1 /* 2 * Copyright 2001 Wasabi Systems, Inc. 3 * All rights reserved. 4 * 5 * Written by Jason R. Thorpe for Wasabi Systems, Inc. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. All advertising materials mentioning features or use of this software 16 * must display the following acknowledgement: 17 * This product includes software developed for the NetBSD Project by 18 * Wasabi Systems, Inc. 19 * 4. The name of Wasabi Systems, Inc. may not be used to endorse 20 * or promote products derived from this software without specific prior 21 * written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 25 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 26 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC 27 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 28 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 29 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 30 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 31 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 32 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 33 * POSSIBILITY OF SUCH DAMAGE. 34 */ 35 36 /* 37 * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net) 38 * All rights reserved. 39 * 40 * Redistribution and use in source and binary forms, with or without 41 * modification, are permitted provided that the following conditions 42 * are met: 43 * 1. Redistributions of source code must retain the above copyright 44 * notice, this list of conditions and the following disclaimer. 45 * 2. Redistributions in binary form must reproduce the above copyright 46 * notice, this list of conditions and the following disclaimer in the 47 * documentation and/or other materials provided with the distribution. 48 * 3. All advertising materials mentioning features or use of this software 49 * must display the following acknowledgement: 50 * This product includes software developed by Jason L. Wright 51 * 4. The name of the author may not be used to endorse or promote products 52 * derived from this software without specific prior written permission. 53 * 54 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 55 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED 56 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 57 * DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, 58 * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 59 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR 60 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, 62 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN 63 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 64 * POSSIBILITY OF SUCH DAMAGE. 65 * 66 * $OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp $ 67 * $NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $ 68 * $FreeBSD: src/sys/net/if_bridge.c,v 1.26 2005/10/13 23:05:55 thompsa Exp $ 69 */ 70 71 /* 72 * Network interface bridge support. 73 * 74 * TODO: 75 * 76 * - Currently only supports Ethernet-like interfaces (Ethernet, 77 * 802.11, VLANs on Ethernet, etc.) Figure out a nice way 78 * to bridge other types of interfaces (FDDI-FDDI, and maybe 79 * consider heterogenous bridges). 80 * 81 * 82 * Bridge's route information is duplicated to each CPUs: 83 * 84 * CPU0 CPU1 CPU2 CPU3 85 * +-----------+ +-----------+ +-----------+ +-----------+ 86 * | rtnode | | rtnode | | rtnode | | rtnode | 87 * | | | | | | | | 88 * | dst eaddr | | dst eaddr | | dst eaddr | | dst eaddr | 89 * +-----------+ +-----------+ +-----------+ +-----------+ 90 * | | | | 91 * | | | | 92 * | | +----------+ | | 93 * | | | rtinfo | | | 94 * | +---->| |<---+ | 95 * | | flags | | 96 * +-------------->| timeout |<-------------+ 97 * | dst_ifp | 98 * +----------+ 99 * 100 * We choose to put timeout and dst_ifp into shared part, so updating 101 * them will be cheaper than using message forwarding. Also there is 102 * not need to use spinlock to protect the updating: timeout and dst_ifp 103 * is not related and specific field's updating order has no importance. 104 * The cache pollution by the share part should not be heavy: in a stable 105 * setup, dst_ifp probably will be not changed in rtnode's life time, 106 * while timeout is refreshed once per second; most of the time, timeout 107 * and dst_ifp are read-only accessed. 108 * 109 * 110 * Bridge route information installation on bridge_input path: 111 * 112 * CPU0 CPU1 CPU2 CPU3 113 * 114 * tcp_thread2 115 * | 116 * alloc nmsg 117 * snd nmsg | 118 * w/o rtinfo | 119 * ifnet0<-----------------------+ 120 * | : 121 * lookup dst : 122 * rtnode exists?(Y)free nmsg : 123 * |(N) : 124 * | 125 * alloc rtinfo 126 * alloc rtnode 127 * install rtnode 128 * | 129 * +---------->ifnet1 130 * : fwd nmsg | 131 * : w/ rtinfo | 132 * : | 133 * : | 134 * alloc rtnode 135 * (w/ nmsg's rtinfo) 136 * install rtnode 137 * | 138 * +---------->ifnet2 139 * : fwd nmsg | 140 * : w/ rtinfo | 141 * : | 142 * : same as ifnet1 143 * | 144 * +---------->ifnet3 145 * : fwd nmsg | 146 * : w/ rtinfo | 147 * : | 148 * : same as ifnet1 149 * free nmsg 150 * : 151 * : 152 * 153 * The netmsgs forwarded between protocol threads and ifnet threads are 154 * allocated with (M_WAITOK|M_NULLOK), so it will not fail under most 155 * cases (route information is too precious to be not installed :). 156 * Since multiple threads may try to install route information for the 157 * same dst eaddr, we look up route information in ifnet0. However, this 158 * looking up only need to be performed on ifnet0, which is the start 159 * point of the route information installation process. 160 * 161 * 162 * Bridge route information deleting/flushing: 163 * 164 * CPU0 CPU1 CPU2 CPU3 165 * 166 * netisr0 167 * | 168 * find suitable rtnodes, 169 * mark their rtinfo dead 170 * | 171 * | domsg <------------------------------------------+ 172 * | | replymsg 173 * | | 174 * V fwdmsg fwdmsg fwdmsg | 175 * ifnet0 --------> ifnet1 --------> ifnet2 --------> ifnet3 176 * delete rtnodes delete rtnodes delete rtnodes delete rtnodes 177 * w/ dead rtinfo w/ dead rtinfo w/ dead rtinfo w/ dead rtinfo 178 * free dead rtinfos 179 * 180 * All deleting/flushing operations are serialized by netisr0, so each 181 * operation only reaps the route information marked dead by itself. 182 * 183 * 184 * Bridge route information adding/deleting/flushing: 185 * Since all operation is serialized by the fixed message flow between 186 * ifnet threads, it is not possible to create corrupted per-cpu route 187 * information. 188 * 189 * 190 * 191 * Percpu member interface list iteration with blocking operation: 192 * Since one bridge could only delete one member interface at a time and 193 * the deleted member interface is not freed after netmsg_service_sync(), 194 * following way is used to make sure that even if the certain member 195 * interface is ripped from the percpu list during the blocking operation, 196 * the iteration still could keep going: 197 * 198 * TAILQ_FOREACH_MUTABLE(bif, sc->sc_iflists[mycpuid], bif_next, nbif) { 199 * blocking operation; 200 * blocking operation; 201 * ... 202 * ... 203 * if (nbif != NULL && !nbif->bif_onlist) { 204 * KKASSERT(bif->bif_onlist); 205 * nbif = TAILQ_NEXT(bif, bif_next); 206 * } 207 * } 208 * 209 * As mentioned above only one member interface could be unlinked from the 210 * percpu member interface list, so either bif or nbif may be not on the list, 211 * but _not_ both. To keep the list iteration, we don't care about bif, but 212 * only nbif. Since removed member interface will only be freed after we 213 * finish our work, it is safe to access any field in an unlinked bif (here 214 * bif_onlist). If nbif is no longer on the list, then bif must be on the 215 * list, so we change nbif to the next element of bif and keep going. 