroy(&V_bridge_list_sx) 671 #define BRIDGE_LIST_LOCK(x) sx_xlock(&V_bridge_list_sx) 672 #define BRIDGE_LIST_UNLOCK(x) sx_xunlock(&V_bridge_list_sx) 673 674 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner); 675 #define V_bridge_cloner VNET(bridge_cloner) 676 677 static const char bridge_name[] = "bridge"; 678 679 static void 680 vnet_bridge_init(const void *unused __unused) 681 { 682 683 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode", 684 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL, 685 UMA_ALIGN_PTR, 0); 686 BRIDGE_LIST_LOCK_INIT(); 687 688 struct if_clone_addreq req = { 689 .create_f = bridge_clone_create, 690 .destroy_f = bridge_clone_destroy, 691 .flags = IFC_F_AUTOUNIT, 692 }; 693 V_bridge_cloner = ifc_attach_cloner(bridge_name, &req); 694 } 695 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY, 696 vnet_bridge_init, NULL); 697 698 static void 699 vnet_bridge_uninit(const void *unused __unused) 700 { 701 702 ifc_detach_cloner(V_bridge_cloner); 703 V_bridge_cloner = NULL; 704 BRIDGE_LIST_LOCK_DESTROY(); 705 706 /* Callbacks may use the UMA zone. */ 707 NET_EPOCH_DRAIN_CALLBACKS(); 708 709 uma_zdestroy(V_bridge_rtnode_zone); 710 } 711 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY, 712 vnet_bridge_uninit, NULL); 713 714 static int 715 bridge_modevent(module_t mod, int type, void *data) 716 { 717 718 switch (type) { 719 case MOD_LOAD: 720 bridge_dn_p = bridge_dummynet; 721 bridge_same_p = bridge_same; 722 bridge_get_softc_p = bridge_get_softc; 723 bridge_member_ifaddrs_p = bridge_member_ifaddrs; 724 bridge_detach_cookie = EVENTHANDLER_REGISTER( 725 ifnet_departure_event, bridge_ifdetach, NULL, 726 EVENTHANDLER_PRI_ANY); 727 break; 728 case MOD_UNLOAD: 729 EVENTHANDLER_DEREGISTER(ifnet_departure_event, 730 bridge_detach_cookie); 731 bridge_dn_p = NULL; 732 bridge_same_p = NULL; 733 bridge_get_softc_p = NULL; 734 bridge_member_ifaddrs_p = NULL; 735 break; 736 default: 737 return (EOPNOTSUPP); 738 } 739 return (0); 740 } 741 742 static moduledata_t bridge_mod = { 743 "if_bridge", 744 bridge_modevent, 745 0 746 }; 747 748 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY); 749 MODULE_VERSION(if_bridge, 1); 750 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1); 751 752 /* 753 * handler for net.link.bridge.ipfw 754 */ 755 static int 756 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS) 757 { 758 int enable = V_pfil_ipfw; 759 int error; 760 761 error = sysctl_handle_int(oidp, &enable, 0, req); 762 enable &= 1; 763 764 if (enable != V_pfil_ipfw) { 765 V_pfil_ipfw = enable; 766 767 /* 768 * Disable pfil so that ipfw doesnt run twice, if the user 769 * really wants both then they can re-enable pfil_bridge and/or 770 * pfil_member. Also allow non-ip packets as ipfw can filter by 771 * layer2 type. 772 */ 773 if (V_pfil_ipfw) { 774 V_pfil_onlyip = 0; 775 V_pfil_bridge = 0; 776 V_pfil_member = 0; 777 } 778 } 779 780 return (error); 781 } 782 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, 783 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT, 784 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I", 785 "Layer2 filter with IPFW"); 786 787 #ifdef VIMAGE 788 static void 789 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg) 790 { 791 struct bridge_softc *sc = ifp->if_softc; 792 struct bridge_iflist *bif; 793 794 BRIDGE_LOCK(sc); 795 796 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL) 797 bridge_delete_member(sc, bif, 0); 798 799 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) { 800 bridge_delete_span(sc, bif); 801 } 802 803 BRIDGE_UNLOCK(sc); 804 } 805 #endif 806 807 /* 808 * bridge_get_softc: 809 * 810 * Return the bridge softc for an ifnet. 811 */ 812 static void * 813 bridge_get_softc(struct ifnet *ifp) 814 { 815 struct bridge_iflist *bif; 816 817 NET_EPOCH_ASSERT(); 818 819 bif = ifp->if_bridge; 820 if (bif == NULL) 821 return (NULL); 822 return (bif->bif_sc); 823 } 824 825 /* 826 * bridge_same: 827 * 828 * Return true if two interfaces are in the same bridge. This is only used by 829 * bridgestp via bridge_same_p. 830 */ 831 static bool 832 bridge_same(const void *bifap, const void *bifbp) 833 { 834 const struct bridge_iflist *bifa = bifap, *bifb = bifbp; 835 836 NET_EPOCH_ASSERT(); 837 838 if (bifa == NULL || bifb == NULL) 839 return (false); 840 841 return (bifa->bif_sc == bifb->bif_sc); 842 } 843 844 /* 845 * bridge_clone_create: 846 * 847 * Create a new bridge instance. 848 */ 849 static int 850 bridge_clone_create(struct if_clone *ifc, char *name, size_t len, 851 struct ifc_data *ifd, struct ifnet **ifpp) 852 { 853 struct bridge_softc *sc; 854 struct ifnet *ifp; 855 856 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO); 857 ifp = sc->sc_ifp = if_alloc(IFT_ETHER); 858 859 BRIDGE_LOCK_INIT(sc); 860 sc->sc_brtmax = BRIDGE_RTABLE_MAX; 861 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT; 862 863 /* Initialize our routing table. */ 864 bridge_rtable_init(sc); 865 866 callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0); 867 868 CK_LIST_INIT(&sc->sc_iflist); 869 CK_LIST_INIT(&sc->sc_spanlist); 870 871 ifp->if_softc = sc; 872 if_initname(ifp, bridge_name, ifd->unit); 873 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; 874 ifp->if_capabilities = ifp->if_capenable = IFCAP_VLAN_HWTAGGING; 875 ifp->if_ioctl = bridge_ioctl; 876 #ifdef ALTQ 877 ifp->if_start = bridge_altq_start; 878 ifp->if_transmit = bridge_altq_transmit; 879 IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen); 880 ifp->if_snd.ifq_drv_maxlen = 0; 881 IFQ_SET_READY(&ifp->if_snd); 882 #else 883 ifp->if_transmit = bridge_transmit; 884 #endif 885 ifp->if_qflush = bridge_qflush; 886 ifp->if_init = bridge_init; 887 ifp->if_type = IFT_BRIDGE; 888 889 ether_gen_addr(ifp, &sc->sc_defaddr); 890 891 bstp_attach(&sc->sc_stp, &bridge_ops); 892 ether_ifattach(ifp, sc->sc_defaddr.octet); 893 /* Now undo some of the damage... */ 894 ifp->if_baudrate = 0; 895 #ifdef VIMAGE 896 ifp->if_reassign = bridge_reassign; 897 #endif 898 sc->sc_if_input = ifp->if_input; /* ether_input */ 899 ifp->if_input = bridge_inject; 900 901 /* 902 * Allow BRIDGE_INPUT() to pass in packets originating from the bridge 903 * itself via bridge_inject(). This is required for netmap but 904 * otherwise has no effect. 905 */ 906 ifp->if_bridge_input = bridge_input; 907 908 BRIDGE_LIST_LOCK(); 909 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list); 910 BRIDGE_LIST_UNLOCK(); 911 *ifpp = ifp; 912 913 return (0); 914 } 915 916 static void 917 bridge_clone_destroy_cb(struct epoch_context *ctx) 918 { 919 struct bridge_softc *sc; 920 921 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx); 922 923 BRIDGE_LOCK_DESTROY(sc); 924 free(sc, M_DEVBUF); 925 } 926 927 /* 928 * bridge_clone_destroy: 929 * 930 * Destroy a bridge instance. 931 */ 932 static int 933 bridge_clone_destroy(struct if_clone *ifc, struct ifnet *ifp, uint32_t flags) 934 { 935 struct bridge_softc *sc = ifp->if_softc; 936 struct bridge_iflist *bif; 937 938 BRIDGE_LOCK(sc); 939 940 bridge_stop(ifp); 941 ifp->if_flags &= ~IFF_UP; 942 943 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL) 944 bridge_delete_member(sc, bif, 0); 945 946 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) { 947 bridge_delete_span(sc, bif); 948 } 949 950 /* Tear down the routing table. */ 951 bridge_rtable_fini(sc); 952 953 BRIDGE_UNLOCK(sc); 954 955 callout_drain(&sc->sc_brcallout); 956 957 BRIDGE_LIST_LOCK(); 958 LIST_REMOVE(sc, sc_list); 959 BRIDGE_LIST_UNLOCK(); 960 961 bstp_detach(&sc->sc_stp); 962 #ifdef ALTQ 963 IFQ_PURGE(&ifp->if_snd); 964 #endif 965 ether_ifdetach(ifp); 966 if_free(ifp); 967 968 NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx); 969 970 return (0); 971 } 972 973 /* 974 * bridge_ioctl: 975 * 976 * Handle a control request from the operator. 977 */ 978 static int 979 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data) 980 { 981 struct bridge_softc *sc = ifp->if_softc; 982 struct ifreq *ifr = (struct ifreq *)data; 983 struct bridge_iflist *bif; 984 struct thread *td = curthread; 985 union { 986 struct ifbreq ifbreq; 987 struct ifbifconf ifbifconf; 988 struct ifbareq ifbareq; 989 struct ifbaconf ifbaconf; 990 struct ifbrparam ifbrparam; 991 struct ifbropreq ifbropreq; 992 struct ifbif_vlan_req ifvlanreq; 993 } args; 994 struct ifdrv *ifd = (struct ifdrv *) data; 995 const struct bridge_control *bc; 996 int error = 0, oldmtu; 997 998 BRIDGE_LOCK(sc); 999 1000 switch (cmd) { 1001 case SIOCADDMULTI: 1002 case SIOCDELMULTI: 1003 break; 1004 1005 case SIOCGDRVSPEC: 1006 case SIOCSDRVSPEC: 1007 if (ifd->ifd_cmd >= bridge_control_table_size) { 1008 error = EXTERROR(EINVAL, "Invalid control command"); 1009 break; 1010 } 1011 bc = &bridge_control_table[ifd->ifd_cmd]; 1012 1013 if (cmd == SIOCGDRVSPEC && 1014 (bc->bc_flags & BC_F_COPYOUT) == 0) { 1015 error = EXTERROR(EINVAL, 1016 "Inappropriate ioctl for command " 1017 "(expected SIOCSDRVSPEC)"); 1018 break; 1019 } 1020 else if (cmd == SIOCSDRVSPEC && 1021 (bc->bc_flags & BC_F_COPYOUT) != 0) { 1022 error = EXTERROR(EINVAL, 1023 "Inappropriate ioctl for command " 1024 "(expected SIOCGDRVSPEC)"); 1025 break; 1026 } 1027 1028 if (bc->bc_flags & BC_F_SUSER) { 1029 error = priv_check(td, PRIV_NET_BRIDGE); 1030 if (error) { 1031 EXTERROR(error, "PRIV_NET_BRIDGE required"); 1032 break; 1033 } 1034 } 1035 1036 if (ifd->ifd_len != bc->bc_argsize || 1037 ifd->ifd_len > sizeof(args)) { 1038 error = EXTERROR(EINVAL, "Invalid argument size"); 1039 break; 1040 } 1041 1042 bzero(&args, sizeof(args)); 1043 if (bc->bc_flags & BC_F_COPYIN) { 1044 error = copyin(ifd->ifd_data, &args, ifd->ifd_len); 1045 if (error) 1046 break; 1047 } 1048 1049 oldmtu = ifp->if_mtu; 1050 error = (*bc->bc_func)(sc, &args); 1051 if (error) 1052 break; 1053 1054 /* 1055 * Bridge MTU may change during addition of the first port. 1056 * If it did, do network layer specific procedure. 1057 */ 1058 if (ifp->if_mtu != oldmtu) 1059 if_notifymtu(ifp); 1060 1061 if (bc->bc_flags & BC_F_COPYOUT) 1062 error = copyout(&args, ifd->ifd_data, ifd->ifd_len); 1063 1064 break; 1065 1066 case SIOCSIFFLAGS: 1067 if (!(ifp->if_flags & IFF_UP) && 1068 (ifp->if_drv_flags & IFF_DRV_RUNNING)) { 1069 /* 1070 * If interface is marked down and it is running, 1071 * then stop it. 1072 */ 1073 bridge_stop(ifp); 1074 } else if ((ifp->if_flags & IFF_UP) && 1075 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) { 1076 /* 1077 * If interface is marked up and it is stopped, then 1078 * start it. 1079 */ 1080 BRIDGE_UNLOCK(sc); 1081 (*ifp->if_init)(sc); 1082 BRIDGE_LOCK(sc); 1083 } 1084 break; 1085 1086 case SIOCSIFMTU: 1087 oldmtu = sc->sc_ifp->if_mtu; 1088 1089 if (ifr->ifr_mtu < IF_MINMTU) { 1090 error = EXTERROR(EINVAL, 1091 "Requested MTU is lower than IF_MINMTU"); 1092 break; 1093 } 1094 if (CK_LIST_EMPTY(&sc->sc_iflist)) { 1095 sc->sc_ifp->if_mtu = ifr->ifr_mtu; 1096 break; 1097 } 1098 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1099 error = (*bif->bif_ifp->if_ioctl)(bif->bif_ifp, 1100 SIOCSIFMTU, (caddr_t)ifr); 1101 if (error != 0) { 1102 log(LOG_NOTICE, "%s: invalid MTU: %u for" 1103 " member %s\n", sc->sc_ifp->if_xname, 1104 ifr->ifr_mtu, 1105 bif->bif_ifp->if_xname); 1106 error = EINVAL; 1107 break; 1108 } 1109 } 1110 if (error) { 1111 /* Restore the previous MTU on all member interfaces. */ 1112 ifr->ifr_mtu = oldmtu; 1113 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1114 (*bif->bif_ifp->if_ioctl)(bif->bif_ifp, 1115 SIOCSIFMTU, (caddr_t)ifr); 1116 } 1117 EXTERROR(error, 1118 "Failed to set MTU on member interface"); 1119 } else { 1120 sc->sc_ifp->if_mtu = ifr->ifr_mtu; 1121 } 1122 break; 1123 default: 1124 /* 1125 * drop the lock as ether_ioctl() will call bridge_start() and 1126 * cause the lock to be recursed. 1127 */ 1128 BRIDGE_UNLOCK(sc); 1129 error = ether_ioctl(ifp, cmd, data); 1130 BRIDGE_LOCK(sc); 1131 break; 1132 } 1133 1134 BRIDGE_UNLOCK(sc); 1135 1136 return (error); 1137 } 1138 1139 /* 1140 * bridge_mutecaps: 1141 * 1142 * Clear or restore unwanted capabilities on the member interface 1143 */ 1144 static void 1145 bridge_mutecaps(struct bridge_softc *sc) 1146 { 1147 struct bridge_iflist *bif; 1148 int enabled, mask; 1149 1150 BRIDGE_LOCK_ASSERT(sc); 1151 1152 /* Initial bitmask of capabilities to test */ 1153 mask = BRIDGE_IFCAPS_MASK; 1154 1155 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1156 /* Every member must support it or it's disabled */ 1157 mask &= bif->bif_savedcaps; 1158 } 1159 1160 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1161 enabled = bif->bif_ifp->if_capenable; 1162 enabled &= ~BRIDGE_IFCAPS_STRIP; 1163 /* Strip off mask bits and enable them again if allowed */ 1164 enabled &= ~BRIDGE_IFCAPS_MASK; 1165 enabled |= mask; 1166 bridge_set_ifcap(sc, bif, enabled); 1167 } 1168 } 1169 1170 static void 1171 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set) 1172 { 1173 struct ifnet *ifp = bif->bif_ifp; 1174 struct ifreq ifr; 1175 int error, mask, stuck; 1176 1177 bzero(&ifr, sizeof(ifr)); 1178 ifr.ifr_reqcap = set; 1179 1180 if (ifp->if_capenable != set) { 1181 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr); 1182 if (error) 1183 if_printf(sc->sc_ifp, 1184 "error setting capabilities on %s: %d\n", 1185 ifp->if_xname, error); 1186 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP; 1187 stuck = ifp->if_capenable & mask & ~set; 1188 if (stuck != 0) 1189 if_printf(sc->sc_ifp, 1190 "can't disable some capabilities on %s: 0x%x\n", 1191 ifp->if_xname, stuck); 1192 } 1193 } 1194 1195 /* 1196 * bridge_lookup_member: 1197 * 1198 * Lookup a bridge member interface. 