216 */ 217 218 #include "opt_inet.h" 219 #include "opt_inet6.h" 220 221 #include <sys/param.h> 222 #include <sys/mbuf.h> 223 #include <sys/malloc.h> 224 #include <sys/protosw.h> 225 #include <sys/systm.h> 226 #include <sys/time.h> 227 #include <sys/socket.h> /* for net/if.h */ 228 #include <sys/sockio.h> 229 #include <sys/ctype.h> /* string functions */ 230 #include <sys/kernel.h> 231 #include <sys/random.h> 232 #include <sys/sysctl.h> 233 #include <sys/module.h> 234 #include <sys/proc.h> 235 #include <sys/priv.h> 236 #include <sys/lock.h> 237 #include <sys/thread.h> 238 #include <sys/thread2.h> 239 #include <sys/mpipe.h> 240 241 #include <net/bpf.h> 242 #include <net/if.h> 243 #include <net/if_dl.h> 244 #include <net/if_types.h> 245 #include <net/if_var.h> 246 #include <net/pfil.h> 247 #include <net/ifq_var.h> 248 #include <net/if_clone.h> 249 250 #include <netinet/in.h> /* for struct arpcom */ 251 #include <netinet/in_systm.h> 252 #include <netinet/in_var.h> 253 #include <netinet/ip.h> 254 #include <netinet/ip_var.h> 255 #ifdef INET6 256 #include <netinet/ip6.h> 257 #include <netinet6/ip6_var.h> 258 #endif 259 #include <netinet/if_ether.h> /* for struct arpcom */ 260 #include <net/bridge/if_bridgevar.h> 261 #include <net/if_llc.h> 262 #include <net/netmsg2.h> 263 #include <net/netisr2.h> 264 265 #include <net/route.h> 266 #include <sys/in_cksum.h> 267 268 /* 269 * Size of the route hash table. Must be a power of two. 270 */ 271 #ifndef BRIDGE_RTHASH_SIZE 272 #define BRIDGE_RTHASH_SIZE 1024 273 #endif 274 275 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1) 276 277 /* 278 * Maximum number of addresses to cache. 279 */ 280 #ifndef BRIDGE_RTABLE_MAX 281 #define BRIDGE_RTABLE_MAX 100 282 #endif 283 284 /* 285 * Spanning tree defaults. 286 */ 287 #define BSTP_DEFAULT_MAX_AGE (20 * 256) 288 #define BSTP_DEFAULT_HELLO_TIME (2 * 256) 289 #define BSTP_DEFAULT_FORWARD_DELAY (15 * 256) 290 #define BSTP_DEFAULT_HOLD_TIME (1 * 256) 291 #define BSTP_DEFAULT_BRIDGE_PRIORITY 0x8000 292 #define BSTP_DEFAULT_PORT_PRIORITY 0x80 293 #define BSTP_DEFAULT_PATH_COST 55 294 295 /* 296 * Timeout (in seconds) for entries learned dynamically. 297 */ 298 #ifndef BRIDGE_RTABLE_TIMEOUT 299 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */ 300 #endif 301 302 /* 303 * Number of seconds between walks of the route list. 304 */ 305 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD 306 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60) 307 #endif 308 309 /* 310 * List of capabilities to mask on the member interface. 311 */ 312 #define BRIDGE_IFCAPS_MASK (IFCAP_TXCSUM | IFCAP_TSO) 313 314 typedef int (*bridge_ctl_t)(struct bridge_softc *, void *); 315 316 struct netmsg_brctl { 317 struct netmsg_base base; 318 bridge_ctl_t bc_func; 319 struct bridge_softc *bc_sc; 320 void *bc_arg; 321 }; 322 323 struct netmsg_brsaddr { 324 struct netmsg_base base; 325 struct bridge_softc *br_softc; 326 struct ifnet *br_dst_if; 327 struct bridge_rtinfo *br_rtinfo; 328 int br_setflags; 329 uint8_t br_dst[ETHER_ADDR_LEN]; 330 uint8_t br_flags; 331 }; 332 333 struct netmsg_braddbif { 334 struct netmsg_base base; 335 struct bridge_softc *br_softc; 336 struct bridge_ifinfo *br_bif_info; 337 struct ifnet *br_bif_ifp; 338 }; 339 340 struct netmsg_brdelbif { 341 struct netmsg_base base; 342 struct bridge_softc *br_softc; 343 struct bridge_ifinfo *br_bif_info; 344 struct bridge_iflist_head *br_bif_list; 345 }; 346 347 struct netmsg_brsflags { 348 struct netmsg_base base; 349 struct bridge_softc *br_softc; 350 struct bridge_ifinfo *br_bif_info; 351 uint32_t br_bif_flags; 352 }; 353 354 eventhandler_tag bridge_detach_cookie = NULL; 355 356 extern struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *); 357 extern int (*bridge_output_p)(struct ifnet *, struct mbuf *); 358 extern void (*bridge_dn_p)(struct mbuf *, struct ifnet *); 359 extern struct ifnet *(*bridge_interface_p)(void *if_bridge); 360 361 static int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD; 362 363 static int bridge_clone_create(struct if_clone *, int, caddr_t); 364 static int bridge_clone_destroy(struct ifnet *); 365 366 static int bridge_ioctl(struct ifnet *, u_long, caddr_t, struct ucred *); 367 static void bridge_mutecaps(struct bridge_ifinfo *, struct ifnet *, int); 368 static void bridge_ifdetach(void *, struct ifnet *); 369 static void bridge_init(void *); 370 static int bridge_from_us(struct bridge_softc *, struct ether_header *); 371 static void bridge_stop(struct ifnet *); 372 static void bridge_start(struct ifnet *, struct ifaltq_subque *); 373 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *); 374 static int bridge_output(struct ifnet *, struct mbuf *); 375 static struct ifnet *bridge_interface(void *if_bridge); 376 377 static void bridge_forward(struct bridge_softc *, struct mbuf *m); 378 379 static void bridge_timer_handler(netmsg_t); 380 static void bridge_timer(void *); 381 382 static void bridge_start_bcast(struct bridge_softc *, struct mbuf *); 383 static void bridge_broadcast(struct bridge_softc *, struct ifnet *, 384 struct mbuf *); 385 static void bridge_span(struct bridge_softc *, struct mbuf *); 386 387 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *, 388 struct ifnet *, uint8_t); 389 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *); 390 static void bridge_rtreap(struct bridge_softc *); 391 static void bridge_rtreap_async(struct bridge_softc *); 392 static void bridge_rttrim(struct bridge_softc *); 393 static int bridge_rtage_finddead(struct bridge_softc *); 394 static void bridge_rtage(struct bridge_softc *); 395 static void bridge_rtflush(struct bridge_softc *, int); 396 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *); 397 static int bridge_rtsaddr(struct bridge_softc *, const uint8_t *, 398 struct ifnet *, uint8_t); 399 static void bridge_rtmsg_sync(struct bridge_softc *sc); 400 static void bridge_rtreap_handler(netmsg_t); 401 static void bridge_rtinstall_handler(netmsg_t); 402 static int bridge_rtinstall_oncpu(struct bridge_softc *, const uint8_t *, 403 struct ifnet *, int, uint8_t, struct bridge_rtinfo **); 404 405 static void bridge_rtable_init(struct bridge_softc *); 406 static void bridge_rtable_fini(struct bridge_softc *); 407 408 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *); 409 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *, 410 const uint8_t *); 411 static void bridge_rtnode_insert(struct bridge_softc *, 412 struct bridge_rtnode *); 413 static void bridge_rtnode_destroy(struct bridge_softc *, 414 struct bridge_rtnode *); 415 416 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *, 417 const char *name); 418 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *, 419 struct ifnet *ifp); 420 static struct bridge_iflist *bridge_lookup_member_ifinfo(struct bridge_softc *, 421 struct bridge_ifinfo *); 422 static void bridge_delete_member(struct bridge_softc *, 423 struct bridge_iflist *, int); 424 static void bridge_delete_span(struct bridge_softc *, 425 struct bridge_iflist *); 426 427 static int bridge_control(struct bridge_softc *, u_long, 428 bridge_ctl_t, void *); 429 static int bridge_ioctl_init(struct bridge_softc *, void *); 430 static int bridge_ioctl_stop(struct bridge_softc *, void *); 431 static int bridge_ioctl_add(struct bridge_softc *, void *); 432 static int bridge_ioctl_del(struct bridge_softc *, void *); 433 static void bridge_ioctl_fillflags(struct bridge_softc *sc, 434 struct bridge_iflist *bif, struct ifbreq *req); 435 static int bridge_ioctl_gifflags(struct bridge_softc *, void *); 436 static int bridge_ioctl_sifflags(struct bridge_softc *, void *); 437 static int bridge_ioctl_scache(struct bridge_softc *, void *); 438 static int bridge_ioctl_gcache(struct bridge_softc *, void *); 439 static int bridge_ioctl_gifs(struct bridge_softc *, void *); 440 static int bridge_ioctl_rts(struct bridge_softc *, void *); 441 static int bridge_ioctl_saddr(struct bridge_softc *, void *); 442 static int bridge_ioctl_sto(struct bridge_softc *, void *); 443 static int bridge_ioctl_gto(struct bridge_softc *, void *); 444 static int bridge_ioctl_daddr(struct bridge_softc *, void *); 445 static int bridge_ioctl_flush(struct bridge_softc *, void *); 446 static int bridge_ioctl_gpri(struct bridge_softc *, void *); 447 static int bridge_ioctl_spri(struct bridge_softc *, void *); 448 static int bridge_ioctl_reinit(struct bridge_softc *, void *); 449 static int bridge_ioctl_ght(struct bridge_softc *, void *); 450 static int bridge_ioctl_sht(struct bridge_softc *, void *); 451 static int bridge_ioctl_gfd(struct bridge_softc *, void *); 452 static int bridge_ioctl_sfd(struct bridge_softc *, void *); 453 static int bridge_ioctl_gma(struct bridge_softc *, void *); 454 static int bridge_ioctl_sma(struct bridge_softc *, void *); 455 static int bridge_ioctl_sifprio(struct bridge_softc *, void *); 456 static int bridge_ioctl_sifcost(struct bridge_softc *, void *); 457 static int bridge_ioctl_addspan(struct bridge_softc *, void *); 458 static int bridge_ioctl_delspan(struct bridge_softc *, void *); 459 static int bridge_ioctl_sifbondwght(struct bridge_softc *, void *); 460 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *, 461 int); 462 static int bridge_ip_checkbasic(struct mbuf **mp); 463 #ifdef INET6 464 static int bridge_ip6_checkbasic(struct mbuf **mp); 465 #endif /* INET6 */ 466 static int bridge_fragment(struct ifnet *, struct mbuf *, 467 struct ether_header *, int, struct llc *); 468 static void bridge_enqueue_handler(netmsg_t); 469 static void bridge_handoff(struct bridge_softc *, struct ifnet *, 470 struct mbuf *, int); 471 472 static void bridge_del_bif_handler(netmsg_t); 473 static void bridge_add_bif_handler(netmsg_t); 474 static void bridge_del_bif(struct bridge_softc *, struct bridge_ifinfo *, 475 struct bridge_iflist_head *); 476 static void bridge_add_bif(struct bridge_softc *, struct bridge_ifinfo *, 477 struct ifnet *); 478 479 SYSCTL_DECL(_net_link); 480 SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge"); 481 482 static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */ 483 static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */ 484 static int pfil_member = 1; /* run pfil hooks on the member interface */ 485 static int bridge_debug; 486 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW, 487 &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled"); 488 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW, 489 &pfil_bridge, 0, "Packet filter on the bridge interface"); 490 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW, 491 &pfil_member, 0, "Packet filter on the member interface"); 492 SYSCTL_INT(_net_link_bridge, OID_AUTO, debug, CTLFLAG_RW, 493 &bridge_debug, 0, "Bridge debug mode"); 494 495 struct bridge_control_arg { 496 union { 497 struct ifbreq ifbreq; 498 struct ifbifconf ifbifconf; 499 struct ifbareq ifbareq; 500 struct ifbaconf ifbaconf; 501 struct ifbrparam ifbrparam; 502 } bca_u; 503 int bca_len; 504 void *bca_uptr; 505 void *bca_kptr; 506 }; 507 508 struct bridge_control { 509 bridge_ctl_t bc_func; 510 int bc_argsize; 511 int bc_flags; 512 }; 513 514 #define BC_F_COPYIN 0x01 /* copy arguments in */ 515 #define BC_F_COPYOUT 0x02 /* copy arguments out */ 516 #define BC_F_SUSER 0x04 /* do super-user check */ 517 518 const struct bridge_control bridge_control_table[] = { 519 { bridge_ioctl_add, sizeof(struct ifbreq), 520 BC_F_COPYIN|BC_F_SUSER }, 521 { bridge_ioctl_del, sizeof(struct ifbreq), 522 BC_F_COPYIN|BC_F_SUSER }, 523 524 { bridge_ioctl_gifflags, sizeof(struct ifbreq), 525 BC_F_COPYIN|BC_F_COPYOUT }, 526 { bridge_ioctl_sifflags, sizeof(struct ifbreq), 527 BC_F_COPYIN|BC_F_SUSER }, 528 529 { bridge_ioctl_scache, sizeof(struct ifbrparam), 530 BC_F_COPYIN|BC_F_SUSER }, 531 { bridge_ioctl_gcache, sizeof(struct ifbrparam), 532 BC_F_COPYOUT }, 533 534 { bridge_ioctl_gifs, sizeof(struct ifbifconf), 535 BC_F_COPYIN|BC_F_COPYOUT }, 536 { bridge_ioctl_rts, sizeof(struct ifbaconf), 537 BC_F_COPYIN|BC_F_COPYOUT }, 538 539 { bridge_ioctl_saddr, sizeof(struct ifbareq), 540 BC_F_COPYIN|BC_F_SUSER }, 541 542 { bridge_ioctl_sto, sizeof(struct ifbrparam), 543 BC_F_COPYIN|BC_F_SUSER }, 544 { bridge_ioctl_gto, sizeof(struct ifbrparam), 545 BC_F_COPYOUT }, 546 547 { bridge_ioctl_daddr, sizeof(struct ifbareq), 548 BC_F_COPYIN|BC_F_SUSER }, 549 550 { bridge_ioctl_flush, sizeof(struct ifbreq), 551 BC_F_COPYIN|BC_F_SUSER }, 552 553 { bridge_ioctl_gpri, sizeof(struct ifbrparam), 554 BC_F_COPYOUT }, 555 { bridge_ioctl_spri, sizeof(struct ifbrparam), 556 BC_F_COPYIN|BC_F_SUSER }, 557 558 { bridge_ioctl_ght, sizeof(struct ifbrparam), 559 BC_F_COPYOUT }, 560 { bridge_ioctl_sht, sizeof(struct ifbrparam), 561 BC_F_COPYIN|BC_F_SUSER }, 562 563 { bridge_ioctl_gfd, sizeof(struct ifbrparam), 564 BC_F_COPYOUT }, 565 { bridge_ioctl_sfd, sizeof(struct ifbrparam), 566 BC_F_COPYIN|BC_F_SUSER }, 567 568 { bridge_ioctl_gma, sizeof(struct ifbrparam), 569 BC_F_COPYOUT }, 570 { bridge_ioctl_sma, sizeof(struct ifbrparam), 571 BC_F_COPYIN|BC_F_SUSER }, 572 573 { bridge_ioctl_sifprio, sizeof(struct ifbreq), 574 BC_F_COPYIN|BC_F_SUSER }, 575 576 { bridge_ioctl_sifcost, sizeof(struct ifbreq), 577 BC_F_COPYIN|BC_F_SUSER }, 578 579 { bridge_ioctl_addspan, sizeof(struct ifbreq), 580 BC_F_COPYIN|BC_F_SUSER }, 581 { bridge_ioctl_delspan, sizeof(struct ifbreq), 582 BC_F_COPYIN|BC_F_SUSER }, 583 584 { bridge_ioctl_sifbondwght, sizeof(struct ifbreq), 585 BC_F_COPYIN|BC_F_SUSER }, 586 587 }; 588 static const int bridge_control_table_size = NELEM(bridge_control_table); 589 590 LIST_HEAD(, bridge_softc) bridge_list; 591 592 struct if_clone bridge_cloner = IF_CLONE_INITIALIZER("bridge", 593 bridge_clone_create, 594 bridge_clone_destroy, 0, IF_MAXUNIT); 595 596 static int 597 bridge_modevent(module_t mod, int type, void *data) 598 { 599 switch (type) { 600 case MOD_LOAD: 601 LIST_INIT(&bridge_list); 602 if_clone_attach(&bridge_cloner); 603 bridge_input_p = bridge_input; 604 bridge_output_p = bridge_output; 605 bridge_interface_p = bridge_interface; 606 bridge_detach_cookie = EVENTHANDLER_REGISTER( 607 ifnet_detach_event, bridge_ifdetach, NULL, 608 EVENTHANDLER_PRI_ANY); 609 #if 0 /* notyet */ 610 bstp_linkstate_p = bstp_linkstate; 611 #endif 612 break; 613 case MOD_UNLOAD: 614 if (!LIST_EMPTY(&bridge_list)) 615 return (EBUSY); 616 EVENTHANDLER_DEREGISTER(ifnet_detach_event, 617 bridge_detach_cookie); 618 if_clone_detach(&bridge_cloner); 619 bridge_input_p = NULL; 620 bridge_output_p = NULL; 621 bridge_interface_p = NULL; 622 #if 0 /* notyet */ 623 bstp_linkstate_p = NULL; 624 #endif 625 break; 626 default: 627 return (EOPNOTSUPP); 628 } 629 return (0); 630 } 631 632 static moduledata_t bridge_mod = { 633 "if_bridge", 634 bridge_modevent, 635 0 636 }; 637 638 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY); 639 640 641 /* 642 * bridge_clone_create: 643 * 644 * Create a new bridge instance. 