1199 */ 1200 static struct bridge_iflist * 1201 bridge_lookup_member(struct bridge_softc *sc, const char *name) 1202 { 1203 struct bridge_iflist *bif; 1204 struct ifnet *ifp; 1205 1206 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc); 1207 1208 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1209 ifp = bif->bif_ifp; 1210 if (strcmp(ifp->if_xname, name) == 0) 1211 return (bif); 1212 } 1213 1214 return (NULL); 1215 } 1216 1217 /* 1218 * bridge_lookup_member_if: 1219 * 1220 * Lookup a bridge member interface by ifnet*. 1221 */ 1222 static struct bridge_iflist * 1223 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp) 1224 { 1225 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc); 1226 return (member_ifp->if_bridge); 1227 } 1228 1229 static void 1230 bridge_delete_member_cb(struct epoch_context *ctx) 1231 { 1232 struct bridge_iflist *bif; 1233 1234 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx); 1235 1236 free(bif, M_DEVBUF); 1237 } 1238 1239 /* 1240 * bridge_delete_member: 1241 * 1242 * Delete the specified member interface. 1243 */ 1244 static void 1245 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif, 1246 int gone) 1247 { 1248 struct ifnet *ifs = bif->bif_ifp; 1249 struct ifnet *fif = NULL; 1250 struct bridge_iflist *bifl; 1251 1252 BRIDGE_LOCK_ASSERT(sc); 1253 1254 if (bif->bif_flags & IFBIF_STP) 1255 bstp_disable(&bif->bif_stp); 1256 1257 ifs->if_bridge = NULL; 1258 CK_LIST_REMOVE(bif, bif_next); 1259 1260 /* 1261 * If removing the interface that gave the bridge its mac address, set 1262 * the mac address of the bridge to the address of the next member, or 1263 * to its default address if no members are left. 1264 */ 1265 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) { 1266 if (CK_LIST_EMPTY(&sc->sc_iflist)) { 1267 bcopy(&sc->sc_defaddr, 1268 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1269 sc->sc_ifaddr = NULL; 1270 } else { 1271 bifl = CK_LIST_FIRST(&sc->sc_iflist); 1272 fif = bifl->bif_ifp; 1273 bcopy(IF_LLADDR(fif), 1274 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1275 sc->sc_ifaddr = fif; 1276 } 1277 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp); 1278 } 1279 1280 bridge_linkcheck(sc); 1281 bridge_mutecaps(sc); /* recalcuate now this interface is removed */ 1282 BRIDGE_RT_LOCK(sc); 1283 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL); 1284 BRIDGE_RT_UNLOCK(sc); 1285 KASSERT(bif->bif_addrcnt == 0, 1286 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt)); 1287 1288 ifs->if_bridge_output = NULL; 1289 ifs->if_bridge_input = NULL; 1290 ifs->if_bridge_linkstate = NULL; 1291 if (!gone) { 1292 switch (ifs->if_type) { 1293 case IFT_ETHER: 1294 case IFT_L2VLAN: 1295 /* 1296 * Take the interface out of promiscuous mode, but only 1297 * if it was promiscuous in the first place. It might 1298 * not be if we're in the bridge_ioctl_add() error path. 1299 */ 1300 if (ifs->if_flags & IFF_PROMISC) 1301 (void) ifpromisc(ifs, 0); 1302 break; 1303 1304 case IFT_GIF: 1305 break; 1306 1307 default: 1308 #ifdef DIAGNOSTIC 1309 panic("bridge_delete_member: impossible"); 1310 #endif 1311 break; 1312 } 1313 /* Re-enable any interface capabilities */ 1314 bridge_set_ifcap(sc, bif, bif->bif_savedcaps); 1315 } 1316 bstp_destroy(&bif->bif_stp); /* prepare to free */ 1317 1318 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx); 1319 } 1320 1321 /* 1322 * bridge_delete_span: 1323 * 1324 * Delete the specified span interface. 1325 */ 1326 static void 1327 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif) 1328 { 1329 BRIDGE_LOCK_ASSERT(sc); 1330 1331 KASSERT(bif->bif_ifp->if_bridge == NULL, 1332 ("%s: not a span interface", __func__)); 1333 1334 CK_LIST_REMOVE(bif, bif_next); 1335 1336 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx); 1337 } 1338 1339 static int 1340 bridge_ioctl_add(struct bridge_softc *sc, void *arg) 1341 { 1342 struct ifbreq *req = arg; 1343 struct bridge_iflist *bif = NULL; 1344 struct ifnet *ifs; 1345 int error = 0; 1346 1347 ifs = ifunit(req->ifbr_ifsname); 1348 if (ifs == NULL) 1349 return (EXTERROR(ENOENT, "No such interface", 1350 req->ifbr_ifsname)); 1351 if (ifs->if_ioctl == NULL) /* must be supported */ 1352 return (EXTERROR(EINVAL, "Interface must support ioctl(2)")); 1353 1354 /* 1355 * If the new interface is a vlan(4), it could be a bridge SVI. 1356 * Don't allow such things to be added to bridges. 1357 */ 1358 if (ifs->if_type == IFT_L2VLAN) { 1359 struct ifnet *parent; 1360 struct epoch_tracker et; 1361 bool is_bridge; 1362 1363 /* 1364 * Entering NET_EPOCH with BRIDGE_LOCK held, but this is okay 1365 * since we don't sleep here. 1366 */ 1367 NET_EPOCH_ENTER(et); 1368 parent = VLAN_TRUNKDEV(ifs); 1369 is_bridge = (parent != NULL && parent->if_type == IFT_BRIDGE); 1370 NET_EPOCH_EXIT(et); 1371 1372 if (is_bridge) 1373 return (EXTERROR(EINVAL, 1374 "Bridge SVI cannot be added to a bridge")); 1375 } 1376 1377 /* If it's in the span list, it can't be a member. */ 1378 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1379 if (ifs == bif->bif_ifp) 1380 return (EXTERROR(EBUSY, 1381 "Span interface cannot be a member")); 1382 1383 if (ifs->if_bridge) { 1384 struct bridge_iflist *sbif = ifs->if_bridge; 1385 if (sbif->bif_sc == sc) 1386 return (EXTERROR(EEXIST, 1387 "Interface is already a member of this bridge")); 1388 1389 return (EXTERROR(EBUSY, 1390 "Interface is already a member of another bridge")); 1391 } 1392 1393 switch (ifs->if_type) { 1394 case IFT_ETHER: 1395 case IFT_L2VLAN: 1396 case IFT_GIF: 1397 /* permitted interface types */ 1398 break; 1399 default: 1400 return (EXTERROR(EINVAL, "Unsupported interface type")); 1401 } 1402 1403 #ifdef INET6 1404 /* 1405 * Two valid inet6 addresses with link-local scope must not be 1406 * on the parent interface and the member interfaces at the 1407 * same time. This restriction is needed to prevent violation 1408 * of link-local scope zone. Attempts to add a member 1409 * interface which has inet6 addresses when the parent has 1410 * inet6 triggers removal of all inet6 addresses on the member 1411 * interface. 1412 */ 1413 1414 /* Check if the parent interface has a link-local scope addr. */ 1415 if (V_allow_llz_overlap == 0 && 1416 in6ifa_llaonifp(sc->sc_ifp) != NULL) { 1417 /* 1418 * If any, remove all inet6 addresses from the member 1419 * interfaces. 1420 */ 1421 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1422 if (in6ifa_llaonifp(bif->bif_ifp)) { 1423 in6_ifdetach(bif->bif_ifp); 1424 if_printf(sc->sc_ifp, 1425 "IPv6 addresses on %s have been removed " 1426 "before adding it as a member to prevent " 1427 "IPv6 address scope violation.\n", 1428 bif->bif_ifp->if_xname); 1429 } 1430 } 1431 if (in6ifa_llaonifp(ifs)) { 1432 in6_ifdetach(ifs); 1433 if_printf(sc->sc_ifp, 1434 "IPv6 addresses on %s have been removed " 1435 "before adding it as a member to prevent " 1436 "IPv6 address scope violation.\n", 1437 ifs->if_xname); 1438 } 1439 } 1440 #endif 1441 1442 /* 1443 * If member_ifaddrs is disabled, do not allow an interface with 1444 * assigned IP addresses to be added to a bridge. Skip this check 1445 * for gif interfaces, because the IP address assigned to a gif 1446 * interface is separate from the bridge's Ethernet segment. 1447 */ 1448 if (ifs->if_type != IFT_GIF) { 1449 struct ifaddr *ifa; 1450 1451 CK_STAILQ_FOREACH(ifa, &ifs->if_addrhead, ifa_link) { 1452 if (ifa->ifa_addr->sa_family != AF_INET && 1453 ifa->ifa_addr->sa_family != AF_INET6) 1454 continue; 1455 1456 if (V_member_ifaddrs) { 1457 if_printf(sc->sc_ifp, 1458 "WARNING: Adding member interface %s which " 1459 "has an IP address assigned is deprecated " 1460 "and will be unsupported in a future " 1461 "release.\n", ifs->if_xname); 1462 break; 1463 } else { 1464 return (EXTERROR(EINVAL, 1465 "Member interface may not have " 1466 "an IP address assigned")); 1467 } 1468 } 1469 } 1470 1471 /* Allow the first Ethernet member to define the MTU */ 1472 if (CK_LIST_EMPTY(&sc->sc_iflist)) 1473 sc->sc_ifp->if_mtu = ifs->if_mtu; 1474 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) { 1475 struct ifreq ifr; 1476 1477 snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s", 1478 ifs->if_xname); 1479 ifr.ifr_mtu = sc->sc_ifp->if_mtu; 1480 1481 error = (*ifs->if_ioctl)(ifs, 1482 SIOCSIFMTU, (caddr_t)&ifr); 1483 if (error != 0) { 1484 log(LOG_NOTICE, "%s: invalid MTU: %u for" 1485 " new member %s\n", sc->sc_ifp->if_xname, 1486 ifr.ifr_mtu, 1487 ifs->if_xname); 1488 return (EXTERROR(EINVAL, 1489 "Failed to set MTU on new member")); 1490 } 1491 } 1492 1493 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO); 1494 if (bif == NULL) 1495 return (ENOMEM); 1496 1497 bif->bif_sc = sc; 1498 bif->bif_ifp = ifs; 1499 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER; 1500 bif->bif_savedcaps = ifs->if_capenable; 1501 bif->bif_vlanproto = ETHERTYPE_VLAN; 1502 bif->bif_pvid = sc->sc_defpvid; 1503 if (sc->sc_flags & IFBRF_DEFQINQ) 1504 bif->bif_flags |= IFBIF_QINQ; 1505 1506 /* 1507 * Assign the interface's MAC address to the bridge if it's the first 1508 * member and the MAC address of the bridge has not been changed from 1509 * the default randomly generated one. 1510 */ 1511 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) && 1512 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) { 1513 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); 1514 sc->sc_ifaddr = ifs; 1515 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp); 1516 } 1517 1518 ifs->if_bridge = bif; 1519 ifs->if_bridge_output = bridge_output; 1520 ifs->if_bridge_input = bridge_input; 1521 ifs->if_bridge_linkstate = bridge_linkstate; 1522 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp); 1523 /* 1524 * XXX: XLOCK HERE!?! 1525 * 1526 * NOTE: insert_***HEAD*** should be safe for the traversals. 1527 */ 1528 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next); 1529 1530 /* Set interface capabilities to the intersection set of all members */ 1531 bridge_mutecaps(sc); 1532 bridge_linkcheck(sc); 1533 1534 /* Place the interface into promiscuous mode */ 1535 switch (ifs->if_type) { 1536 case IFT_ETHER: 1537 case IFT_L2VLAN: 1538 error = ifpromisc(ifs, 1); 1539 break; 1540 } 1541 1542 if (error) 1543 bridge_delete_member(sc, bif, 0); 1544 return (error); 1545 } 1546 1547 static int 1548 bridge_ioctl_del(struct bridge_softc *sc, void *arg) 1549 { 1550 struct ifbreq *req = arg; 1551 struct bridge_iflist *bif; 1552 1553 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1554 if (bif == NULL) 1555 return (EXTERROR(ENOENT, "Interface is not a bridge member")); 1556 1557 bridge_delete_member(sc, bif, 0); 1558 1559 return (0); 1560 } 1561 1562 static int 1563 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg) 1564 { 1565 struct ifbreq *req = arg; 1566 struct bridge_iflist *bif; 1567 struct bstp_port *bp; 1568 1569 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1570 if (bif == NULL) 1571 return (EXTERROR(ENOENT, "Interface is not a bridge member")); 1572 1573 bp = &bif->bif_stp; 1574 req->ifbr_ifsflags = bif->bif_flags; 1575 req->ifbr_state = bp->bp_state; 1576 req->ifbr_priority = bp->bp_priority; 1577 req->ifbr_path_cost = bp->bp_path_cost; 1578 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff; 1579 req->ifbr_proto = bp->bp_protover; 1580 req->ifbr_role = bp->bp_role; 1581 req->ifbr_stpflags = bp->bp_flags; 1582 req->ifbr_addrcnt = bif->bif_addrcnt; 1583 req->ifbr_addrmax = bif->bif_addrmax; 1584 req->ifbr_addrexceeded = bif->bif_addrexceeded; 1585 req->ifbr_pvid = bif->bif_pvid; 1586 req->ifbr_vlanproto = bif->bif_vlanproto; 1587 1588 /* Copy STP state options as flags */ 1589 if (bp->bp_operedge) 1590 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE; 1591 if (bp->bp_flags & BSTP_PORT_AUTOEDGE) 1592 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE; 1593 if (bp->bp_ptp_link) 1594 req->ifbr_ifsflags |= IFBIF_BSTP_PTP; 1595 if (bp->bp_flags & BSTP_PORT_AUTOPTP) 1596 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP; 1597 if (bp->bp_flags & BSTP_PORT_ADMEDGE) 1598 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE; 1599 if (bp->bp_flags & BSTP_PORT_ADMCOST) 1600 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST; 1601 return (0); 1602 } 1603 1604 static int 1605 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg) 1606 { 1607 struct epoch_tracker et; 1608 struct ifbreq *req = arg; 1609 struct bridge_iflist *bif; 1610 struct bstp_port *bp; 1611 int error; 1612 1613 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1614 if (bif == NULL) 1615 return (EXTERROR(ENOENT, "Interface is not a bridge member")); 1616 bp = &bif->bif_stp; 1617 1618 if (req->ifbr_ifsflags & IFBIF_SPAN) 1619 /* SPAN is readonly */ 1620 return (EXTERROR(EINVAL, "Span interface cannot be modified")); 1621 1622 NET_EPOCH_ENTER(et); 1623 1624 if (req->ifbr_ifsflags & IFBIF_STP) { 1625 if ((bif->bif_flags & IFBIF_STP) == 0) { 1626 error = bstp_enable(&bif->bif_stp); 1627 if (error) { 1628 NET_EPOCH_EXIT(et); 1629 return (EXTERROR(error, 1630 "Failed to enable STP")); 1631 } 1632 } 1633 } else { 1634 if ((bif->bif_flags & IFBIF_STP) != 0) 1635 bstp_disable(&bif->bif_stp); 1636 } 1637 1638 /* Pass on STP flags */ 1639 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0); 1640 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0); 1641 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0); 1642 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0); 1643 1644 /* Save the bits relating to the bridge */ 1645 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK; 1646 1647 NET_EPOCH_EXIT(et); 1648 1649 return (0); 1650 } 1651 1652 static int 1653 bridge_ioctl_scache(struct bridge_softc *sc, void *arg) 1654 { 1655 struct ifbrparam *param = arg; 1656 1657 sc->sc_brtmax = param->ifbrp_csize; 1658 bridge_rttrim(sc); 1659 1660 return (0); 1661 } 1662 1663 static int 1664 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg) 1665 { 1666 struct ifbrparam *param = arg; 1667 1668 param->ifbrp_csize = sc->sc_brtmax; 1669 1670 return (0); 1671 } 1672 1673 static int 1674 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg) 1675 { 1676 struct ifbifconf *bifc = arg; 1677 struct bridge_iflist *bif; 1678 struct ifbreq breq; 1679 char *buf, *outbuf; 1680 int count, buflen, len, error = 0; 1681 1682 count = 0; 1683 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) 1684 count++; 1685 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 1686 count++; 1687 1688 buflen = sizeof(breq) * count; 1689 if (bifc->ifbic_len == 0) { 1690 bifc->ifbic_len = buflen; 1691 return (0); 1692 } 1693 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO); 1694 if (outbuf == NULL) 1695 return (ENOMEM); 1696 1697 count = 0; 1698 buf = outbuf; 1699 len = min(bifc->ifbic_len, buflen); 1700 bzero(&breq, sizeof(breq)); 1701 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 1702 if (len < sizeof(breq)) 1703 break; 1704 1705 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname, 1706 sizeof(breq.ifbr_ifsname)); 1707 /* Fill in the ifbreq structure */ 1708 error = bridge_ioctl_gifflags(sc, &breq); 1709 if (error) 1710 break; 1711 memcpy(buf, &breq, sizeof(breq)); 1712 count++; 1713 buf += sizeof(breq); 1714 len -= sizeof(breq); 1715 } 1716 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) { 1717 if (len < sizeof(breq)) 1718 break; 1719 1720 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname, 1721 sizeof(breq.ifbr_ifsname)); 1722 breq.ifbr_ifsflags = bif->bif_flags; 1723 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff; 1724 memcpy(buf, &breq, sizeof(breq)); 1725 count++; 1726 buf += sizeof(breq); 1727 len -= sizeof(breq); 1728 } 1729 1730 bifc->ifbic_len = sizeof(breq) * count; 1731 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len); 1732 free(outbuf, M_TEMP); 1733 return (error); 1734 } 1735 1736 static int 1737 bridge_ioctl_rts(struct bridge_softc *sc, void *arg) 1738 { 1739 struct ifbaconf *bac = arg; 1740 struct bridge_rtnode *brt; 1741 struct ifbareq bareq; 1742 char *buf, *outbuf; 1743 int count, buflen, len, error = 0; 1744 1745 if (bac->ifbac_len == 0) 1746 return (0); 1747 1748 count = 0; 1749 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) 1750 count++; 1751 buflen = sizeof(bareq) * count; 1752 1753 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO); 1754 if (outbuf == NULL) 1755 return (ENOMEM); 1756 1757 count = 0; 1758 buf = outbuf; 1759 len = min(bac->ifbac_len, buflen); 1760 bzero(&bareq, sizeof(bareq)); 1761 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) { 1762 if (len < sizeof(bareq)) 1763 goto out; 1764 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname, 1765 sizeof(bareq.ifba_ifsname)); 1766 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr)); 1767 bareq.ifba_vlan = brt->brt_vlan; 1768 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 1769 time_uptime < brt->brt_expire) 1770 bareq.ifba_expire = brt->brt_expire - time_uptime; 1771 else 1772 bareq.ifba_expire = 0; 1773 bareq.ifba_flags = brt->brt_flags; 1774 1775 memcpy(buf, &bareq, sizeof(bareq)); 1776 count++; 1777 buf += sizeof(bareq); 1778 len -= sizeof(bareq); 1779 } 1780 out: 1781 bac->ifbac_len = sizeof(bareq) * count; 1782 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len); 1783 free(outbuf, M_TEMP); 1784 return (error); 1785 } 1786 1787 static int 1788 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg) 1789 { 1790 struct ifbareq *req = arg; 1791 struct bridge_iflist *bif; 1792 struct epoch_tracker et; 1793 int error; 1794 1795 NET_EPOCH_ENTER(et); 1796 bif = bridge_lookup_member(sc, req->ifba_ifsname); 1797 if (bif == NULL) { 1798 NET_EPOCH_EXIT(et); 1799 return (EXTERROR(ENOENT, "Interface is not a bridge member")); 1800 } 1801 1802 /* bridge_rtupdate() may acquire the lock. */ 1803 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1, 1804 req->ifba_flags); 1805 NET_EPOCH_EXIT(et); 1806 1807 return (error); 1808 } 1809 1810 static int 1811 bridge_ioctl_sto(struct bridge_softc *sc, void *arg) 1812 { 1813 struct ifbrparam *param = arg; 1814 1815 sc->sc_brttimeout = param->ifbrp_ctime; 1816 return (0); 1817 } 1818 1819 static int 1820 bridge_ioctl_gto(struct bridge_softc *sc, void *arg) 1821 { 1822 struct ifbrparam *param = arg; 1823 1824 param->ifbrp_ctime = sc->sc_brttimeout; 1825 return (0); 1826 } 1827 1828 static int 1829 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg) 1830 { 1831 struct ifbareq *req = arg; 1832 int vlan = req->ifba_vlan; 1833 1834 /* Userspace uses '0' to mean 'any vlan' */ 1835 if (vlan == 0) 1836 vlan = DOT1Q_VID_RSVD_IMPL; 1837 1838 return (bridge_rtdaddr(sc, req->ifba_dst, vlan)); 1839 } 1840 1841 static int 1842 bridge_ioctl_flush(struct bridge_softc *sc, void *arg) 1843 { 1844 struct ifbreq *req = arg; 1845 1846 BRIDGE_RT_LOCK(sc); 1847 bridge_rtflush(sc, req->ifbr_ifsflags); 1848 BRIDGE_RT_UNLOCK(sc); 1849 1850 return (0); 1851 } 1852 1853 static int 1854 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg) 1855 { 1856 struct ifbrparam *param = arg; 1857 struct bstp_state *bs = &sc->sc_stp; 1858 1859 param->ifbrp_prio = bs->bs_bridge_priority; 1860 return (0); 1861 } 1862 1863 static int 1864 bridge_ioctl_spri(struct bridge_softc *sc, void *arg) 1865 { 1866 struct ifbrparam *param = arg; 1867 1868 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio)); 1869 } 1870 1871 static int 1872 bridge_ioctl_ght(struct bridge_softc *sc, void *arg) 1873 { 1874 struct ifbrparam *param = arg; 1875 struct bstp_state *bs = &sc->sc_stp; 1876 1877 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8; 1878 return (0); 1879 } 1880 1881 static int 1882 bridge_ioctl_sht(struct bridge_softc *sc, void *arg) 1883 { 1884 struct ifbrparam *param = arg; 1885 1886 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime)); 1887 } 1888 1889 static int 1890 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg) 1891 { 1892 struct ifbrparam *param = arg; 1893 struct bstp_state *bs = &sc->sc_stp; 1894 1895 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8; 1896 return (0); 1897 } 1898 1899 static int 1900 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg) 1901 { 1902 struct ifbrparam *param = arg; 1903 1904 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay)); 1905 } 1906 1907 static int 1908 bridge_ioctl_gma(struct bridge_softc *sc, void *arg) 1909 { 1910 struct ifbrparam *param = arg; 1911 struct bstp_state *bs = &sc->sc_stp; 1912 1913 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8; 1914 return (0); 1915 } 1916 1917 static int 1918 bridge_ioctl_sma(struct bridge_softc *sc, void *arg) 1919 { 1920 struct ifbrparam *param = arg; 1921 1922 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage)); 1923 } 1924 1925 static int 1926 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg) 1927 { 1928 struct ifbreq *req = arg; 1929 struct bridge_iflist *bif; 1930 1931 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1932 if (bif == NULL) 1933 return (EXTERROR(ENOENT, "Interface is not a bridge member")); 1934 1935 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority)); 1936 } 1937 1938 static int 1939 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg) 1940 { 1941 struct ifbreq *req = arg; 1942 struct bridge_iflist *bif; 1943 1944 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1945 if (bif == NULL) 1946 return (EXTERROR(ENOENT, "Interface is not a bridge member")); 1947 1948 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost)); 1949 } 1950 1951 static int 1952 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg) 1953 { 1954 struct ifbreq *req = arg; 1955 struct bridge_iflist *bif; 1956 1957 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1958 if (bif == NULL) 1959 return (EXTERROR(ENOENT, "Interface is not a bridge member")); 1960 1961 bif->bif_addrmax = req->ifbr_addrmax; 1962 return (0); 1963 } 1964 1965 static int 1966 bridge_ioctl_sifpvid(struct bridge_softc *sc, void *arg) 1967 { 1968 struct ifbreq *req = arg; 1969 struct bridge_iflist *bif; 1970 1971 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1972 if (bif == NULL) 1973 return (EXTERROR(ENOENT, "Interface is not a bridge member")); 1974 1975 if (req->ifbr_pvid > DOT1Q_VID_MAX) 1976 return (EXTERROR(EINVAL, "Invalid VLAN ID")); 1977 1978 bif->bif_pvid = req->ifbr_pvid; 1979 return (0); 1980 } 1981 1982 static int 1983 bridge_ioctl_sifvlanset(struct bridge_softc *sc, void *arg) 1984 { 1985 struct ifbif_vlan_req *req = arg; 1986 struct bridge_iflist *bif; 1987 1988 bif = bridge_lookup_member(sc, req->bv_ifname); 1989 if (bif == NULL) 1990 return (EXTERROR(ENOENT, "Interface is not a bridge member")); 1991 1992 /* Reject invalid VIDs. */ 1993 if (BRVLAN_TEST(&req->bv_set, DOT1Q_VID_NULL) || 1994 BRVLAN_TEST(&req->bv_set, DOT1Q_VID_RSVD_IMPL)) 1995 return (EXTERROR(EINVAL, "Invalid VLAN ID in set")); 1996 1997 switch (req->bv_op) { 1998 /* Replace the existing vlan set with the new set */ 1999 case BRDG_VLAN_OP_SET: 2000 BIT_COPY(BRVLAN_SETSIZE, &req->bv_set, &bif->bif_vlan_set); 2001 break; 2002 2003 /* Modify the existing vlan set to add the given vlans */ 2004 case BRDG_VLAN_OP_ADD: 2005 BIT_OR(BRVLAN_SETSIZE, &bif->bif_vlan_set, &req->bv_set); 2006 break; 2007 2008 /* Modify the existing vlan set to remove the given vlans */ 2009 case BRDG_VLAN_OP_DEL: 2010 BIT_ANDNOT(BRVLAN_SETSIZE, &bif->bif_vlan_set, &req->bv_set); 2011 break; 2012 2013 /* Invalid or unknown operation */ 2014 default: 2015 return (EXTERROR(EINVAL, 2016 "Unsupported BRDGSIFVLANSET operation")); 2017 } 2018 2019 return (0); 2020 } 2021 2022 static int 2023 bridge_ioctl_gifvlanset(struct bridge_softc *sc, void *arg) 2024 { 2025 struct ifbif_vlan_req *req = arg; 2026 struct bridge_iflist *bif; 2027 2028 bif = bridge_lookup_member(sc, req->bv_ifname); 2029 if (bif == NULL) 2030 return (EXTERROR(ENOENT, "Interface is not a bridge member")); 2031 2032 BIT_COPY(BRVLAN_SETSIZE, &bif->bif_vlan_set, &req->bv_set); 2033 return (0); 2034 } 2035 2036 static int 2037 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg) 2038 { 2039 struct ifbreq *req = arg; 2040 struct bridge_iflist *bif = NULL; 2041 struct ifnet *ifs; 2042 2043 ifs = ifunit(req->ifbr_ifsname); 2044 if (ifs == NULL) 2045 return (EXTERROR(ENOENT, "No such interface")); 2046 2047 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 2048 if (ifs == bif->bif_ifp) 2049 return (EXTERROR(EBUSY, 2050 "Interface is already a span port")); 2051 2052 if (ifs->if_bridge != NULL) 2053 return (EXTERROR(EEXIST, 2054 "Interface is already a bridge member")); 2055 2056 switch (ifs->if_type) { 2057 case IFT_ETHER: 2058 case IFT_GIF: 2059 case IFT_L2VLAN: 2060 break; 2061 default: 2062 return (EXTERROR(EINVAL, "Unsupported interface type")); 2063 } 2064 2065 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO); 2066 if (bif == NULL) 2067 return (ENOMEM); 