645 */ 646 static int 647 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t param __unused) 648 { 649 struct bridge_softc *sc; 650 struct ifnet *ifp; 651 u_char eaddr[6]; 652 int cpu, rnd; 653 654 sc = kmalloc(sizeof(*sc), M_DEVBUF, M_WAITOK | M_ZERO); 655 ifp = sc->sc_ifp = &sc->sc_if; 656 657 sc->sc_brtmax = BRIDGE_RTABLE_MAX; 658 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT; 659 sc->sc_bridge_max_age = BSTP_DEFAULT_MAX_AGE; 660 sc->sc_bridge_hello_time = BSTP_DEFAULT_HELLO_TIME; 661 sc->sc_bridge_forward_delay = BSTP_DEFAULT_FORWARD_DELAY; 662 sc->sc_bridge_priority = BSTP_DEFAULT_BRIDGE_PRIORITY; 663 sc->sc_hold_time = BSTP_DEFAULT_HOLD_TIME; 664 665 /* Initialize our routing table. */ 666 bridge_rtable_init(sc); 667 668 callout_init(&sc->sc_brcallout); 669 netmsg_init(&sc->sc_brtimemsg, NULL, &netisr_adone_rport, 670 MSGF_DROPABLE, bridge_timer_handler); 671 sc->sc_brtimemsg.lmsg.u.ms_resultp = sc; 672 673 callout_init(&sc->sc_bstpcallout); 674 netmsg_init(&sc->sc_bstptimemsg, NULL, &netisr_adone_rport, 675 MSGF_DROPABLE, bstp_tick_handler); 676 sc->sc_bstptimemsg.lmsg.u.ms_resultp = sc; 677 678 /* Initialize per-cpu member iface lists */ 679 sc->sc_iflists = kmalloc(sizeof(*sc->sc_iflists) * ncpus, 680 M_DEVBUF, M_WAITOK); 681 for (cpu = 0; cpu < ncpus; ++cpu) 682 TAILQ_INIT(&sc->sc_iflists[cpu]); 683 684 TAILQ_INIT(&sc->sc_spanlist); 685 686 ifp->if_softc = sc; 687 if_initname(ifp, ifc->ifc_name, unit); 688 ifp->if_mtu = ETHERMTU; 689 ifp->if_flags = IFF_BROADCAST | IFF_MULTICAST; 690 ifp->if_ioctl = bridge_ioctl; 691 ifp->if_start = bridge_start; 692 ifp->if_init = bridge_init; 693 ifp->if_type = IFT_ETHER; 694 ifq_set_maxlen(&ifp->if_snd, ifqmaxlen); 695 ifq_set_ready(&ifp->if_snd); 696 ifp->if_hdrlen = ETHER_HDR_LEN; 697 698 /* 699 * Generate a random ethernet address and use the private AC:DE:48 700 * OUI code. 701 */ 702 rnd = karc4random(); 703 bcopy(&rnd, &eaddr[0], 4); /* ETHER_ADDR_LEN == 6 */ 704 rnd = karc4random(); 705 bcopy(&rnd, &eaddr[2], 4); /* ETHER_ADDR_LEN == 6 */ 706 707 eaddr[0] &= ~1; /* clear multicast bit */ 708 eaddr[0] |= 2; /* set the LAA bit */ 709 710 ether_ifattach(ifp, eaddr, NULL); 711 /* Now undo some of the damage... */ 712 ifp->if_baudrate = 0; 713 /*ifp->if_type = IFT_BRIDGE;*/ 714 715 crit_enter(); /* XXX MP */ 716 LIST_INSERT_HEAD(&bridge_list, sc, sc_list); 717 crit_exit(); 718 719 return (0); 720 } 721 722 static void 723 bridge_delete_dispatch(netmsg_t msg) 724 { 725 struct bridge_softc *sc = msg->lmsg.u.ms_resultp; 726 struct ifnet *bifp = sc->sc_ifp; 727 struct bridge_iflist *bif; 728 729 ifnet_serialize_all(bifp); 730 731 while ((bif = TAILQ_FIRST(&sc->sc_iflists[mycpuid])) != NULL) 732 bridge_delete_member(sc, bif, 0); 733 734 while ((bif = TAILQ_FIRST(&sc->sc_spanlist)) != NULL) 735 bridge_delete_span(sc, bif); 736 737 ifnet_deserialize_all(bifp); 738 739 lwkt_replymsg(&msg->lmsg, 0); 740 } 741 742 /* 743 * bridge_clone_destroy: 744 * 745 * Destroy a bridge instance. 746 */ 747 static int 748 bridge_clone_destroy(struct ifnet *ifp) 749 { 750 struct bridge_softc *sc = ifp->if_softc; 751 struct netmsg_base msg; 752 753 ifnet_serialize_all(ifp); 754 755 bridge_stop(ifp); 756 ifp->if_flags &= ~IFF_UP; 757 758 ifnet_deserialize_all(ifp); 759 760 netmsg_init(&msg, NULL, &curthread->td_msgport, 761 0, bridge_delete_dispatch); 762 msg.lmsg.u.ms_resultp = sc; 763 lwkt_domsg(BRIDGE_CFGPORT, &msg.lmsg, 0); 764 765 crit_enter(); /* XXX MP */ 766 LIST_REMOVE(sc, sc_list); 767 crit_exit(); 768 769 ether_ifdetach(ifp); 770 771 /* Tear down the routing table. */ 772 bridge_rtable_fini(sc); 773 774 /* Free per-cpu member iface lists */ 775 kfree(sc->sc_iflists, M_DEVBUF); 776 777 kfree(sc, M_DEVBUF); 778 779 return 0; 780 } 781 782 /* 783 * bridge_ioctl: 784 * 785 * Handle a control request from the operator. 786 */ 787 static int 788 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data, struct ucred *cr) 789 { 790 struct bridge_softc *sc = ifp->if_softc; 791 struct bridge_control_arg args; 792 struct ifdrv *ifd = (struct ifdrv *) data; 793 const struct bridge_control *bc; 794 int error = 0; 795 796 ASSERT_IFNET_SERIALIZED_ALL(ifp); 797 798 switch (cmd) { 799 case SIOCADDMULTI: 800 case SIOCDELMULTI: 801 break; 802 803 case SIOCGDRVSPEC: 804 case SIOCSDRVSPEC: 805 if (ifd->ifd_cmd >= bridge_control_table_size) { 806 error = EINVAL; 807 break; 808 } 809 bc = &bridge_control_table[ifd->ifd_cmd]; 810 811 if (cmd == SIOCGDRVSPEC && 812 (bc->bc_flags & BC_F_COPYOUT) == 0) { 813 error = EINVAL; 814 break; 815 } else if (cmd == SIOCSDRVSPEC && 816 (bc->bc_flags & BC_F_COPYOUT)) { 817 error = EINVAL; 818 break; 819 } 820 821 if (bc->bc_flags & BC_F_SUSER) { 822 error = priv_check_cred(cr, PRIV_ROOT, NULL_CRED_OKAY); 823 if (error) 824 break; 825 } 826 827 if (ifd->ifd_len != bc->bc_argsize || 828 ifd->ifd_len > sizeof(args.bca_u)) { 829 error = EINVAL; 830 break; 831 } 832 833 memset(&args, 0, sizeof(args)); 834 if (bc->bc_flags & BC_F_COPYIN) { 835 error = copyin(ifd->ifd_data, &args.bca_u, 836 ifd->ifd_len); 837 if (error) 838 break; 839 } 840 841 error = bridge_control(sc, cmd, bc->bc_func, &args); 842 if (error) { 843 KKASSERT(args.bca_len == 0 && args.bca_kptr == NULL); 844 break; 845 } 846 847 if (bc->bc_flags & BC_F_COPYOUT) { 848 error = copyout(&args, ifd->ifd_data, ifd->ifd_len); 849 if (args.bca_len != 0) { 850 KKASSERT(args.bca_kptr != NULL); 851 if (!error) { 852 error = copyout(args.bca_kptr, 853 args.bca_uptr, args.bca_len); 854 } 855 kfree(args.bca_kptr, M_TEMP); 856 } else { 857 KKASSERT(args.bca_kptr == NULL); 858 } 859 } else { 860 KKASSERT(args.bca_len == 0 && args.bca_kptr == NULL); 861 } 862 break; 863 864 case SIOCSIFFLAGS: 865 if (!(ifp->if_flags & IFF_UP) && 866 (ifp->if_flags & IFF_RUNNING)) { 867 /* 868 * If interface is marked down and it is running, 869 * then stop it. 870 */ 871 bridge_stop(ifp); 872 } else if ((ifp->if_flags & IFF_UP) && 873 !(ifp->if_flags & IFF_RUNNING)) { 874 /* 875 * If interface is marked up and it is stopped, then 876 * start it. 877 */ 878 ifp->if_init(sc); 879 } 880 881 /* 882 * If running and link flag state change we have to 883 * reinitialize as well. 884 */ 885 if ((ifp->if_flags & IFF_RUNNING) && 886 (ifp->if_flags & (IFF_LINK0|IFF_LINK1|IFF_LINK2)) != 887 sc->sc_copy_flags) { 888 sc->sc_copy_flags = ifp->if_flags & 889 (IFF_LINK0|IFF_LINK1|IFF_LINK2); 890 bridge_control(sc, 0, bridge_ioctl_reinit, NULL); 891 } 892 893 break; 894 895 case SIOCSIFMTU: 896 /* Do not allow the MTU to be changed on the bridge */ 897 error = EINVAL; 898 break; 899 900 default: 901 error = ether_ioctl(ifp, cmd, data); 902 break; 903 } 904 return (error); 905 } 906 907 /* 908 * bridge_mutecaps: 909 * 910 * Clear or restore unwanted capabilities on the member interface 911 */ 912 static void 913 bridge_mutecaps(struct bridge_ifinfo *bif_info, struct ifnet *ifp, int mute) 914 { 915 struct ifreq ifr; 916 917 if (ifp->if_ioctl == NULL) 918 return; 919 920 bzero(&ifr, sizeof(ifr)); 921 ifr.ifr_reqcap = ifp->if_capenable; 922 923 if (mute) { 924 /* mask off and save capabilities */ 925 bif_info->bifi_mutecap = ifr.ifr_reqcap & BRIDGE_IFCAPS_MASK; 926 if (bif_info->bifi_mutecap != 0) 927 ifr.ifr_reqcap &= ~BRIDGE_IFCAPS_MASK; 928 } else { 929 /* restore muted capabilities */ 930 ifr.ifr_reqcap |= bif_info->bifi_mutecap; 931 } 932 933 if (bif_info->bifi_mutecap != 0) { 934 ifnet_serialize_all(ifp); 935 ifp->if_ioctl(ifp, SIOCSIFCAP, (caddr_t)&ifr, NULL); 936 ifnet_deserialize_all(ifp); 937 } 938 } 939 940 /* 941 * bridge_lookup_member: 942 * 943 * Lookup a bridge member interface. 944 */ 945 static struct bridge_iflist * 946 bridge_lookup_member(struct bridge_softc *sc, const char *name) 947 { 948 struct bridge_iflist *bif; 949 950 TAILQ_FOREACH(bif, &sc->sc_iflists[mycpuid], bif_next) { 951 if (strcmp(bif->bif_ifp->if_xname, name) == 0) 952 return (bif); 953 } 954 return (NULL); 955 } 956 957 /* 958 * bridge_lookup_member_if: 959 * 960 * Lookup a bridge member interface by ifnet*. 961 */ 962 static struct bridge_iflist * 963 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp) 964 { 965 struct bridge_iflist *bif; 966 967 TAILQ_FOREACH(bif, &sc->sc_iflists[mycpuid], bif_next) { 968 if (bif->bif_ifp == member_ifp) 969 return (bif); 970 } 971 return (NULL); 972 } 973 974 /* 975 * bridge_lookup_member_ifinfo: 976 * 977 * Lookup a bridge member interface by bridge_ifinfo. 978 */ 979 static struct bridge_iflist * 980 bridge_lookup_member_ifinfo(struct bridge_softc *sc, 981 struct bridge_ifinfo *bif_info) 982 { 983 struct bridge_iflist *bif; 984 985 TAILQ_FOREACH(bif, &sc->sc_iflists[mycpuid], bif_next) { 986 if (bif->bif_info == bif_info) 987 return (bif); 988 } 989 return (NULL); 990 } 991 992 /* 993 * bridge_delete_member: 994 * 995 * Delete the specified member interface. 996 */ 997 static void 998 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif, 999 int gone) 1000 { 1001 struct ifnet *ifs = bif->bif_ifp; 1002 struct ifnet *bifp = sc->sc_ifp; 1003 struct bridge_ifinfo *bif_info = bif->bif_info; 1004 struct bridge_iflist_head saved_bifs; 1005 1006 ASSERT_IFNET_SERIALIZED_ALL(bifp); 1007 KKASSERT(bif_info != NULL); 1008 1009 ifs->if_bridge = NULL; 1010 1011 /* 1012 * Release bridge interface's serializer: 1013 * - To avoid possible dead lock. 1014 * - Various sync operation will block the current thread. 1015 */ 1016 ifnet_deserialize_all(bifp); 1017 1018 if (!gone) { 1019 switch (ifs->if_type) { 1020 case IFT_ETHER: 1021 case IFT_L2VLAN: 1022 /* 1023 * Take the interface out of promiscuous mode. 1024 */ 1025 ifpromisc(ifs, 0); 1026 bridge_mutecaps(bif_info, ifs, 0); 1027 break; 1028 1029 case IFT_GIF: 1030 break; 1031 1032 default: 1033 panic("bridge_delete_member: impossible"); 1034 break; 1035 } 1036 } 1037 1038 /* 1039 * Remove bifs from percpu linked list. 1040 * 1041 * Removed bifs are not freed immediately, instead, 1042 * they are saved in saved_bifs. They will be freed 1043 * after we make sure that no one is accessing them, 1044 * i.e. after following netmsg_service_sync() 1045 */ 1046 TAILQ_INIT(&saved_bifs); 1047 bridge_del_bif(sc, bif_info, &saved_bifs); 1048 1049 /* 1050 * Make sure that all protocol threads: 1051 * o see 'ifs' if_bridge is changed 1052 * o know that bif is removed from the percpu linked list 1053 */ 1054 netmsg_service_sync(); 1055 1056 /* 1057 * Free the removed bifs 1058 */ 1059 KKASSERT(!TAILQ_EMPTY(&saved_bifs)); 1060 while ((bif = TAILQ_FIRST(&saved_bifs)) != NULL) { 1061 TAILQ_REMOVE(&saved_bifs, bif, bif_next); 1062 kfree(bif, M_DEVBUF); 1063 } 1064 1065 /* See the comment in bridge_ioctl_stop() */ 1066 bridge_rtmsg_sync(sc); 1067 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL | IFBF_FLUSHSYNC); 1068 1069 ifnet_serialize_all(bifp); 1070 1071 if (bifp->if_flags & IFF_RUNNING) 1072 bstp_initialization(sc); 1073 1074 /* 1075 * Free the bif_info after bstp_initialization(), so that 1076 * bridge_softc.sc_root_port will not reference a dangling 1077 * pointer. 1078 */ 1079 kfree(bif_info, M_DEVBUF); 1080 } 1081 1082 /* 1083 * bridge_delete_span: 1084 * 1085 * Delete the specified span interface. 1086 */ 1087 static void 1088 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif) 1089 { 1090 KASSERT(bif->bif_ifp->if_bridge == NULL, 1091 ("%s: not a span interface", __func__)); 1092 1093 TAILQ_REMOVE(&sc->sc_iflists[mycpuid], bif, bif_next); 1094 kfree(bif, M_DEVBUF); 1095 } 1096 1097 static int 1098 bridge_ioctl_init(struct bridge_softc *sc, void *arg __unused) 1099 { 1100 struct ifnet *ifp = sc->sc_ifp; 1101 1102 if (ifp->if_flags & IFF_RUNNING) 1103 return 0; 1104 1105 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz, 1106 bridge_timer, sc); 1107 1108 ifp->if_flags |= IFF_RUNNING; 1109 bstp_initialization(sc); 1110 return 0; 1111 } 1112 1113 static int 1114 bridge_ioctl_stop(struct bridge_softc *sc, void *arg __unused) 1115 { 1116 struct ifnet *ifp = sc->sc_ifp; 1117 1118 if ((ifp->if_flags & IFF_RUNNING) == 0) 1119 return 0; 1120 1121 callout_stop(&sc->sc_brcallout); 1122 1123 crit_enter(); 1124 lwkt_dropmsg(&sc->sc_brtimemsg.lmsg); 1125 crit_exit(); 1126 1127 bstp_stop(sc); 1128 1129 ifp->if_flags &= ~IFF_RUNNING; 1130 1131 ifnet_deserialize_all(ifp); 1132 1133 /* Let everyone know that we are stopped */ 1134 netmsg_service_sync(); 1135 1136 /* 1137 * Sync ifnetX msgports in the order we forward rtnode 1138 * installation message. This is used to make sure that 1139 * all rtnode installation messages sent by bridge_rtupdate() 1140 * during above netmsg_service_sync() are flushed. 1141 */ 1142 bridge_rtmsg_sync(sc); 1143 bridge_rtflush(sc, IFBF_FLUSHDYN | IFBF_FLUSHSYNC); 1144 1145 ifnet_serialize_all(ifp); 1146 return 0; 1147 } 1148 1149 static int 1150 bridge_ioctl_add(struct bridge_softc *sc, void *arg) 1151 { 1152 struct ifbreq *req = arg; 1153 struct bridge_iflist *bif; 1154 struct bridge_ifinfo *bif_info; 1155 struct ifnet *ifs, *bifp; 1156 int error = 0; 1157 1158 bifp = sc->sc_ifp; 1159 ASSERT_IFNET_SERIALIZED_ALL(bifp); 1160 1161 ifs = ifunit(req->ifbr_ifsname); 1162 if (ifs == NULL) 1163 return (ENOENT); 1164 1165 /* If it's in the span list, it can't be a member. */ 1166 TAILQ_FOREACH(bif, &sc->sc_spanlist, bif_next) 1167 if (ifs == bif->bif_ifp) 1168 return (EBUSY); 1169 1170 /* Allow the first Ethernet member to define the MTU */ 1171 if (ifs->if_type != IFT_GIF) { 1172 if (TAILQ_EMPTY(&sc->sc_iflists[mycpuid])) { 1173 bifp->if_mtu = ifs->if_mtu; 1174 } else if (bifp->if_mtu != ifs->if_mtu) { 1175 if_printf(bifp, "invalid MTU for %s\n", ifs->if_xname); 1176 return (EINVAL); 1177 } 1178 } 1179 1180 if (ifs->if_bridge == sc) 1181 return (EEXIST); 1182 1183 if (ifs->if_bridge != NULL) 1184 return (EBUSY); 1185 1186 bif_info = kmalloc(sizeof(*bif_info), M_DEVBUF, M_WAITOK | M_ZERO); 1187 bif_info->bifi_priority = BSTP_DEFAULT_PORT_PRIORITY; 1188 bif_info->bifi_path_cost = BSTP_DEFAULT_PATH_COST; 1189 bif_info->bifi_ifp = ifs; 1190 bif_info->bifi_bond_weight = 1; 1191 1192 /* 1193 * Release bridge interface's serializer: 1194 * - To avoid possible dead lock. 1195 * - Various sync operation will block the current thread. 