2068 2069 bif->bif_ifp = ifs; 2070 bif->bif_flags = IFBIF_SPAN; 2071 2072 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next); 2073 2074 return (0); 2075 } 2076 2077 static int 2078 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg) 2079 { 2080 struct ifbreq *req = arg; 2081 struct bridge_iflist *bif; 2082 struct ifnet *ifs; 2083 2084 ifs = ifunit(req->ifbr_ifsname); 2085 if (ifs == NULL) 2086 return (EXTERROR(ENOENT, "No such interface")); 2087 2088 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 2089 if (ifs == bif->bif_ifp) 2090 break; 2091 2092 if (bif == NULL) 2093 return (EXTERROR(ENOENT, "Interface is not a span port")); 2094 2095 bridge_delete_span(sc, bif); 2096 2097 return (0); 2098 } 2099 2100 static int 2101 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg) 2102 { 2103 struct ifbropreq *req = arg; 2104 struct bstp_state *bs = &sc->sc_stp; 2105 struct bstp_port *root_port; 2106 2107 req->ifbop_maxage = bs->bs_bridge_max_age >> 8; 2108 req->ifbop_hellotime = bs->bs_bridge_htime >> 8; 2109 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8; 2110 2111 root_port = bs->bs_root_port; 2112 if (root_port == NULL) 2113 req->ifbop_root_port = 0; 2114 else 2115 req->ifbop_root_port = root_port->bp_ifp->if_index; 2116 2117 req->ifbop_holdcount = bs->bs_txholdcount; 2118 req->ifbop_priority = bs->bs_bridge_priority; 2119 req->ifbop_protocol = bs->bs_protover; 2120 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost; 2121 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id; 2122 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id; 2123 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id; 2124 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec; 2125 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec; 2126 2127 return (0); 2128 } 2129 2130 static int 2131 bridge_ioctl_grte(struct bridge_softc *sc, void *arg) 2132 { 2133 struct ifbrparam *param = arg; 2134 2135 param->ifbrp_cexceeded = sc->sc_brtexceeded; 2136 return (0); 2137 } 2138 2139 static int 2140 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg) 2141 { 2142 struct ifbpstpconf *bifstp = arg; 2143 struct bridge_iflist *bif; 2144 struct bstp_port *bp; 2145 struct ifbpstpreq bpreq; 2146 char *buf, *outbuf; 2147 int count, buflen, len, error = 0; 2148 2149 count = 0; 2150 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 2151 if ((bif->bif_flags & IFBIF_STP) != 0) 2152 count++; 2153 } 2154 2155 buflen = sizeof(bpreq) * count; 2156 if (bifstp->ifbpstp_len == 0) { 2157 bifstp->ifbpstp_len = buflen; 2158 return (0); 2159 } 2160 2161 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO); 2162 if (outbuf == NULL) 2163 return (ENOMEM); 2164 2165 count = 0; 2166 buf = outbuf; 2167 len = min(bifstp->ifbpstp_len, buflen); 2168 bzero(&bpreq, sizeof(bpreq)); 2169 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 2170 if (len < sizeof(bpreq)) 2171 break; 2172 2173 if ((bif->bif_flags & IFBIF_STP) == 0) 2174 continue; 2175 2176 bp = &bif->bif_stp; 2177 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff; 2178 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions; 2179 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost; 2180 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id; 2181 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id; 2182 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id; 2183 2184 memcpy(buf, &bpreq, sizeof(bpreq)); 2185 count++; 2186 buf += sizeof(bpreq); 2187 len -= sizeof(bpreq); 2188 } 2189 2190 bifstp->ifbpstp_len = sizeof(bpreq) * count; 2191 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len); 2192 free(outbuf, M_TEMP); 2193 return (error); 2194 } 2195 2196 static int 2197 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg) 2198 { 2199 struct ifbrparam *param = arg; 2200 2201 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto)); 2202 } 2203 2204 static int 2205 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg) 2206 { 2207 struct ifbrparam *param = arg; 2208 2209 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc)); 2210 } 2211 2212 static int 2213 bridge_ioctl_gflags(struct bridge_softc *sc, void *arg) 2214 { 2215 struct ifbrparam *param = arg; 2216 2217 param->ifbrp_flags = sc->sc_flags; 2218 2219 return (0); 2220 } 2221 2222 static int 2223 bridge_ioctl_sflags(struct bridge_softc *sc, void *arg) 2224 { 2225 struct ifbrparam *param = arg; 2226 2227 sc->sc_flags = param->ifbrp_flags; 2228 2229 return (0); 2230 } 2231 2232 static int 2233 bridge_ioctl_gdefpvid(struct bridge_softc *sc, void *arg) 2234 { 2235 struct ifbrparam *param = arg; 2236 2237 param->ifbrp_defpvid = sc->sc_defpvid; 2238 2239 return (0); 2240 } 2241 2242 static int 2243 bridge_ioctl_sdefpvid(struct bridge_softc *sc, void *arg) 2244 { 2245 struct ifbrparam *param = arg; 2246 2247 /* Reject invalid VIDs, but allow 0 to mean 'none'. */ 2248 if (param->ifbrp_defpvid > DOT1Q_VID_MAX) 2249 return (EINVAL); 2250 2251 sc->sc_defpvid = param->ifbrp_defpvid; 2252 2253 return (0); 2254 } 2255 2256 static int 2257 bridge_ioctl_svlanproto(struct bridge_softc *sc, void *arg) 2258 { 2259 struct ifbreq *req = arg; 2260 struct bridge_iflist *bif; 2261 2262 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 2263 if (bif == NULL) 2264 return (EXTERROR(ENOENT, "Interface is not a bridge member")); 2265 2266 if (req->ifbr_vlanproto != ETHERTYPE_VLAN && 2267 req->ifbr_vlanproto != ETHERTYPE_QINQ) 2268 return (EXTERROR(EINVAL, "Invalid VLAN protocol")); 2269 2270 bif->bif_vlanproto = req->ifbr_vlanproto; 2271 2272 return (0); 2273 } 2274 /* 2275 * bridge_ifdetach: 2276 * 2277 * Detach an interface from a bridge. Called when a member 2278 * interface is detaching. 2279 */ 2280 static void 2281 bridge_ifdetach(void *arg __unused, struct ifnet *ifp) 2282 { 2283 struct bridge_iflist *bif = ifp->if_bridge; 2284 struct bridge_softc *sc = NULL; 2285 2286 if (bif) 2287 sc = bif->bif_sc; 2288 2289 if (V_bridge_cloner == NULL) { 2290 /* 2291 * This detach handler can be called after 2292 * vnet_bridge_uninit(). Just return in that case. 2293 */ 2294 return; 2295 } 2296 /* Check if the interface is a bridge member */ 2297 if (sc != NULL) { 2298 BRIDGE_LOCK(sc); 2299 bridge_delete_member(sc, bif, 1); 2300 BRIDGE_UNLOCK(sc); 2301 return; 2302 } 2303 2304 /* Check if the interface is a span port */ 2305 BRIDGE_LIST_LOCK(); 2306 LIST_FOREACH(sc, &V_bridge_list, sc_list) { 2307 BRIDGE_LOCK(sc); 2308 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) 2309 if (ifp == bif->bif_ifp) { 2310 bridge_delete_span(sc, bif); 2311 break; 2312 } 2313 2314 BRIDGE_UNLOCK(sc); 2315 } 2316 BRIDGE_LIST_UNLOCK(); 2317 } 2318 2319 /* 2320 * bridge_init: 2321 * 2322 * Initialize a bridge interface. 2323 */ 2324 static void 2325 bridge_init(void *xsc) 2326 { 2327 struct bridge_softc *sc = (struct bridge_softc *)xsc; 2328 struct ifnet *ifp = sc->sc_ifp; 2329 2330 if (ifp->if_drv_flags & IFF_DRV_RUNNING) 2331 return; 2332 2333 BRIDGE_LOCK(sc); 2334 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz, 2335 bridge_timer, sc); 2336 2337 ifp->if_drv_flags |= IFF_DRV_RUNNING; 2338 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */ 2339 2340 BRIDGE_UNLOCK(sc); 2341 } 2342 2343 /* 2344 * bridge_stop: 2345 * 2346 * Stop the bridge interface. 2347 */ 2348 static void 2349 bridge_stop(struct ifnet *ifp) 2350 { 2351 struct bridge_softc *sc = ifp->if_softc; 2352 2353 BRIDGE_LOCK_ASSERT(sc); 2354 2355 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) 2356 return; 2357 2358 BRIDGE_RT_LOCK(sc); 2359 callout_stop(&sc->sc_brcallout); 2360 2361 bstp_stop(&sc->sc_stp); 2362 2363 bridge_rtflush(sc, IFBF_FLUSHDYN); 2364 BRIDGE_RT_UNLOCK(sc); 2365 2366 ifp->if_drv_flags &= ~IFF_DRV_RUNNING; 2367 } 2368 2369 /* 2370 * bridge_enqueue: 2371 * 2372 * Enqueue a packet on a bridge member interface. 2373 * 2374 */ 2375 static int 2376 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m, 2377 struct bridge_iflist *bif) 2378 { 2379 int len, err = 0; 2380 short mflags; 2381 struct mbuf *m0; 2382 2383 /* 2384 * Find the bridge member port this packet is being sent on, if the 2385 * caller didn't already provide it. 2386 */ 2387 if (bif == NULL) 2388 bif = bridge_lookup_member_if(sc, dst_ifp); 2389 if (bif == NULL) { 2390 /* Perhaps the interface was removed from the bridge */ 2391 m_freem(m); 2392 return (EINVAL); 2393 } 2394 2395 /* Do VLAN filtering. */ 2396 if (!bridge_vfilter_out(bif, m)) { 2397 m_freem(m); 2398 return (0); 2399 } 2400 2401 /* We may be sending a fragment so traverse the mbuf */ 2402 for (; m; m = m0) { 2403 m0 = m->m_nextpkt; 2404 m->m_nextpkt = NULL; 2405 len = m->m_pkthdr.len; 2406 mflags = m->m_flags; 2407 2408 /* 2409 * If the native VLAN ID of the outgoing interface matches the 2410 * VLAN ID of the frame, remove the VLAN tag. 2411 */ 2412 if (bif->bif_pvid != DOT1Q_VID_NULL && 2413 VLANTAGOF(m) == bif->bif_pvid) { 2414 m->m_flags &= ~M_VLANTAG; 2415 m->m_pkthdr.ether_vtag = 0; 2416 } 2417 2418 /* 2419 * There are two cases where we have to insert our own tag: 2420 * if the member interface doesn't support hardware tagging, 2421 * or if the tag proto is not 802.1q. 2422 */ 2423 if ((m->m_flags & M_VLANTAG) && 2424 ((dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0 || 2425 bif->bif_vlanproto != ETHERTYPE_VLAN)) { 2426 m = ether_vlanencap_proto(m, m->m_pkthdr.ether_vtag, 2427 bif->bif_vlanproto); 2428 if (m == NULL) { 2429 if_printf(dst_ifp, 2430 "unable to prepend VLAN header\n"); 2431 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1); 2432 continue; 2433 } 2434 m->m_flags &= ~M_VLANTAG; 2435 } 2436 2437 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */ 2438 /* 2439 * XXXZL: gif(4) requires the af to be saved in csum_data field 2440 * so that gif_transmit() routine can pull it back. 2441 */ 2442 if (dst_ifp->if_type == IFT_GIF) 2443 m->m_pkthdr.csum_data = AF_LINK; 2444 if ((err = dst_ifp->if_transmit(dst_ifp, m))) { 2445 int n; 2446 2447 for (m = m0, n = 1; m != NULL; m = m0, n++) { 2448 m0 = m->m_nextpkt; 2449 m_freem(m); 2450 } 2451 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, n); 2452 break; 2453 } 2454 2455 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1); 2456 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len); 2457 if (mflags & M_MCAST) 2458 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1); 2459 } 2460 2461 return (err); 2462 } 2463 2464 /* 2465 * bridge_dummynet: 2466 * 2467 * Receive a queued packet from dummynet and pass it on to the output 2468 * interface. 2469 * 2470 * The mbuf has the Ethernet header already attached. 2471 */ 2472 static void 2473 bridge_dummynet(struct mbuf *m, struct ifnet *ifp) 2474 { 2475 struct bridge_iflist *bif = ifp->if_bridge; 2476 struct bridge_softc *sc = NULL; 2477 2478 if (bif) 2479 sc = bif->bif_sc; 2480 2481 /* 2482 * The packet didnt originate from a member interface. This should only 2483 * ever happen if a member interface is removed while packets are 2484 * queued for it. 2485 */ 2486 if (sc == NULL) { 2487 m_freem(m); 2488 return; 2489 } 2490 2491 if (PFIL_HOOKED_OUT_46) { 2492 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0) 2493 return; 2494 if (m == NULL) 2495 return; 2496 } 2497 2498 bridge_enqueue(sc, ifp, m, NULL); 2499 } 2500 2501 /* 2502 * bridge_output: 2503 * 2504 * Send output from a bridge member interface. This 2505 * performs the bridging function for locally originated 2506 * packets. 2507 * 2508 * The mbuf has the Ethernet header already attached. We must 2509 * enqueue or free the mbuf before returning. 2510 */ 2511 static int 2512 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa, 2513 struct rtentry *rt) 2514 { 2515 struct ether_header *eh; 2516 struct bridge_iflist *sbif; 2517 struct ifnet *bifp, *dst_if; 2518 struct bridge_softc *sc; 2519 ether_vlanid_t vlan; 2520 2521 NET_EPOCH_ASSERT(); 2522 2523 if ((m = m_pullup(m, ETHER_HDR_LEN)) == NULL) 2524 return (0); 2525 2526 sbif = ifp->if_bridge; 2527 sc = sbif->bif_sc; 2528 bifp = sc->sc_ifp; 2529 2530 eh = mtod(m, struct ether_header *); 2531 vlan = VLANTAGOF(m); 2532 2533 /* 2534 * If bridge is down, but the original output interface is up, 2535 * go ahead and send out that interface. Otherwise, the packet 2536 * is dropped below. 2537 */ 2538 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { 2539 dst_if = ifp; 2540 goto sendunicast; 2541 } 2542 2543 /* 2544 * If the packet is a multicast, or we don't know a better way to 2545 * get there, send to all interfaces. 