1196 */ 1197 ifnet_deserialize_all(bifp); 1198 1199 switch (ifs->if_type) { 1200 case IFT_ETHER: 1201 case IFT_L2VLAN: 1202 /* 1203 * Place the interface into promiscuous mode. 1204 */ 1205 error = ifpromisc(ifs, 1); 1206 if (error) { 1207 ifnet_serialize_all(bifp); 1208 goto out; 1209 } 1210 bridge_mutecaps(bif_info, ifs, 1); 1211 break; 1212 1213 case IFT_GIF: /* :^) */ 1214 break; 1215 1216 default: 1217 error = EINVAL; 1218 ifnet_serialize_all(bifp); 1219 goto out; 1220 } 1221 1222 /* 1223 * Add bifs to percpu linked lists 1224 */ 1225 bridge_add_bif(sc, bif_info, ifs); 1226 1227 ifnet_serialize_all(bifp); 1228 1229 if (bifp->if_flags & IFF_RUNNING) 1230 bstp_initialization(sc); 1231 else 1232 bstp_stop(sc); 1233 1234 /* 1235 * Everything has been setup, so let the member interface 1236 * deliver packets to this bridge on its input/output path. 1237 */ 1238 ifs->if_bridge = sc; 1239 out: 1240 if (error) { 1241 if (bif_info != NULL) 1242 kfree(bif_info, M_DEVBUF); 1243 } 1244 return (error); 1245 } 1246 1247 static int 1248 bridge_ioctl_del(struct bridge_softc *sc, void *arg) 1249 { 1250 struct ifbreq *req = arg; 1251 struct bridge_iflist *bif; 1252 1253 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1254 if (bif == NULL) 1255 return (ENOENT); 1256 1257 bridge_delete_member(sc, bif, 0); 1258 1259 return (0); 1260 } 1261 1262 static int 1263 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg) 1264 { 1265 struct ifbreq *req = arg; 1266 struct bridge_iflist *bif; 1267 1268 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1269 if (bif == NULL) 1270 return (ENOENT); 1271 bridge_ioctl_fillflags(sc, bif, req); 1272 return (0); 1273 } 1274 1275 static void 1276 bridge_ioctl_fillflags(struct bridge_softc *sc, struct bridge_iflist *bif, 1277 struct ifbreq *req) 1278 { 1279 req->ifbr_ifsflags = bif->bif_flags; 1280 req->ifbr_state = bif->bif_state; 1281 req->ifbr_priority = bif->bif_priority; 1282 req->ifbr_path_cost = bif->bif_path_cost; 1283 req->ifbr_bond_weight = bif->bif_bond_weight; 1284 req->ifbr_portno = bif->bif_ifp->if_index & 0xff; 1285 if (bif->bif_flags & IFBIF_STP) { 1286 req->ifbr_peer_root = bif->bif_peer_root; 1287 req->ifbr_peer_bridge = bif->bif_peer_bridge; 1288 req->ifbr_peer_cost = bif->bif_peer_cost; 1289 req->ifbr_peer_port = bif->bif_peer_port; 1290 if (bstp_supersedes_port_info(sc, bif)) { 1291 req->ifbr_designated_root = bif->bif_peer_root; 1292 req->ifbr_designated_bridge = bif->bif_peer_bridge; 1293 req->ifbr_designated_cost = bif->bif_peer_cost; 1294 req->ifbr_designated_port = bif->bif_peer_port; 1295 } else { 1296 req->ifbr_designated_root = sc->sc_bridge_id; 1297 req->ifbr_designated_bridge = sc->sc_bridge_id; 1298 req->ifbr_designated_cost = bif->bif_path_cost + 1299 bif->bif_peer_cost; 1300 req->ifbr_designated_port = bif->bif_port_id; 1301 } 1302 } else { 1303 req->ifbr_peer_root = 0; 1304 req->ifbr_peer_bridge = 0; 1305 req->ifbr_peer_cost = 0; 1306 req->ifbr_peer_port = 0; 1307 req->ifbr_designated_root = 0; 1308 req->ifbr_designated_bridge = 0; 1309 req->ifbr_designated_cost = 0; 1310 req->ifbr_designated_port = 0; 1311 } 1312 } 1313 1314 static int 1315 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg) 1316 { 1317 struct ifbreq *req = arg; 1318 struct bridge_iflist *bif; 1319 struct ifnet *bifp = sc->sc_ifp; 1320 1321 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1322 if (bif == NULL) 1323 return (ENOENT); 1324 1325 if (req->ifbr_ifsflags & IFBIF_SPAN) { 1326 /* SPAN is readonly */ 1327 return (EINVAL); 1328 } 1329 1330 if (req->ifbr_ifsflags & IFBIF_STP) { 1331 switch (bif->bif_ifp->if_type) { 1332 case IFT_ETHER: 1333 /* These can do spanning tree. */ 1334 break; 1335 1336 default: 1337 /* Nothing else can. */ 1338 return (EINVAL); 1339 } 1340 } 1341 1342 bif->bif_flags = (bif->bif_flags & IFBIF_KEEPMASK) | 1343 (req->ifbr_ifsflags & ~IFBIF_KEEPMASK); 1344 if (bifp->if_flags & IFF_RUNNING) 1345 bstp_initialization(sc); 1346 1347 return (0); 1348 } 1349 1350 static int 1351 bridge_ioctl_scache(struct bridge_softc *sc, void *arg) 1352 { 1353 struct ifbrparam *param = arg; 1354 struct ifnet *ifp = sc->sc_ifp; 1355 1356 sc->sc_brtmax = param->ifbrp_csize; 1357 1358 ifnet_deserialize_all(ifp); 1359 bridge_rttrim(sc); 1360 ifnet_serialize_all(ifp); 1361 1362 return (0); 1363 } 1364 1365 static int 1366 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg) 1367 { 1368 struct ifbrparam *param = arg; 1369 1370 param->ifbrp_csize = sc->sc_brtmax; 1371 1372 return (0); 1373 } 1374 1375 static int 1376 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg) 1377 { 1378 struct bridge_control_arg *bc_arg = arg; 1379 struct ifbifconf *bifc = arg; 1380 struct bridge_iflist *bif; 1381 struct ifbreq *breq; 1382 int count, len; 1383 1384 count = 0; 1385 TAILQ_FOREACH(bif, &sc->sc_iflists[mycpuid], bif_next) 1386 count++; 1387 TAILQ_FOREACH(bif, &sc->sc_spanlist, bif_next) 1388 count++; 1389 1390 if (bifc->ifbic_len == 0) { 1391 bifc->ifbic_len = sizeof(*breq) * count; 1392 return 0; 1393 } else if (count == 0 || bifc->ifbic_len < sizeof(*breq)) { 1394 bifc->ifbic_len = 0; 1395 return 0; 1396 } 1397 1398 len = min(bifc->ifbic_len, sizeof(*breq) * count); 1399 KKASSERT(len >= sizeof(*breq)); 1400 1401 breq = kmalloc(len, M_TEMP, M_WAITOK | M_NULLOK | M_ZERO); 1402 if (breq == NULL) { 1403 bifc->ifbic_len = 0; 1404 return ENOMEM; 1405 } 1406 bc_arg->bca_kptr = breq; 1407 1408 count = 0; 1409 TAILQ_FOREACH(bif, &sc->sc_iflists[mycpuid], bif_next) { 1410 if (len < sizeof(*breq)) 1411 break; 1412 1413 strlcpy(breq->ifbr_ifsname, bif->bif_ifp->if_xname, 1414 sizeof(breq->ifbr_ifsname)); 1415 bridge_ioctl_fillflags(sc, bif, breq); 1416 breq++; 1417 count++; 1418 len -= sizeof(*breq); 1419 } 1420 TAILQ_FOREACH(bif, &sc->sc_spanlist, bif_next) { 1421 if (len < sizeof(*breq)) 1422 break; 1423 1424 strlcpy(breq->ifbr_ifsname, bif->bif_ifp->if_xname, 1425 sizeof(breq->ifbr_ifsname)); 1426 breq->ifbr_ifsflags = bif->bif_flags; 1427 breq->ifbr_portno = bif->bif_ifp->if_index & 0xff; 1428 breq++; 1429 count++; 1430 len -= sizeof(*breq); 1431 } 1432 1433 bifc->ifbic_len = sizeof(*breq) * count; 1434 KKASSERT(bifc->ifbic_len > 0); 1435 1436 bc_arg->bca_len = bifc->ifbic_len; 1437 bc_arg->bca_uptr = bifc->ifbic_req; 1438 return 0; 1439 } 1440 1441 static int 1442 bridge_ioctl_rts(struct bridge_softc *sc, void *arg) 1443 { 1444 struct bridge_control_arg *bc_arg = arg; 1445 struct ifbaconf *bac = arg; 1446 struct bridge_rtnode *brt; 1447 struct ifbareq *bareq; 1448 int count, len; 1449 1450 count = 0; 1451 LIST_FOREACH(brt, &sc->sc_rtlists[mycpuid], brt_list) 1452 count++; 1453 1454 if (bac->ifbac_len == 0) { 1455 bac->ifbac_len = sizeof(*bareq) * count; 1456 return 0; 1457 } else if (count == 0 || bac->ifbac_len < sizeof(*bareq)) { 1458 bac->ifbac_len = 0; 1459 return 0; 1460 } 1461 1462 len = min(bac->ifbac_len, sizeof(*bareq) * count); 1463 KKASSERT(len >= sizeof(*bareq)); 1464 1465 bareq = kmalloc(len, M_TEMP, M_WAITOK | M_NULLOK | M_ZERO); 1466 if (bareq == NULL) { 1467 bac->ifbac_len = 0; 1468 return ENOMEM; 1469 } 1470 bc_arg->bca_kptr = bareq; 1471 1472 count = 0; 1473 LIST_FOREACH(brt, &sc->sc_rtlists[mycpuid], brt_list) { 1474 struct bridge_rtinfo *bri = brt->brt_info; 1475 unsigned long expire; 1476 1477 if (len < sizeof(*bareq)) 1478 break; 1479 1480 strlcpy(bareq->ifba_ifsname, bri->bri_ifp->if_xname, 1481 sizeof(bareq->ifba_ifsname)); 1482 memcpy(bareq->ifba_dst, brt->brt_addr, sizeof(brt->brt_addr)); 1483 expire = bri->bri_expire; 1484 if ((bri->bri_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 1485 time_second < expire) 1486 bareq->ifba_expire = expire - time_second; 1487 else 1488 bareq->ifba_expire = 0; 1489 bareq->ifba_flags = bri->bri_flags; 1490 bareq++; 1491 count++; 1492 len -= sizeof(*bareq); 1493 } 1494 1495 bac->ifbac_len = sizeof(*bareq) * count; 1496 KKASSERT(bac->ifbac_len > 0); 1497 1498 bc_arg->bca_len = bac->ifbac_len; 1499 bc_arg->bca_uptr = bac->ifbac_req; 1500 return 0; 1501 } 1502 1503 static int 1504 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg) 1505 { 1506 struct ifbareq *req = arg; 1507 struct bridge_iflist *bif; 1508 struct ifnet *ifp = sc->sc_ifp; 1509 int error; 1510 1511 ASSERT_IFNET_SERIALIZED_ALL(ifp); 1512 1513 bif = bridge_lookup_member(sc, req->ifba_ifsname); 1514 if (bif == NULL) 1515 return (ENOENT); 1516 1517 ifnet_deserialize_all(ifp); 1518 error = bridge_rtsaddr(sc, req->ifba_dst, bif->bif_ifp, 1519 req->ifba_flags); 1520 ifnet_serialize_all(ifp); 1521 return (error); 1522 } 1523 1524 static int 1525 bridge_ioctl_sto(struct bridge_softc *sc, void *arg) 1526 { 1527 struct ifbrparam *param = arg; 1528 1529 sc->sc_brttimeout = param->ifbrp_ctime; 1530 1531 return (0); 1532 } 1533 1534 static int 1535 bridge_ioctl_gto(struct bridge_softc *sc, void *arg) 1536 { 1537 struct ifbrparam *param = arg; 1538 1539 param->ifbrp_ctime = sc->sc_brttimeout; 1540 1541 return (0); 1542 } 1543 1544 static int 1545 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg) 1546 { 1547 struct ifbareq *req = arg; 1548 struct ifnet *ifp = sc->sc_ifp; 1549 int error; 1550 1551 ifnet_deserialize_all(ifp); 1552 error = bridge_rtdaddr(sc, req->ifba_dst); 1553 ifnet_serialize_all(ifp); 1554 return error; 1555 } 1556 1557 static int 1558 bridge_ioctl_flush(struct bridge_softc *sc, void *arg) 1559 { 1560 struct ifbreq *req = arg; 1561 struct ifnet *ifp = sc->sc_ifp; 1562 1563 ifnet_deserialize_all(ifp); 1564 bridge_rtflush(sc, req->ifbr_ifsflags | IFBF_FLUSHSYNC); 1565 ifnet_serialize_all(ifp); 1566 1567 return (0); 1568 } 1569 1570 static int 1571 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg) 1572 { 1573 struct ifbrparam *param = arg; 1574 1575 param->ifbrp_prio = sc->sc_bridge_priority; 1576 1577 return (0); 1578 } 1579 1580 static int 1581 bridge_ioctl_spri(struct bridge_softc *sc, void *arg) 1582 { 1583 struct ifbrparam *param = arg; 1584 1585 sc->sc_bridge_priority = param->ifbrp_prio; 1586 1587 if (sc->sc_ifp->if_flags & IFF_RUNNING) 1588 bstp_initialization(sc); 1589 1590 return (0); 1591 } 1592 1593 static int 1594 bridge_ioctl_reinit(struct bridge_softc *sc, void *arg __unused) 1595 { 1596 if (sc->sc_ifp->if_flags & IFF_RUNNING) 1597 bstp_initialization(sc); 1598 return (0); 1599 } 1600 1601 static int 1602 bridge_ioctl_ght(struct bridge_softc *sc, void *arg) 1603 { 1604 struct ifbrparam *param = arg; 1605 1606 param->ifbrp_hellotime = sc->sc_bridge_hello_time >> 8; 1607 1608 return (0); 1609 } 1610 1611 static int 1612 bridge_ioctl_sht(struct bridge_softc *sc, void *arg) 1613 { 1614 struct ifbrparam *param = arg; 1615 1616 if (param->ifbrp_hellotime == 0) 1617 return (EINVAL); 1618 sc->sc_bridge_hello_time = param->ifbrp_hellotime << 8; 1619 1620 if (sc->sc_ifp->if_flags & IFF_RUNNING) 1621 bstp_initialization(sc); 1622 1623 return (0); 1624 } 1625 1626 static int 1627 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg) 1628 { 1629 struct ifbrparam *param = arg; 1630 1631 param->ifbrp_fwddelay = sc->sc_bridge_forward_delay >> 8; 1632 1633 return (0); 1634 } 1635 1636 static int 1637 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg) 1638 { 1639 struct ifbrparam *param = arg; 1640 1641 if (param->ifbrp_fwddelay == 0) 1642 return (EINVAL); 1643 sc->sc_bridge_forward_delay = param->ifbrp_fwddelay << 8; 1644 1645 if (sc->sc_ifp->if_flags & IFF_RUNNING) 1646 bstp_initialization(sc); 1647 1648 return (0); 1649 } 1650 1651 static int 1652 bridge_ioctl_gma(struct bridge_softc *sc, void *arg) 1653 { 1654 struct ifbrparam *param = arg; 1655 1656 param->ifbrp_maxage = sc->sc_bridge_max_age >> 8; 1657 1658 return (0); 1659 } 1660 1661 static int 1662 bridge_ioctl_sma(struct bridge_softc *sc, void *arg) 1663 { 1664 struct ifbrparam *param = arg; 1665 1666 if (param->ifbrp_maxage == 0) 1667 return (EINVAL); 1668 sc->sc_bridge_max_age = param->ifbrp_maxage << 8; 1669 1670 if (sc->sc_ifp->if_flags & IFF_RUNNING) 1671 bstp_initialization(sc); 1672 1673 return (0); 1674 } 1675 1676 static int 1677 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg) 1678 { 1679 struct ifbreq *req = arg; 1680 struct bridge_iflist *bif; 1681 1682 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1683 if (bif == NULL) 1684 return (ENOENT); 1685 1686 bif->bif_priority = req->ifbr_priority; 1687 1688 if (sc->sc_ifp->if_flags & IFF_RUNNING) 1689 bstp_initialization(sc); 1690 1691 return (0); 1692 } 1693 1694 static int 1695 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg) 1696 { 1697 struct ifbreq *req = arg; 1698 struct bridge_iflist *bif; 1699 1700 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1701 if (bif == NULL) 1702 return (ENOENT); 1703 1704 bif->bif_path_cost = req->ifbr_path_cost; 1705 1706 if (sc->sc_ifp->if_flags & IFF_RUNNING) 1707 bstp_initialization(sc); 1708 1709 return (0); 1710 } 1711 1712 static int 1713 bridge_ioctl_sifbondwght(struct bridge_softc *sc, void *arg) 1714 { 1715 struct ifbreq *req = arg; 1716 struct bridge_iflist *bif; 1717 1718 bif = bridge_lookup_member(sc, req->ifbr_ifsname); 1719 if (bif == NULL) 1720 return (ENOENT); 1721 1722 bif->bif_bond_weight = req->ifbr_bond_weight; 1723 1724 /* no reinit needed */ 1725 1726 return (0); 1727 } 1728 1729 static int 1730 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg) 1731 { 1732 struct ifbreq *req = arg; 1733 struct bridge_iflist *bif; 1734 struct ifnet *ifs; 1735 struct bridge_ifinfo *bif_info; 1736 1737 ifs = ifunit(req->ifbr_ifsname); 1738 if (ifs == NULL) 1739 return (ENOENT); 1740 1741 TAILQ_FOREACH(bif, &sc->sc_spanlist, bif_next) 1742 if (ifs == bif->bif_ifp) 1743 return (EBUSY); 1744 1745 if (ifs->if_bridge != NULL) 1746 return (EBUSY); 1747 1748 switch (ifs->if_type) { 1749 case IFT_ETHER: 1750 case IFT_GIF: 1751 case IFT_L2VLAN: 1752 break; 1753 1754 default: 1755 return (EINVAL); 1756 } 1757 1758 /* 1759 * bif_info is needed for bif_flags 1760 */ 1761 bif_info = kmalloc(sizeof(*bif_info), M_DEVBUF, M_WAITOK | M_ZERO); 1762 bif_info->bifi_ifp = ifs; 1763 1764 bif = kmalloc(sizeof(*bif), M_DEVBUF, M_WAITOK | M_ZERO); 1765 bif->bif_ifp = ifs; 1766 bif->bif_info = bif_info; 1767 bif->bif_flags = IFBIF_SPAN; 1768 /* NOTE: span bif does not need bridge_ifinfo */ 1769 1770 TAILQ_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next); 1771 1772 sc->sc_span = 1; 1773 1774 return (0); 1775 } 1776 1777 static int 1778 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg) 1779 { 1780 struct ifbreq *req = arg; 1781 struct bridge_iflist *bif; 1782 struct ifnet *ifs; 1783 1784 ifs = ifunit(req->ifbr_ifsname); 1785 if (ifs == NULL) 1786 return (ENOENT); 1787 1788 TAILQ_FOREACH(bif, &sc->sc_spanlist, bif_next) 1789 if (ifs == bif->bif_ifp) 1790 break; 1791 1792 if (bif == NULL) 1793 return (ENOENT); 1794 1795 bridge_delete_span(sc, bif); 1796 1797 if (TAILQ_EMPTY(&sc->sc_spanlist)) 1798 sc->sc_span = 0; 1799 1800 return (0); 1801 } 1802 1803 static void 1804 bridge_ifdetach_dispatch(netmsg_t msg) 1805 { 1806 struct ifnet *ifp, *bifp; 1807 struct bridge_softc *sc; 1808 struct bridge_iflist *bif; 1809 1810 ifp = msg->lmsg.u.ms_resultp; 1811 sc = ifp->if_bridge; 1812 1813 /* Check if the interface is a bridge member */ 1814 if (sc != NULL) { 1815 bifp = sc->sc_ifp; 1816 1817 ifnet_serialize_all(bifp); 1818 1819 bif = bridge_lookup_member_if(sc, ifp); 1820 if (bif != NULL) { 1821 bridge_delete_member(sc, bif, 1); 1822 } else { 1823 /* XXX Why bif will be NULL? */ 1824 } 1825 1826 ifnet_deserialize_all(bifp); 1827 goto reply; 1828 } 1829 1830 crit_enter(); /* XXX MP */ 1831 1832 /* Check if the interface is a span port */ 1833 LIST_FOREACH(sc, &bridge_list, sc_list) { 1834 bifp = sc->sc_ifp; 1835 1836 ifnet_serialize_all(bifp); 1837 1838 TAILQ_FOREACH(bif, &sc->sc_spanlist, bif_next) 1839 if (ifp == bif->bif_ifp) { 1840 bridge_delete_span(sc, bif); 1841 break; 1842 } 1843 1844 ifnet_deserialize_all(bifp); 1845 } 1846 1847 crit_exit(); 1848 1849 reply: 1850 lwkt_replymsg(&msg->lmsg, 0); 1851 } 1852 1853 /* 1854 * bridge_ifdetach: 1855 * 1856 * Detach an interface from a bridge. Called when a member 1857 * interface is detaching. 