2546 */ 2547 if (ETHER_IS_MULTICAST(eh->ether_dhost)) 2548 dst_if = NULL; 2549 else 2550 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan); 2551 /* Tap any traffic not passing back out the originating interface */ 2552 if (dst_if != ifp) 2553 ETHER_BPF_MTAP(bifp, m); 2554 if (dst_if == NULL) { 2555 struct bridge_iflist *bif; 2556 struct mbuf *mc; 2557 int used = 0; 2558 2559 bridge_span(sc, m); 2560 2561 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 2562 dst_if = bif->bif_ifp; 2563 2564 if (dst_if->if_type == IFT_GIF) 2565 continue; 2566 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2567 continue; 2568 2569 /* 2570 * If this is not the original output interface, 2571 * and the interface is participating in spanning 2572 * tree, make sure the port is in a state that 2573 * allows forwarding. 2574 */ 2575 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) && 2576 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2577 continue; 2578 2579 if (CK_LIST_NEXT(bif, bif_next) == NULL) { 2580 used = 1; 2581 mc = m; 2582 } else { 2583 mc = m_dup(m, M_NOWAIT); 2584 if (mc == NULL) { 2585 if_inc_counter(bifp, IFCOUNTER_OERRORS, 1); 2586 continue; 2587 } 2588 } 2589 2590 bridge_enqueue(sc, dst_if, mc, bif); 2591 } 2592 if (used == 0) 2593 m_freem(m); 2594 return (0); 2595 } 2596 2597 sendunicast: 2598 /* 2599 * XXX Spanning tree consideration here? 2600 */ 2601 2602 bridge_span(sc, m); 2603 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) { 2604 m_freem(m); 2605 return (0); 2606 } 2607 2608 bridge_enqueue(sc, dst_if, m, NULL); 2609 return (0); 2610 } 2611 2612 /* 2613 * bridge_transmit: 2614 * 2615 * Do output on a bridge. 2616 * 2617 */ 2618 static int 2619 bridge_transmit(struct ifnet *ifp, struct mbuf *m) 2620 { 2621 struct bridge_softc *sc; 2622 struct ether_header *eh; 2623 struct ifnet *dst_if; 2624 int error = 0; 2625 ether_vlanid_t vlan; 2626 2627 sc = ifp->if_softc; 2628 2629 ETHER_BPF_MTAP(ifp, m); 2630 2631 eh = mtod(m, struct ether_header *); 2632 vlan = VLANTAGOF(m); 2633 2634 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) && 2635 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan)) != NULL) { 2636 error = bridge_enqueue(sc, dst_if, m, NULL); 2637 } else 2638 bridge_broadcast(sc, ifp, m, 0); 2639 2640 return (error); 2641 } 2642 2643 #ifdef ALTQ 2644 static void 2645 bridge_altq_start(if_t ifp) 2646 { 2647 struct ifaltq *ifq = &ifp->if_snd; 2648 struct mbuf *m; 2649 2650 IFQ_LOCK(ifq); 2651 IFQ_DEQUEUE_NOLOCK(ifq, m); 2652 while (m != NULL) { 2653 bridge_transmit(ifp, m); 2654 IFQ_DEQUEUE_NOLOCK(ifq, m); 2655 } 2656 IFQ_UNLOCK(ifq); 2657 } 2658 2659 static int 2660 bridge_altq_transmit(if_t ifp, struct mbuf *m) 2661 { 2662 int err; 2663 2664 if (ALTQ_IS_ENABLED(&ifp->if_snd)) { 2665 IFQ_ENQUEUE(&ifp->if_snd, m, err); 2666 if (err == 0) 2667 bridge_altq_start(ifp); 2668 } else 2669 err = bridge_transmit(ifp, m); 2670 2671 return (err); 2672 } 2673 #endif /* ALTQ */ 2674 2675 /* 2676 * The ifp->if_qflush entry point for if_bridge(4) is no-op. 2677 */ 2678 static void 2679 bridge_qflush(struct ifnet *ifp __unused) 2680 { 2681 } 2682 2683 /* 2684 * bridge_forward: 2685 * 2686 * The forwarding function of the bridge. 2687 * 2688 * NOTE: Releases the lock on return. 2689 */ 2690 static void 2691 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif, 2692 struct mbuf *m) 2693 { 2694 struct bridge_iflist *dbif; 2695 struct ifnet *src_if, *dst_if, *ifp; 2696 struct ether_header *eh; 2697 uint8_t *dst; 2698 int error; 2699 ether_vlanid_t vlan; 2700 2701 NET_EPOCH_ASSERT(); 2702 2703 src_if = m->m_pkthdr.rcvif; 2704 ifp = sc->sc_ifp; 2705 vlan = VLANTAGOF(m); 2706 2707 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1); 2708 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); 2709 2710 if ((sbif->bif_flags & IFBIF_STP) && 2711 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2712 goto drop; 2713 2714 eh = mtod(m, struct ether_header *); 2715 dst = eh->ether_dhost; 2716 2717 /* If the interface is learning, record the address. */ 2718 if (sbif->bif_flags & IFBIF_LEARNING) { 2719 error = bridge_rtupdate(sc, eh->ether_shost, vlan, 2720 sbif, 0, IFBAF_DYNAMIC); 2721 /* 2722 * If the interface has addresses limits then deny any source 2723 * that is not in the cache. 2724 */ 2725 if (error && sbif->bif_addrmax) 2726 goto drop; 2727 } 2728 2729 if ((sbif->bif_flags & IFBIF_STP) != 0 && 2730 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING) 2731 goto drop; 2732 2733 #ifdef DEV_NETMAP 2734 /* 2735 * Hand the packet to netmap only if it wasn't injected by netmap 2736 * itself. 2737 */ 2738 if ((m->m_flags & M_BRIDGE_INJECT) == 0 && 2739 (if_getcapenable(ifp) & IFCAP_NETMAP) != 0) { 2740 ifp->if_input(ifp, m); 2741 return; 2742 } 2743 m->m_flags &= ~M_BRIDGE_INJECT; 2744 #endif 2745 2746 /* 2747 * At this point, the port either doesn't participate 2748 * in spanning tree or it is in the forwarding state. 2749 */ 2750 2751 /* 2752 * If the packet is unicast, destined for someone on 2753 * "this" side of the bridge, drop it. 2754 */ 2755 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) { 2756 dst_if = bridge_rtlookup(sc, dst, vlan); 2757 if (src_if == dst_if) 2758 goto drop; 2759 } else { 2760 /* 2761 * Check if its a reserved multicast address, any address 2762 * listed in 802.1D section 7.12.6 may not be forwarded by the 2763 * bridge. 2764 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F 2765 */ 2766 if (dst[0] == 0x01 && dst[1] == 0x80 && 2767 dst[2] == 0xc2 && dst[3] == 0x00 && 2768 dst[4] == 0x00 && dst[5] <= 0x0f) 2769 goto drop; 2770 2771 /* ...forward it to all interfaces. */ 2772 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1); 2773 dst_if = NULL; 2774 } 2775 2776 /* 2777 * If we have a destination interface which is a member of our bridge, 2778 * OR this is a unicast packet, push it through the bpf(4) machinery. 2779 * For broadcast or multicast packets, don't bother because it will 2780 * be reinjected into ether_input. We do this before we pass the packets 2781 * through the pfil(9) framework, as it is possible that pfil(9) will 2782 * drop the packet, or possibly modify it, making it difficult to debug 2783 * firewall issues on the bridge. 2784 */ 2785 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0) 2786 ETHER_BPF_MTAP(ifp, m); 2787 2788 /* run the packet filter */ 2789 if (PFIL_HOOKED_IN_46) { 2790 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0) 2791 return; 2792 if (m == NULL) 2793 return; 2794 } 2795 2796 if (dst_if == NULL) { 2797 bridge_broadcast(sc, src_if, m, 1); 2798 return; 2799 } 2800 2801 /* 2802 * At this point, we're dealing with a unicast frame 2803 * going to a different interface. 2804 */ 2805 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 2806 goto drop; 2807 2808 dbif = bridge_lookup_member_if(sc, dst_if); 2809 if (dbif == NULL) 2810 /* Not a member of the bridge (anymore?) */ 2811 goto drop; 2812 2813 /* Private segments can not talk to each other */ 2814 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE) 2815 goto drop; 2816 2817 if ((dbif->bif_flags & IFBIF_STP) && 2818 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 2819 goto drop; 2820 2821 if (PFIL_HOOKED_OUT_46) { 2822 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0) 2823 return; 2824 if (m == NULL) 2825 return; 2826 } 2827 2828 bridge_enqueue(sc, dst_if, m, dbif); 2829 return; 2830 2831 drop: 2832 m_freem(m); 2833 } 2834 2835 /* 2836 * bridge_input: 2837 * 2838 * Receive input from a member interface. Queue the packet for 2839 * bridging if it is not for us. 2840 */ 2841 static struct mbuf * 2842 bridge_input(struct ifnet *ifp, struct mbuf *m) 2843 { 2844 struct bridge_softc *sc = NULL; 2845 struct bridge_iflist *bif, *bif2; 2846 struct ifnet *bifp; 2847 struct ether_header *eh; 2848 struct mbuf *mc, *mc2; 2849 ether_vlanid_t vlan; 2850 int error; 2851 2852 NET_EPOCH_ASSERT(); 2853 2854 /* We need the Ethernet header later, so make sure we have it now. */ 2855 if ((m = m_pullup(m, ETHER_HDR_LEN)) == NULL) 2856 return (NULL); 2857 2858 eh = mtod(m, struct ether_header *); 2859 vlan = VLANTAGOF(m); 2860 2861 /* 2862 * If this frame has a VLAN tag and the receiving interface has a 2863 * vlan(4) trunk, then it is is destined for vlan(4), not for us. 2864 * This means if vlan(4) and bridge(4) are configured on the same 2865 * interface, vlan(4) is preferred, which is what users typically 2866 * expect. 2867 */ 2868 if (vlan != DOT1Q_VID_NULL && ifp->if_vlantrunk != NULL) 2869 return (m); 2870 2871 bif = ifp->if_bridge; 2872 if (bif) 2873 sc = bif->bif_sc; 2874 2875 if (sc == NULL) { 2876 /* 2877 * This packet originated from the bridge itself, so it must 2878 * have been transmitted by netmap. Derive the "source" 2879 * interface from the source address and drop the packet if the 2880 * source address isn't known. 2881 */ 2882 KASSERT((m->m_flags & M_BRIDGE_INJECT) != 0, 2883 ("%s: ifnet %p missing a bridge softc", __func__, ifp)); 2884 sc = if_getsoftc(ifp); 2885 ifp = bridge_rtlookup(sc, eh->ether_shost, vlan); 2886 if (ifp == NULL) { 2887 if_inc_counter(sc->sc_ifp, IFCOUNTER_IERRORS, 1); 2888 m_freem(m); 2889 return (NULL); 2890 } 2891 m->m_pkthdr.rcvif = ifp; 2892 } 2893 bifp = sc->sc_ifp; 2894 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) 2895 return (m); 2896 2897 /* 2898 * Implement support for bridge monitoring. If this flag has been 2899 * set on this interface, discard the packet once we push it through 2900 * the bpf(4) machinery, but before we do, increment the byte and 2901 * packet counters associated with this interface. 2902 */ 2903 if ((bifp->if_flags & IFF_MONITOR) != 0) { 2904 m->m_pkthdr.rcvif = bifp; 2905 ETHER_BPF_MTAP(bifp, m); 2906 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); 2907 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); 2908 m_freem(m); 2909 return (NULL); 2910 } 2911 2912 /* Do VLAN filtering. */ 2913 if (!bridge_vfilter_in(bif, m)) { 2914 if_inc_counter(sc->sc_ifp, IFCOUNTER_IERRORS, 1); 2915 m_freem(m); 2916 return (NULL); 2917 } 2918 /* bridge_vfilter_in() may add a tag */ 2919 vlan = VLANTAGOF(m); 2920 2921 bridge_span(sc, m); 2922 2923 if (m->m_flags & (M_BCAST|M_MCAST)) { 2924 /* Tap off 802.1D packets; they do not get forwarded. */ 2925 if (memcmp(eh->ether_dhost, bstp_etheraddr, 2926 ETHER_ADDR_LEN) == 0) { 2927 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */ 2928 return (NULL); 2929 } 2930 2931 if ((bif->bif_flags & IFBIF_STP) && 2932 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) { 2933 return (m); 2934 } 2935 2936 /* 2937 * Make a deep copy of the packet and enqueue the copy 2938 * for bridge processing; return the original packet for 2939 * local processing. 2940 */ 2941 mc = m_dup(m, M_NOWAIT); 2942 if (mc == NULL) { 2943 return (m); 2944 } 2945 2946 /* Perform the bridge forwarding function with the copy. */ 2947 bridge_forward(sc, bif, mc); 2948 2949 #ifdef DEV_NETMAP 2950 /* 2951 * If netmap is enabled and has not already seen this packet, 2952 * then it will be consumed by bridge_forward(). 2953 */ 2954 if ((if_getcapenable(bifp) & IFCAP_NETMAP) != 0 && 2955 (m->m_flags & M_BRIDGE_INJECT) == 0) { 2956 m_freem(m); 2957 return (NULL); 2958 } 2959 #endif 2960 2961 /* 2962 * Reinject the mbuf as arriving on the bridge so we have a 2963 * chance at claiming multicast packets. We can not loop back 2964 * here from ether_input as a bridge is never a member of a 2965 * bridge. 2966 */ 2967 KASSERT(bifp->if_bridge == NULL, 2968 ("loop created in bridge_input")); 2969 mc2 = m_dup(m, M_NOWAIT); 2970 if (mc2 != NULL) { 2971 /* Keep the layer3 header aligned */ 2972 int i = min(mc2->m_pkthdr.len, max_protohdr); 2973 mc2 = m_copyup(mc2, i, ETHER_ALIGN); 2974 } 2975 if (mc2 != NULL) { 2976 mc2->m_pkthdr.rcvif = bifp; 2977 mc2->m_flags &= ~M_BRIDGE_INJECT; 2978 sc->sc_if_input(bifp, mc2); 2979 } 2980 2981 /* Return the original packet for local processing. */ 2982 return (m); 2983 } 2984 2985 if ((bif->bif_flags & IFBIF_STP) && 2986 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) { 2987 return (m); 2988 } 2989 2990 #if defined(INET) || defined(INET6) 2991 #define CARP_CHECK_WE_ARE_DST(iface) \ 2992 ((iface)->if_carp && (*carp_forus_p)((iface), eh->ether_dhost)) 2993 #define CARP_CHECK_WE_ARE_SRC(iface) \ 2994 ((iface)->if_carp && (*carp_forus_p)((iface), eh->ether_shost)) 2995 #else 2996 #define CARP_CHECK_WE_ARE_DST(iface) false 2997 #define CARP_CHECK_WE_ARE_SRC(iface) false 2998 #endif 2999 3000 #ifdef DEV_NETMAP 3001 #define GRAB_FOR_NETMAP(ifp, m) do { \ 3002 if ((if_getcapenable(ifp) & IFCAP_NETMAP) != 0 && \ 3003 ((m)->m_flags & M_BRIDGE_INJECT) == 0) { \ 3004 (ifp)->if_input(ifp, m); \ 3005 return (NULL); \ 3006 } \ 3007 } while (0) 3008 #else 3009 #define GRAB_FOR_NETMAP(ifp, m) 3010 #endif 3011 3012 #define GRAB_OUR_PACKETS(iface) \ 3013 if ((iface)->if_type == IFT_GIF) \ 3014 continue; \ 3015 /* It is destined for us. */ \ 3016 if (memcmp(IF_LLADDR(iface), eh->ether_dhost, ETHER_ADDR_LEN) == 0 || \ 3017 CARP_CHECK_WE_ARE_DST(iface)) { \ 3018 if (bif->bif_flags & IFBIF_LEARNING) { \ 3019 error = bridge_rtupdate(sc, eh->ether_shost, \ 3020 vlan, bif, 0, IFBAF_DYNAMIC); \ 3021 if (error && bif->bif_addrmax) { \ 3022 m_freem(m); \ 3023 return (NULL); \ 3024 } \ 3025 } \ 3026 m->m_pkthdr.rcvif = iface; \ 3027 if ((iface) == ifp) { \ 3028 /* Skip bridge processing... src == dest */ \ 3029 return (m); \ 3030 } \ 3031 /* It's passing over or to the bridge, locally. */ \ 3032 ETHER_BPF_MTAP(bifp, m); \ 3033 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \ 3034 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);\ 3035 /* Hand the packet over to netmap if necessary. */ \ 3036 GRAB_FOR_NETMAP(bifp, m); \ 3037 /* Filter on the physical interface. */ \ 3038 if (V_pfil_local_phys && PFIL_HOOKED_IN_46) { \ 3039 if (bridge_pfil(&m, NULL, ifp, \ 3040 PFIL_IN) != 0 || m == NULL) { \ 3041 return (NULL); \ 3042 } \ 3043 } \ 3044 if ((iface) != bifp) \ 3045 ETHER_BPF_MTAP(iface, m); \ 3046 /* Pass tagged packets to if_vlan, if it's loaded */ \ 3047 if (VLANTAGOF(m) != 0) { \ 3048 if (bifp->if_vlantrunk == NULL) { \ 3049 m_freem(m); \ 3050 return (NULL); \ 3051 } \ 3052 (*vlan_input_p)(bifp, m); \ 3053 return (NULL); \ 3054 } \ 3055 return (m); \ 3056 } \ 3057 \ 3058 /* We just received a packet that we sent out. */ \ 3059 if (memcmp(IF_LLADDR(iface), eh->ether_shost, ETHER_ADDR_LEN) == 0 || \ 3060 CARP_CHECK_WE_ARE_SRC(iface)) { \ 3061 m_freem(m); \ 3062 return (NULL); \ 3063 } 3064 3065 /* 3066 * Unicast. Make sure it's not for the bridge. 3067 */ 3068 do { GRAB_OUR_PACKETS(bifp) } while (0); 3069 3070 /* 3071 * If member_ifaddrs is enabled, see if the packet is destined for 3072 * one of the members' addresses. 3073 */ 3074 if (V_member_ifaddrs) { 3075 /* Check the interface the packet arrived on. */ 3076 do { GRAB_OUR_PACKETS(ifp) } while (0); 3077 3078 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) { 3079 GRAB_OUR_PACKETS(bif2->bif_ifp) 3080 } 3081 } 3082 3083 #undef CARP_CHECK_WE_ARE_DST 3084 #undef CARP_CHECK_WE_ARE_SRC 3085 #undef GRAB_FOR_NETMAP 3086 #undef GRAB_OUR_PACKETS 3087 3088 /* Perform the bridge forwarding function. */ 3089 bridge_forward(sc, bif, m); 3090 3091 return (NULL); 3092 } 3093 3094 /* 3095 * Inject a packet back into the host ethernet stack. This will generally only 3096 * be used by netmap when an application writes to the host TX ring. The 3097 * M_BRIDGE_INJECT flag ensures that the packet is re-routed to the bridge 3098 * interface after ethernet processing. 3099 */ 3100 static void 3101 bridge_inject(struct ifnet *ifp, struct mbuf *m) 3102 { 3103 struct bridge_softc *sc; 3104 3105 if (ifp->if_type == IFT_L2VLAN) { 3106 /* 3107 * vlan(4) gives us the vlan ifnet, so we need to get the 3108 * bridge softc to get a pointer to ether_input to send the 3109 * packet to. 3110 */ 3111 struct ifnet *bifp = NULL; 3112 3113 if (vlan_trunkdev_p == NULL) { 3114 m_freem(m); 3115 return; 3116 } 3117 3118 bifp = vlan_trunkdev_p(ifp); 3119 if (bifp == NULL) { 3120 m_freem(m); 3121 return; 3122 } 3123 3124 sc = if_getsoftc(bifp); 3125 sc->sc_if_input(ifp, m); 3126 return; 3127 } 3128 3129 KASSERT((if_getcapenable(ifp) & IFCAP_NETMAP) != 0, 3130 ("%s: iface %s is not running in netmap mode", 3131 __func__, if_name(ifp))); 3132 KASSERT((m->m_flags & M_BRIDGE_INJECT) == 0, 3133 ("%s: mbuf %p has M_BRIDGE_INJECT set", __func__, m)); 3134 3135 m->m_flags |= M_BRIDGE_INJECT; 3136 sc = if_getsoftc(ifp); 3137 sc->sc_if_input(ifp, m); 3138 } 3139 3140 /* 3141 * bridge_broadcast: 3142 * 3143 * Send a frame to all interfaces that are members of 3144 * the bridge, except for the one on which the packet 3145 * arrived. 3146 * 3147 * NOTE: Releases the lock on return. 3148 */ 3149 static void 3150 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if, 3151 struct mbuf *m, int runfilt) 3152 { 3153 struct bridge_iflist *dbif, *sbif; 3154 struct mbuf *mc; 3155 struct ifnet *dst_if; 3156 int used = 0, i; 3157 3158 NET_EPOCH_ASSERT(); 3159 3160 sbif = bridge_lookup_member_if(sc, src_if); 3161 3162 /* Filter on the bridge interface before broadcasting */ 3163 if (runfilt && PFIL_HOOKED_OUT_46) { 3164 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0) 3165 return; 3166 if (m == NULL) 3167 return; 3168 } 3169 3170 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) { 3171 dst_if = dbif->bif_ifp; 3172 if (dst_if == src_if) 3173 continue; 3174 3175 /* Private segments can not talk to each other */ 3176 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)) 3177 continue; 3178 3179 if ((dbif->bif_flags & IFBIF_STP) && 3180 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) 3181 continue; 3182 3183 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 && 3184 (m->m_flags & (M_BCAST|M_MCAST)) == 0) 3185 continue; 3186 3187 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 3188 continue; 3189 3190 if (CK_LIST_NEXT(dbif, bif_next) == NULL) { 3191 mc = m; 3192 used = 1; 3193 } else { 3194 mc = m_dup(m, M_NOWAIT); 3195 if (mc == NULL) { 3196 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 3197 continue; 3198 } 3199 } 3200 3201 /* 3202 * Filter on the output interface. Pass a NULL bridge interface 3203 * pointer so we do not redundantly filter on the bridge for 3204 * each interface we broadcast on. 3205 */ 3206 if (runfilt && PFIL_HOOKED_OUT_46) { 3207 if (used == 0) { 3208 /* Keep the layer3 header aligned */ 3209 i = min(mc->m_pkthdr.len, max_protohdr); 3210 mc = m_copyup(mc, i, ETHER_ALIGN); 3211 if (mc == NULL) { 3212 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 3213 continue; 3214 } 3215 } 3216 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0) 3217 continue; 3218 if (mc == NULL) 3219 continue; 3220 } 3221 3222 bridge_enqueue(sc, dst_if, mc, dbif); 3223 } 3224 if (used == 0) 3225 m_freem(m); 3226 } 3227 3228 /* 3229 * bridge_span: 3230 * 3231 * Duplicate a packet out one or more interfaces that are in span mode, 3232 * the original mbuf is unmodified. 3233 */ 3234 static void 3235 bridge_span(struct bridge_softc *sc, struct mbuf *m) 3236 { 3237 struct bridge_iflist *bif; 3238 struct ifnet *dst_if; 3239 struct mbuf *mc; 3240 3241 NET_EPOCH_ASSERT(); 3242 3243 if (CK_LIST_EMPTY(&sc->sc_spanlist)) 3244 return; 3245 3246 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) { 3247 dst_if = bif->bif_ifp; 3248 3249 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) 3250 continue; 3251 3252 mc = m_dup(m, M_NOWAIT); 3253 if (mc == NULL) { 3254 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); 3255 continue; 3256 } 3257 3258 bridge_enqueue(sc, dst_if, mc, bif); 3259 } 3260 } 3261 3262 /* 3263 * Incoming VLAN filtering. Given a frame and the member interface it was 3264 * received on, decide whether the port configuration allows it. 3265 */ 3266 static bool 3267 bridge_vfilter_in(const struct bridge_iflist *sbif, struct mbuf *m) 3268 { 3269 ether_vlanid_t vlan; 3270 3271 vlan = VLANTAGOF(m); 3272 /* Make sure the vlan id is reasonable. */ 3273 if (vlan > DOT1Q_VID_MAX) 3274 return (false); 3275 3276 /* 3277 * If VLAN filtering isn't enabled, pass everything, but add a tag 3278 * if the port has a pvid configured. 3279 */ 3280 if ((sbif->bif_sc->sc_flags & IFBRF_VLANFILTER) == 0) { 3281 if (vlan == DOT1Q_VID_NULL && 3282 sbif->bif_pvid != DOT1Q_VID_NULL) { 3283 m->m_pkthdr.ether_vtag = sbif->bif_pvid; 3284 m->m_flags |= M_VLANTAG; 3285 } 3286 3287 return (true); 3288 } 3289 3290 /* If Q-in-Q is disabled, check for stacked tags. */ 3291 if ((sbif->bif_flags & IFBIF_QINQ) == 0) { 3292 struct ether_header *eh; 3293 uint16_t proto; 3294 3295 eh = mtod(m, struct ether_header *); 3296 proto = ntohs(eh->ether_type); 3297 3298 if (proto == ETHERTYPE_VLAN || proto == ETHERTYPE_QINQ) 3299 return (false); 3300 } 3301 3302 if (vlan == DOT1Q_VID_NULL) { 3303 /* 3304 * The frame doesn't have a tag. If the interface does not 3305 * have an untagged vlan configured, drop the frame. 3306 */ 3307 if (sbif->bif_pvid == DOT1Q_VID_NULL) 3308 return (false); 3309 3310 /* 3311 * Otherwise, insert a new tag based on the interface's 3312 * untagged vlan id. 3313 */ 3314 m->m_pkthdr.ether_vtag = sbif->bif_pvid; 3315 m->m_flags |= M_VLANTAG; 3316 } else { 3317 /* 3318 * The frame has a tag, so check it matches the interface's 3319 * vlan access list. We explicitly do not accept tagged 3320 * frames for the untagged vlan id here (unless it's also 3321 * in the access list). 3322 */ 3323 if (!BRVLAN_TEST(&sbif->bif_vlan_set, vlan)) 3324 return (false); 3325 } 3326 3327 /* Accept the frame. */ 3328 return (true); 3329 } 3330 3331 /* 3332 * Outgoing VLAN filtering. Given a frame, its vlan, and the member interface 3333 * we intend to send it to, decide whether the port configuration allows it to 3334 * be sent. 3335 */ 3336 static bool 3337 bridge_vfilter_out(const struct bridge_iflist *dbif, const struct mbuf *m) 3338 { 3339 struct ether_header *eh; 3340 ether_vlanid_t vlan; 3341 3342 NET_EPOCH_ASSERT(); 3343 3344 /* 3345 * If the interface is in span mode, then bif_sc will be NULL. 3346 * Since the purpose of span interfaces is to receive all frames, 3347 * pass everything. 3348 */ 3349 if (dbif->bif_sc == NULL) 3350 return (true); 3351 3352 /* If VLAN filtering isn't enabled, pass everything. */ 3353 if ((dbif->bif_sc->sc_flags & IFBRF_VLANFILTER) == 0) 3354 return (true); 3355 3356 vlan = VLANTAGOF(m); 3357 3358 /* 3359 * Always allow untagged 802.1D STP frames, even if they would 3360 * otherwise be dropped. This is required for STP to work on 3361 * a filtering bridge. 3362 * 3363 * Tagged STP (Cisco PVST+) is a non-standard extension, so 3364 * handle those frames via the normal filtering path. 3365 */ 3366 eh = mtod(m, struct ether_header *); 3367 if (vlan == DOT1Q_VID_NULL && 3368 memcmp(eh->ether_dhost, bstp_etheraddr, ETHER_ADDR_LEN) == 0) 3369 return (true); 3370 3371 /* 3372 * If the frame wasn't assigned to a vlan at ingress, drop it. 3373 * We can't forward these frames to filtering ports because we 3374 * don't know what VLAN they're supposed to be in. 3375 */ 3376 if (vlan == DOT1Q_VID_NULL) 3377 return (false); 3378 3379 /* 3380 * If the frame's vlan matches the interfaces's untagged vlan, 3381 * allow it. 3382 */ 3383 if (vlan == dbif->bif_pvid) 3384 return (true); 3385 3386 /* 3387 * If the frame's vlan is on the interface's tagged access list, 3388 * allow it. 3389 */ 3390 if (BRVLAN_TEST(&dbif->bif_vlan_set, vlan)) 3391 return (true); 3392 3393 /* The frame was not permitted, so drop it. */ 3394 return (false); 3395 } 3396 3397 /* 3398 * bridge_rtupdate: 3399 * 3400 * Add a bridge routing entry. 3401 */ 3402 static int 3403 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, 3404 ether_vlanid_t vlan, struct bridge_iflist *bif, 3405 int setflags, uint8_t flags) 3406 { 3407 struct bridge_rtnode *brt; 3408 struct bridge_iflist *obif; 3409 int error; 3410 3411 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc); 3412 3413 /* Check the source address is valid and not multicast. */ 3414 if (ETHER_IS_MULTICAST(dst)) 3415 return (EXTERROR(EINVAL, "Multicast address not permitted")); 3416 if (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 && 3417 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) 3418 return (EXTERROR(EINVAL, "Zero address not permitted")); 3419 3420 /* 3421 * A route for this destination might already exist. If so, 3422 * update it, otherwise create a new one. 3423 */ 3424 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) { 3425 BRIDGE_RT_LOCK(sc); 3426 3427 /* Check again, now that we have the lock. There could have 3428 * been a race and we only want to insert this once. */ 3429 if (bridge_rtnode_lookup(sc, dst, vlan) != NULL) { 3430 BRIDGE_RT_UNLOCK(sc); 3431 return (0); 3432 } 3433 3434 if (sc->sc_brtcnt >= sc->sc_brtmax) { 3435 sc->sc_brtexceeded++; 3436 BRIDGE_RT_UNLOCK(sc); 3437 return (EXTERROR(ENOSPC, "Address table is full")); 3438 } 3439 /* Check per interface address limits (if enabled) */ 3440 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) { 3441 bif->bif_addrexceeded++; 3442 BRIDGE_RT_UNLOCK(sc); 3443 return (EXTERROR(ENOSPC, 3444 "Interface address limit exceeded")); 3445 } 3446 3447 /* 3448 * Allocate a new bridge forwarding node, and 3449 * initialize the expiration time and Ethernet 3450 * address. 3451 */ 3452 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO); 3453 if (brt == NULL) { 3454 BRIDGE_RT_UNLOCK(sc); 3455 return (EXTERROR(ENOMEM, 3456 "Cannot allocate address node")); 3457 } 3458 brt->brt_vnet = curvnet; 3459 3460 if (bif->bif_flags & IFBIF_STICKY) 3461 brt->brt_flags = IFBAF_STICKY; 3462 else 3463 brt->brt_flags = IFBAF_DYNAMIC; 3464 3465 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN); 3466 brt->brt_vlan = vlan; 3467 3468 brt->brt_dst = bif; 3469 if ((error = bridge_rtnode_insert(sc, brt)) != 0) { 3470 uma_zfree(V_bridge_rtnode_zone, brt); 3471 BRIDGE_RT_UNLOCK(sc); 3472 return (error); 3473 } 3474 bif->bif_addrcnt++; 3475 3476 BRIDGE_RT_UNLOCK(sc); 3477 } 3478 3479 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 3480 (obif = brt->brt_dst) != bif) { 3481 MPASS(obif != NULL); 3482 3483 BRIDGE_RT_LOCK(sc); 3484 brt->brt_dst->bif_addrcnt--; 3485 brt->brt_dst = bif; 3486 brt->brt_dst->bif_addrcnt++; 3487 BRIDGE_RT_UNLOCK(sc); 3488 3489 if (V_log_mac_flap && 3490 ppsratecheck(&V_log_last, &V_log_count, V_log_interval)) { 3491 log(LOG_NOTICE, 3492 "%s: mac address %6D vlan %d moved from %s to %s\n", 3493 sc->sc_ifp->if_xname, 3494 &brt->brt_addr[0], ":", 3495 brt->brt_vlan, 3496 obif->bif_ifp->if_xname, 3497 bif->bif_ifp->if_xname); 3498 } 3499 } 3500 3501 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 3502 brt->brt_expire = time_uptime + sc->sc_brttimeout; 3503 if (setflags) 3504 brt->brt_flags = flags; 3505 3506 return (0); 3507 } 3508 3509 /* 3510 * bridge_rtlookup: 3511 * 3512 * Lookup the destination interface for an address. 3513 */ 3514 static struct ifnet * 3515 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, 3516 ether_vlanid_t vlan) 3517 { 3518 struct bridge_rtnode *brt; 3519 3520 NET_EPOCH_ASSERT(); 3521 3522 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL) 3523 return (NULL); 3524 3525 return (brt->brt_ifp); 3526 } 3527 3528 /* 3529 * bridge_rttrim: 3530 * 3531 * Trim the routine table so that we have a number 3532 * of routing entries less than or equal to the 3533 * maximum number. 3534 */ 3535 static void 3536 bridge_rttrim(struct bridge_softc *sc) 3537 { 3538 struct bridge_rtnode *brt, *nbrt; 3539 3540 NET_EPOCH_ASSERT(); 3541 BRIDGE_RT_LOCK_ASSERT(sc); 3542 3543 /* Make sure we actually need to do this. */ 3544 if (sc->sc_brtcnt <= sc->sc_brtmax) 3545 return; 3546 3547 /* Force an aging cycle; this might trim enough addresses. */ 3548 bridge_rtage(sc); 3549 if (sc->sc_brtcnt <= sc->sc_brtmax) 3550 return; 3551 3552 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 3553 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { 3554 bridge_rtnode_destroy(sc, brt); 3555 if (sc->sc_brtcnt <= sc->sc_brtmax) 3556 return; 3557 } 3558 } 3559 } 3560 3561 /* 3562 * bridge_timer: 3563 * 3564 * Aging timer for the bridge. 3565 */ 3566 static void 3567 bridge_timer(void *arg) 3568 { 3569 struct bridge_softc *sc = arg; 3570 3571 BRIDGE_RT_LOCK_ASSERT(sc); 3572 3573 /* Destruction of rtnodes requires a proper vnet context */ 3574 CURVNET_SET(sc->sc_ifp->if_vnet); 3575 bridge_rtage(sc); 3576 3577 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) 3578 callout_reset(&sc->sc_brcallout, 3579 bridge_rtable_prune_period * hz, bridge_timer, sc); 3580 CURVNET_RESTORE(); 3581 } 3582 3583 /* 3584 * bridge_rtage: 3585 * 3586 * Perform an aging cycle. 3587 */ 3588 static void 3589 bridge_rtage(struct bridge_softc *sc) 3590 { 3591 struct bridge_rtnode *brt, *nbrt; 3592 3593 BRIDGE_RT_LOCK_ASSERT(sc); 3594 3595 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 3596 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { 3597 if (time_uptime >= brt->brt_expire) 3598 bridge_rtnode_destroy(sc, brt); 3599 } 3600 } 3601 } 3602 3603 /* 3604 * bridge_rtflush: 3605 * 3606 * Remove all dynamic addresses from the bridge. 3607 */ 3608 static void 3609 bridge_rtflush(struct bridge_softc *sc, int full) 3610 { 3611 struct bridge_rtnode *brt, *nbrt; 3612 3613 BRIDGE_RT_LOCK_ASSERT(sc); 3614 3615 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 3616 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 3617 bridge_rtnode_destroy(sc, brt); 3618 } 3619 } 3620 3621 /* 3622 * bridge_rtdaddr: 3623 * 3624 * Remove an address from the table. 3625 */ 3626 static int 3627 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, 3628 ether_vlanid_t vlan) 3629 { 3630 struct bridge_rtnode *brt; 3631 int found = 0; 3632 3633 BRIDGE_RT_LOCK(sc); 3634 3635 /* 3636 * If vlan is DOT1Q_VID_RSVD_IMPL then we want to delete for all vlans 3637 * so the lookup may return more than one. 3638 */ 3639 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) { 3640 bridge_rtnode_destroy(sc, brt); 3641 found = 1; 3642 } 3643 3644 BRIDGE_RT_UNLOCK(sc); 3645 3646 return (found ? 0 : ENOENT); 3647 } 3648 3649 /* 3650 * bridge_rtdelete: 3651 * 3652 * Delete routes to a speicifc member interface. 3653 */ 3654 static void 3655 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full) 3656 { 3657 struct bridge_rtnode *brt, *nbrt; 3658 3659 BRIDGE_RT_LOCK_ASSERT(sc); 3660 3661 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { 3662 if (brt->brt_ifp == ifp && (full || 3663 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)) 3664 bridge_rtnode_destroy(sc, brt); 3665 } 3666 } 3667 3668 /* 3669 * bridge_rtable_init: 3670 * 3671 * Initialize the route table for this bridge. 3672 */ 3673 static void 3674 bridge_rtable_init(struct bridge_softc *sc) 3675 { 3676 int i; 3677 3678 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE, 3679 M_DEVBUF, M_WAITOK); 3680 3681 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++) 3682 CK_LIST_INIT(&sc->sc_rthash[i]); 3683 3684 sc->sc_rthash_key = arc4random(); 3685 CK_LIST_INIT(&sc->sc_rtlist); 3686 } 3687 3688 /* 3689 * bridge_rtable_fini: 3690 * 3691 * Deconstruct the route table for this bridge. 3692 */ 3693 static void 3694 bridge_rtable_fini(struct bridge_softc *sc) 3695 { 3696 3697 KASSERT(sc->sc_brtcnt == 0, 3698 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt)); 3699 free(sc->sc_rthash, M_DEVBUF); 3700 } 3701 3702 /* 3703 * The following hash function is adapted from "Hash Functions" by Bob Jenkins 3704 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997). 3705 */ 3706 #define mix(a, b, c) \ 3707 do { \ 3708 a -= b; a -= c; a ^= (c >> 13); \ 3709 b -= c; b -= a; b ^= (a << 8); \ 3710 c -= a; c -= b; c ^= (b >> 13); \ 3711 a -= b; a -= c; a ^= (c >> 12); \ 3712 b -= c; b -= a; b ^= (a << 16); \ 3713 c -= a; c -= b; c ^= (b >> 5); \ 3714 a -= b; a -= c; a ^= (c >> 3); \ 3715 b -= c; b -= a; b ^= (a << 10); \ 3716 c -= a; c -= b; c ^= (b >> 15); \ 3717 } while (/*CONSTCOND*/0) 3718 3719 static __inline uint32_t 3720 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr) 3721 { 3722 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key; 3723 3724 b += addr[5] << 8; 3725 b += addr[4]; 3726 a += addr[3] << 24; 3727 a += addr[2] << 16; 3728 a += addr[1] << 8; 3729 a += addr[0]; 3730 3731 mix(a, b, c); 3732 3733 return (c & BRIDGE_RTHASH_MASK); 3734 } 3735 3736 #undef mix 3737 3738 static int 3739 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b) 3740 { 3741 int i, d; 3742 3743 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) { 3744 d = ((int)a[i]) - ((int)b[i]); 3745 } 3746 3747 return (d); 3748 } 3749 3750 /* 3751 * bridge_rtnode_lookup: 3752 * 3753 * Look up a bridge route node for the specified destination. Compare the 3754 * vlan id or if zero then just return the first match. 3755 */ 3756 static struct bridge_rtnode * 3757 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, 3758 ether_vlanid_t vlan) 3759 { 3760 struct bridge_rtnode *brt; 3761 uint32_t hash; 3762 int dir; 3763 3764 BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc); 3765 3766 hash = bridge_rthash(sc, addr); 3767 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) { 3768 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr); 3769 if (dir == 0 && (brt->brt_vlan == vlan || vlan == DOT1Q_VID_RSVD_IMPL)) 3770 return (brt); 3771 if (dir > 0) 3772 return (NULL); 3773 } 3774 3775 return (NULL); 3776 } 3777 3778 /* 3779 * bridge_rtnode_insert: 3780 * 3781 * Insert the specified bridge node into the route table. We 3782 * assume the entry is not already in the table. 3783 */ 3784 static int 3785 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt) 3786 { 3787 struct bridge_rtnode *lbrt; 3788 uint32_t hash; 3789 int dir; 3790 3791 BRIDGE_RT_LOCK_ASSERT(sc); 3792 3793 hash = bridge_rthash(sc, brt->brt_addr); 3794 3795 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]); 3796 if (lbrt == NULL) { 3797 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash); 3798 goto out; 3799 } 3800 3801 do { 3802 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr); 3803 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan) 3804 return (EXTERROR(EEXIST, "Address already exists")); 3805 if (dir > 0) { 3806 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash); 3807 goto out; 3808 } 3809 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) { 3810 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash); 3811 goto out; 3812 } 3813 lbrt = CK_LIST_NEXT(lbrt, brt_hash); 3814 } while (lbrt != NULL); 3815 3816 #ifdef DIAGNOSTIC 3817 panic("bridge_rtnode_insert: impossible"); 3818 #endif 3819 3820 out: 3821 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list); 3822 sc->sc_brtcnt++; 3823 3824 return (0); 3825 } 3826 3827 static void 3828 bridge_rtnode_destroy_cb(struct epoch_context *ctx) 3829 { 3830 struct bridge_rtnode *brt; 3831 3832 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx); 3833 3834 CURVNET_SET(brt->brt_vnet); 3835 uma_zfree(V_bridge_rtnode_zone, brt); 3836 CURVNET_RESTORE(); 3837 } 3838 3839 /* 3840 * bridge_rtnode_destroy: 3841 * 3842 * Destroy a bridge rtnode. 3843 */ 3844 static void 3845 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt) 3846 { 3847 BRIDGE_RT_LOCK_ASSERT(sc); 3848 3849 CK_LIST_REMOVE(brt, brt_hash); 3850 3851 CK_LIST_REMOVE(brt, brt_list); 3852 sc->sc_brtcnt--; 3853 brt->brt_dst->bif_addrcnt--; 3854 3855 NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx); 3856 } 3857 3858 /* 3859 * bridge_rtable_expire: 3860 * 3861 * Set the expiry time for all routes on an interface. 3862 */ 3863 static void 3864 bridge_rtable_expire(struct ifnet *ifp, int age) 3865 { 3866 struct bridge_iflist *bif = NULL; 3867 struct bridge_softc *sc = NULL; 3868 struct bridge_rtnode *brt; 3869 3870 CURVNET_SET(ifp->if_vnet); 3871 3872 bif = ifp->if_bridge; 3873 if (bif) 3874 sc = bif->bif_sc; 3875 MPASS(sc != NULL); 3876 BRIDGE_RT_LOCK(sc); 3877 3878 /* 3879 * If the age is zero then flush, otherwise set all the expiry times to 3880 * age for the interface 3881 */ 3882 if (age == 0) 3883 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN); 3884 else { 3885 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) { 3886 /* Cap the expiry time to 'age' */ 3887 if (brt->brt_ifp == ifp && 3888 brt->brt_expire > time_uptime + age && 3889 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) 3890 brt->brt_expire = time_uptime + age; 3891 } 3892 } 3893 BRIDGE_RT_UNLOCK(sc); 3894 CURVNET_RESTORE(); 3895 } 3896 3897 /* 3898 * bridge_state_change: 3899 * 3900 * Callback from the bridgestp code when a port changes states. 3901 */ 3902 static void 3903 bridge_state_change(struct ifnet *ifp, int state) 3904 { 3905 struct bridge_iflist *bif = ifp->if_bridge; 3906 struct bridge_softc *sc = bif->bif_sc; 3907 static const char *stpstates[] = { 3908 "disabled", 3909 "listening", 3910 "learning", 3911 "forwarding", 3912 "blocking", 3913 "discarding" 3914 }; 3915 3916 CURVNET_SET(ifp->if_vnet); 3917 if (V_log_stp) 3918 log(LOG_NOTICE, "%s: state changed to %s on %s\n", 3919 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname); 3920 CURVNET_RESTORE(); 3921 } 3922 3923 /* 3924 * Send bridge packets through pfil if they are one of the types pfil can deal 3925 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without 3926 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for 3927 * that interface. 3928 */ 3929 static int 3930 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir) 3931 { 3932 int snap, error, i; 3933 struct ether_header *eh1, eh2; 3934 struct llc llc1; 3935 u_int16_t ether_type; 3936 pfil_return_t rv; 3937 #ifdef INET 3938 struct ip *ip = NULL; 3939 int hlen = 0; 3940 struct ifnet *errifp = (bifp != NULL) ? bifp : ifp; 3941 #endif 3942 3943 snap = 0; 3944 error = -1; /* Default error if not error == 0 */ 3945 3946 #if 0 3947 /* we may return with the IP fields swapped, ensure its not shared */ 3948 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__)); 3949 #endif 3950 3951 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0) 3952 return (0); /* filtering is disabled */ 3953 3954 if ((*mp = m_pullup(*mp, ETHER_HDR_LEN)) == NULL) { 3955 printf("%s: m_pullup failed\n", __func__); 3956 return (-1); 3957 } 3958 3959 eh1 = mtod(*mp, struct ether_header *); 3960 ether_type = ntohs(eh1->ether_type); 3961 3962 /* 3963 * Check for SNAP/LLC. 3964 */ 3965 if (ether_type < ETHERMTU) { 3966 struct llc *llc2; 3967 3968 i = min((*mp)->m_pkthdr.len, 3969 ETHER_HDR_LEN + sizeof(struct llc)); 3970 if ((*mp = m_pullup(*mp, i)) == NULL) { 3971 printf("%s: m_pullup failed\n", __func__); 3972 return (-1); 3973 } 3974 eh1 = mtod(*mp, struct ether_header *); 3975 3976 if ((*mp)->m_len >= ETHER_HDR_LEN + sizeof(struct llc)) { 3977 llc2 = (struct llc *)(eh1 + 1); 3978 if (llc2->llc_dsap == LLC_SNAP_LSAP && 3979 llc2->llc_ssap == LLC_SNAP_LSAP && 3980 llc2->llc_control == LLC_UI) { 3981 ether_type = 3982 htons(llc2->llc_un.type_snap.ether_type); 3983 snap = 1; 3984 } 3985 } 3986 } 3987 3988 /* 3989 * If we're trying to filter bridge traffic, only look at traffic for 3990 * protocols available in the kernel (IPv4 and/or IPv6) to avoid 3991 * passing traffic for an unsupported protocol to the filter. This is 3992 * lame since if we really wanted, say, an AppleTalk filter, we are 3993 * hosed, but of course we don't have an AppleTalk filter to begin 3994 * with. (Note that since pfil doesn't understand ARP it will pass 3995 * *ALL* ARP traffic.) 3996 */ 3997 switch (ether_type) { 3998 #ifdef INET 3999 case ETHERTYPE_ARP: 4000 case ETHERTYPE_REVARP: 4001 if (V_pfil_ipfw_arp == 0) 4002 return (0); /* Automatically pass */ 4003 4004 /* FALLTHROUGH */ 4005 case ETHERTYPE_IP: 4006 #endif 4007 #ifdef INET6 4008 case ETHERTYPE_IPV6: 4009 #endif /* INET6 */ 4010 break; 4011 4012 default: 4013 /* 4014 * We get here if the packet isn't from a supported 4015 * protocol. Check to see if the user wants to pass 4016 * non-IP packets, these will not be checked by pfil(9) 4017 * and passed unconditionally so the default is to 4018 * drop. 4019 */ 4020 if (V_pfil_onlyip) 4021 goto bad; 4022 } 4023 4024 /* Run the packet through pfil before stripping link headers */ 4025 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 && 4026 dir == PFIL_OUT && ifp != NULL) { 4027 switch (pfil_mbuf_out(V_link_pfil_head, mp, ifp, NULL)) { 4028 case PFIL_DROPPED: 4029 return (EACCES); 4030 case PFIL_CONSUMED: 4031 return (0); 4032 } 4033 } 4034 4035 /* Strip off the Ethernet header and keep a copy. */ 4036 m_copydata(*mp, 0, ETHER_HDR_LEN, &eh2); 4037 m_adj(*mp, ETHER_HDR_LEN); 4038 4039 /* Strip off snap header, if present */ 4040 if (snap) { 4041 m_copydata(*mp, 0, sizeof(struct llc), &llc1); 4042 m_adj(*mp, sizeof(struct llc)); 4043 } 4044 4045 /* 4046 * Check the IP header for alignment and errors 4047 */ 4048 if (dir == PFIL_IN) { 4049 switch (ether_type) { 4050 #ifdef INET 4051 case ETHERTYPE_IP: 4052 error = bridge_ip_checkbasic(mp); 4053 break; 4054 #endif 4055 #ifdef INET6 4056 case ETHERTYPE_IPV6: 4057 error = bridge_ip6_checkbasic(mp); 4058 break; 4059 #endif /* INET6 */ 4060 default: 4061 error = 0; 4062 } 4063 if (error) 4064 goto bad; 4065 } 4066 4067 error = 0; 4068 4069 /* 4070 * Run the packet through pfil 4071 */ 4072 rv = PFIL_PASS; 4073 switch (ether_type) { 4074 #ifdef INET 4075 case ETHERTYPE_IP: 4076 /* 4077 * Run pfil on the member interface and the bridge, both can 4078 * be skipped by clearing pfil_member or pfil_bridge. 4079 * 4080 * Keep the order: 4081 * in_if -> bridge_if -> out_if 4082 */ 4083 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv = 4084 pfil_mbuf_out(V_inet_pfil_head, mp, bifp, NULL)) != 4085 PFIL_PASS) 4086 break; 4087 4088 if (V_pfil_member && ifp != NULL) { 4089 rv = (dir == PFIL_OUT) ? 4090 pfil_mbuf_out(V_inet_pfil_head, mp, ifp, NULL) : 4091 pfil_mbuf_in(V_inet_pfil_head, mp, ifp, NULL); 4092 if (rv != PFIL_PASS) 4093 break; 4094 } 4095 4096 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv = 4097 pfil_mbuf_in(V_inet_pfil_head, mp, bifp, NULL)) != 4098 PFIL_PASS) 4099 break; 4100 4101 /* check if we need to fragment the packet */ 4102 /* bridge_fragment generates a mbuf chain of packets */ 4103 /* that already include eth headers */ 4104 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) { 4105 i = (*mp)->m_pkthdr.len; 4106 if (i > ifp->if_mtu) { 4107 error = bridge_fragment(ifp, mp, &eh2, snap, 4108 &llc1); 4109 if (error != 0) 4110 if_inc_counter(errifp, 4111 IFCOUNTER_OERRORS, 1); 4112 return (error); 4113 } 4114 } 4115 4116 /* Recalculate the ip checksum. */ 4117 ip = mtod(*mp, struct ip *); 4118 hlen = ip->ip_hl << 2; 4119 if (hlen < sizeof(struct ip)) 4120 goto bad; 4121 if ((*mp = m_pullup(*mp, hlen)) == NULL) 4122 goto bad; 4123 ip = mtod(*mp, struct ip *); 4124 ip->ip_sum = 0; 4125 if (hlen == sizeof(struct ip)) 4126 ip->ip_sum = in_cksum_hdr(ip); 4127 else 4128 ip->ip_sum = in_cksum(*mp, hlen); 4129 4130 break; 4131 #endif /* INET */ 4132 #ifdef INET6 4133 case ETHERTYPE_IPV6: 4134 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv = 4135 pfil_mbuf_out(V_inet6_pfil_head, mp, bifp, NULL)) != 4136 PFIL_PASS) 4137 break; 4138 4139 if (V_pfil_member && ifp != NULL) { 4140 rv = (dir == PFIL_OUT) ? 4141 pfil_mbuf_out(V_inet6_pfil_head, mp, ifp, NULL) : 4142 pfil_mbuf_in(V_inet6_pfil_head, mp, ifp, NULL); 4143 if (rv != PFIL_PASS) 4144 break; 4145 } 4146 4147 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv = 4148 pfil_mbuf_in(V_inet6_pfil_head, mp, bifp, NULL)) != 4149 PFIL_PASS) 4150 break; 4151 break; 4152 #endif 4153 } 4154 4155 switch (rv) { 4156 case PFIL_CONSUMED: 4157 return (0); 4158 case PFIL_DROPPED: 4159 return (EACCES); 4160 default: 4161 break; 4162 } 4163 4164 error = -1; 4165 4166 /* 4167 * Finally, put everything back the way it was and return 4168 */ 4169 if (snap) { 4170 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT); 4171 if (*mp == NULL) 4172 return (error); 4173 bcopy(&llc1, mtod(*mp, void *), sizeof(struct llc)); 4174 } 4175 4176 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT); 4177 if (*mp == NULL) 4178 return (error); 4179 bcopy(&eh2, mtod(*mp, void *), ETHER_HDR_LEN); 4180 4181 return (0); 4182 4183 bad: 4184 m_freem(*mp); 4185 *mp = NULL; 4186 return (error); 4187 } 4188 4189 #ifdef INET 4190 /* 4191 * Perform basic checks on header size since 4192 * pfil assumes ip_input has already processed 4193 * it for it. Cut-and-pasted from ip_input.c. 4194 * Given how simple the IPv6 version is, 4195 * does the IPv4 version really need to be 4196 * this complicated? 4197 * 4198 * XXX Should we update ipstat here, or not? 4199 * XXX Right now we update ipstat but not 4200 * XXX csum_counter. 4201 */ 4202 static int 4203 bridge_ip_checkbasic(struct mbuf **mp) 4204 { 4205 struct mbuf *m = *mp; 4206 struct ip *ip; 4207 int len, hlen; 4208 u_short sum; 4209 4210 if (*mp == NULL) 4211 return (-1); 4212 4213 if (IP_HDR_ALIGNED_P(mtod(m, char *)) == 0) { 4214 if ((m = m_copyup(m, sizeof(struct ip), 4215 (max_linkhdr + 3) & ~3)) == NULL) { 4216 /* XXXJRT new stat, please */ 4217 KMOD_IPSTAT_INC(ips_toosmall); 4218 goto bad; 4219 } 4220 } else if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 4221 KMOD_IPSTAT_INC(ips_toosmall); 4222 goto bad; 4223 } 4224 ip = mtod(m, struct ip *); 4225 if (ip == NULL) goto bad; 4226 4227 if (ip->ip_v != IPVERSION) { 4228 KMOD_IPSTAT_INC(ips_badvers); 4229 goto bad; 4230 } 4231 hlen = ip->ip_hl << 2; 4232 if (hlen < sizeof(struct ip)) { /* minimum header length */ 4233 KMOD_IPSTAT_INC(ips_badhlen); 4234 goto bad; 4235 } 4236 if ((m = m_pullup(m, hlen)) == NULL) { 4237 KMOD_IPSTAT_INC(ips_badhlen); 4238 goto bad; 4239 } 4240 ip = mtod(m, struct ip *); 4241 4242 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) { 4243 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID); 4244 } else { 4245 if (hlen == sizeof(struct ip)) { 4246 sum = in_cksum_hdr(ip); 4247 } else { 4248 sum = in_cksum(m, hlen); 4249 } 4250 } 4251 if (sum) { 4252 KMOD_IPSTAT_INC(ips_badsum); 4253 goto bad; 4254 } 4255 4256 /* Retrieve the packet length. */ 4257 len = ntohs(ip->ip_len); 4258 4259 /* 4260 * Check for additional length bogosity 4261 */ 4262 if (len < hlen) { 4263 KMOD_IPSTAT_INC(ips_badlen); 4264 goto bad; 4265 } 4266 4267 /* 4268 * Check that the amount of data in the buffers 4269 * is as at least much as the IP header would have us expect. 4270 * Drop packet if shorter than we expect. 4271 */ 4272 if (m->m_pkthdr.len < len) { 4273 KMOD_IPSTAT_INC(ips_tooshort); 4274 goto bad; 4275 } 4276 4277 /* Checks out, proceed */ 4278 *mp = m; 4279 return (0); 4280 4281 bad: 4282 *mp = m; 4283 return (-1); 4284 } 4285 #endif /* INET */ 4286 4287 #ifdef INET6 4288 /* 4289 * Same as above, but for IPv6. 4290 * Cut-and-pasted from ip6_input.c. 4291 * XXX Should we update ip6stat, or not? 4292 */ 4293 static int 4294 bridge_ip6_checkbasic(struct mbuf **mp) 4295 { 4296 struct mbuf *m = *mp; 4297 struct ifnet *inifp = m->m_pkthdr.rcvif; 4298 struct ip6_hdr *ip6; 4299 4300 /* 4301 * If the IPv6 header is not aligned, slurp it up into a new 4302 * mbuf with space for link headers, in the event we forward 4303 * it. Otherwise, if it is aligned, make sure the entire base 4304 * IPv6 header is in the first mbuf of the chain. 4305 */ 4306 if (IP6_HDR_ALIGNED_P(mtod(m, char *)) == 0) { 4307 if ((m = m_copyup(m, sizeof(struct ip6_hdr), 4308 (max_linkhdr + 3) & ~3)) == NULL) { 4309 /* XXXJRT new stat, please */ 4310 IP6STAT_INC(ip6s_toosmall); 4311 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 4312 goto bad; 4313 } 4314 } else if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 4315 IP6STAT_INC(ip6s_toosmall); 4316 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 4317 goto bad; 4318 } 4319 4320 ip6 = mtod(m, struct ip6_hdr *); 4321 4322 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 4323 IP6STAT_INC(ip6s_badvers); 4324 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 4325 goto bad; 4326 } 4327 4328 /* Checks out, proceed */ 4329 *mp = m; 4330 return (0); 4331 4332 bad: 4333 *mp = m; 4334 return (-1); 4335 } 4336 #endif /* INET6 */ 4337 4338 #ifdef INET 4339 /* 4340 * bridge_fragment: 4341 * 4342 * Fragment mbuf chain in multiple packets and prepend ethernet header. 4343 */ 4344 static int 4345 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh, 4346 int snap, struct llc *llc) 4347 { 4348 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL; 4349 struct ip *ip; 4350 int error = -1; 4351 4352 if ((m = m_pullup(m, sizeof(struct ip))) == NULL) { 4353 KMOD_IPSTAT_INC(ips_odropped); 4354 goto dropit; 4355 } 4356 ip = mtod(m, struct ip *); 4357 4358 m->m_pkthdr.csum_flags |= CSUM_IP; 4359 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist); 4360 if (error) 4361 goto dropit; 4362 4363 /* 4364 * Walk the chain and re-add the Ethernet header for 4365 * each mbuf packet. 4366 */ 4367 for (mcur = m; mcur; mcur = mcur->m_nextpkt) { 4368 nextpkt = mcur->m_nextpkt; 4369 mcur->m_nextpkt = NULL; 4370 if (snap) { 4371 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT); 4372 if (mcur == NULL) { 4373 error = ENOBUFS; 4374 KMOD_IPSTAT_INC(ips_odropped); 4375 if (mprev != NULL) 4376 mprev->m_nextpkt = nextpkt; 4377 goto dropit; 4378 } 4379 bcopy(llc, mtod(mcur, void *), sizeof(struct llc)); 4380 } 4381 4382 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT); 4383 if (mcur == NULL) { 4384 error = ENOBUFS; 4385 KMOD_IPSTAT_INC(ips_odropped); 4386 if (mprev != NULL) 4387 mprev->m_nextpkt = nextpkt; 4388 goto dropit; 4389 } 4390 bcopy(eh, mtod(mcur, void *), ETHER_HDR_LEN); 4391 4392 /* 4393 * The previous two M_PREPEND could have inserted one or two 4394 * mbufs in front so we have to update the previous packet's 4395 * m_nextpkt. 4396 */ 4397 mcur->m_nextpkt = nextpkt; 4398 if (mprev != NULL) 4399 mprev->m_nextpkt = mcur; 4400 else { 4401 /* The first mbuf in the original chain needs to be 4402 * updated. */ 4403 *mp = mcur; 4404 } 4405 mprev = mcur; 4406 } 4407 4408 KMOD_IPSTAT_INC(ips_fragmented); 4409 return (error); 4410 4411 dropit: 4412 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */ 4413 m = mcur->m_nextpkt; 4414 m_freem(mcur); 4415 } 4416 return (error); 4417 } 4418 #endif /* INET */ 4419 4420 static void 4421 bridge_linkstate(struct ifnet *ifp) 4422 { 4423 struct bridge_softc *sc = NULL; 4424 struct bridge_iflist *bif; 4425 struct epoch_tracker et; 4426 4427 NET_EPOCH_ENTER(et); 4428 4429 bif = ifp->if_bridge; 4430 if (bif) 4431 sc = bif->bif_sc; 4432 4433 if (sc != NULL) { 4434 bridge_linkcheck(sc); 4435 bstp_linkstate(&bif->bif_stp); 4436 } 4437 4438 NET_EPOCH_EXIT(et); 4439 } 4440 4441 static void 4442 bridge_linkcheck(struct bridge_softc *sc) 4443 { 4444 struct bridge_iflist *bif; 4445 int new_link, hasls; 4446 4447 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc); 4448 4449 new_link = LINK_STATE_DOWN; 4450 hasls = 0; 4451 /* Our link is considered up if at least one of our ports is active */ 4452 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { 4453 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE) 4454 hasls++; 4455 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) { 4456 new_link = LINK_STATE_UP; 4457 break; 4458 } 4459 } 4460 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) { 4461 /* If no interfaces support link-state then we default to up */ 4462 new_link = LINK_STATE_UP; 4463 } 4464 if_link_state_change(sc->sc_ifp, new_link); 4465 } 4466