1858 */ 1859 static void 1860 bridge_ifdetach(void *arg __unused, struct ifnet *ifp) 1861 { 1862 struct netmsg_base msg; 1863 1864 netmsg_init(&msg, NULL, &curthread->td_msgport, 1865 0, bridge_ifdetach_dispatch); 1866 msg.lmsg.u.ms_resultp = ifp; 1867 1868 lwkt_domsg(BRIDGE_CFGPORT, &msg.lmsg, 0); 1869 } 1870 1871 /* 1872 * bridge_init: 1873 * 1874 * Initialize a bridge interface. 1875 */ 1876 static void 1877 bridge_init(void *xsc) 1878 { 1879 bridge_control(xsc, SIOCSIFFLAGS, bridge_ioctl_init, NULL); 1880 } 1881 1882 /* 1883 * bridge_stop: 1884 * 1885 * Stop the bridge interface. 1886 */ 1887 static void 1888 bridge_stop(struct ifnet *ifp) 1889 { 1890 bridge_control(ifp->if_softc, SIOCSIFFLAGS, bridge_ioctl_stop, NULL); 1891 } 1892 1893 /* 1894 * Returns TRUE if the packet is being sent 'from us'... from our bridge 1895 * interface or from any member of our bridge interface. This is used 1896 * later on to force the MAC to be the MAC of our bridge interface. 1897 */ 1898 static int 1899 bridge_from_us(struct bridge_softc *sc, struct ether_header *eh) 1900 { 1901 struct bridge_iflist *bif; 1902 1903 if (memcmp(eh->ether_shost, IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN) == 0) 1904 return (1); 1905 1906 TAILQ_FOREACH(bif, &sc->sc_iflists[mycpuid], bif_next) { 1907 if (memcmp(eh->ether_shost, IF_LLADDR(bif->bif_ifp), 1908 ETHER_ADDR_LEN) == 0) { 1909 return (1); 1910 } 1911 } 1912 return (0); 1913 } 1914 1915 /* 1916 * bridge_enqueue: 1917 * 1918 * Enqueue a packet on a bridge member interface. 1919 * 1920 */ 1921 void 1922 bridge_enqueue(struct ifnet *dst_ifp, struct mbuf *m) 1923 { 1924 struct netmsg_packet *nmp; 1925 1926 mbuftrackid(m, 64); 1927 1928 nmp = &m->m_hdr.mh_netmsg; 1929 netmsg_init(&nmp->base, NULL, &netisr_apanic_rport, 1930 0, bridge_enqueue_handler); 1931 nmp->nm_packet = m; 1932 nmp->base.lmsg.u.ms_resultp = dst_ifp; 1933 1934 lwkt_sendmsg(netisr_cpuport(mycpuid), &nmp->base.lmsg); 1935 } 1936 1937 /* 1938 * After looking up dst_if in our forwarding table we still have to 1939 * deal with channel bonding. Find the best interface in the bonding set. 1940 */ 1941 static struct ifnet * 1942 bridge_select_unicast(struct bridge_softc *sc, struct ifnet *dst_if, 1943 int from_blocking, struct mbuf *m) 1944 { 1945 struct bridge_iflist *bif, *nbif; 1946 struct ifnet *alt_if; 1947 int alt_priority; 1948 int priority; 1949 1950 /* 1951 * Unicast, kinda replicates the output side of bridge_output(). 1952 * 1953 * Even though this is a uni-cast packet we may have to select 1954 * an interface from a bonding set. 1955 */ 1956 bif = bridge_lookup_member_if(sc, dst_if); 1957 if (bif == NULL) { 1958 /* Not a member of the bridge (anymore?) */ 1959 return NULL; 1960 } 1961 1962 /* 1963 * If STP is enabled on the target we are an equal opportunity 1964 * employer and do not necessarily output to dst_if. Instead 1965 * scan available links with the same MAC as the current dst_if 1966 * and choose the best one. 1967 * 1968 * We also need to do this because arp entries tag onto a particular 1969 * interface and if it happens to be dead then the packets will 1970 * go into a bit bucket. 1971 * 1972 * If LINK2 is set the matching links are bonded and we-round robin. 1973 * (the MAC address must be the same for the participating links). 1974 * In this case links in a STP FORWARDING or BONDED state are 1975 * allowed for unicast packets. 1976 */ 1977 if (bif->bif_flags & IFBIF_STP) { 1978 alt_if = NULL; 1979 alt_priority = 0; 1980 priority = 0; 1981 1982 TAILQ_FOREACH_MUTABLE(bif, &sc->sc_iflists[mycpuid], 1983 bif_next, nbif) { 1984 /* 1985 * dst_if may imply a bonding set so we must compare 1986 * MAC addresses. 1987 */ 1988 if (memcmp(IF_LLADDR(bif->bif_ifp), 1989 IF_LLADDR(dst_if), 1990 ETHER_ADDR_LEN) != 0) { 1991 continue; 1992 } 1993 1994 if ((bif->bif_ifp->if_flags & IFF_RUNNING) == 0) 1995 continue; 1996 1997 /* 1998 * NOTE: We allow tranmissions through a BLOCKING 1999 * or LEARNING interface only as a last resort. 2000 * We DISALLOW both cases if the receiving 2001 * 2002 * NOTE: If we send a packet through a learning 2003 * interface the receiving end (if also in 2004 * LEARNING) will throw it away, so this is 2005 * the ultimate last resort. 2006 */ 2007 switch(bif->bif_state) { 2008 case BSTP_IFSTATE_BLOCKING: 2009 if (from_blocking == 0 && 2010 bif->bif_priority + 256 > alt_priority) { 2011 alt_priority = bif->bif_priority + 256; 2012 alt_if = bif->bif_ifp; 2013 } 2014 continue; 2015 case BSTP_IFSTATE_LEARNING: 2016 if (from_blocking == 0 && 2017 bif->bif_priority > alt_priority) { 2018 alt_priority = bif->bif_priority; 2019 alt_if = bif->bif_ifp; 2020 } 2021 continue; 2022 case BSTP_IFSTATE_L1BLOCKING: 2023 case BSTP_IFSTATE_LISTENING: 2024 case BSTP_IFSTATE_DISABLED: 2025 continue; 2026 default: 2027 /* FORWARDING, BONDED */ 2028 break; 2029 } 2030 2031 /* 2032 * XXX we need to use the toepliz hash or 2033 * something like that instead of 2034 * round-robining. 2035 */ 2036 if (sc->sc_ifp->if_flags & IFF_LINK2) { 2037 dst_if = bif->bif_ifp; 2038 if (++bif->bif_bond_count >= 2039 bif->bif_bond_weight) { 2040 bif->bif_bond_count = 0; 2041 TAILQ_REMOVE(&sc->sc_iflists[mycpuid], 2042 bif, bif_next); 2043 TAILQ_INSERT_TAIL( 2044 &sc->sc_iflists[mycpuid], 2045 bif, bif_next); 2046 } 2047 priority = 1; 2048 break; 2049 } 2050 2051 /* 2052 * Select best interface in the FORWARDING or 2053 * BONDED set. Well, there shouldn't be any 2054 * in a BONDED state if LINK2 is not set (they 2055 * will all be in a BLOCKING) state, but there 2056 * could be a transitory condition here. 2057 */ 2058 if (bif->bif_priority > priority) { 2059 priority = bif->bif_priority; 2060 dst_if = bif->bif_ifp; 2061 } 2062 } 2063 2064 /* 2065 * If no suitable interfaces were found but a suitable 2066 * alternative interface was found, use the alternative 2067 * interface. 2068 */ 2069 if (priority == 0 && alt_if) 2070 dst_if = alt_if; 2071 } 2072 2073 /* 2074 * At this point, we're dealing with a unicast frame 2075 * going to a different interface. 2076 */ 2077 if ((dst_if->if_flags & IFF_RUNNING) == 0) 2078 dst_if = NULL; 2079 return (dst_if); 2080 } 2081 2082 2083 /* 2084 * bridge_output: 2085 * 2086 * Send output from a bridge member interface. This 2087 * performs the bridging function for locally originated 2088 * packets. 2089 * 2090 * The mbuf has the Ethernet header already attached. We must 2091 * enqueue or free the mbuf before returning. 2092 */ 2093 static int 2094 bridge_output(struct ifnet *ifp, struct mbuf *m) 2095 { 2096 struct bridge_softc *sc = ifp->if_bridge; 2097 struct bridge_iflist *bif, *nbif; 2098 struct ether_header *eh; 2099 struct ifnet *dst_if, *alt_if, *bifp; 2100 int from_us; 2101 int alt_priority; 2102 2103 ASSERT_IFNET_NOT_SERIALIZED_ALL(ifp); 2104 mbuftrackid(m, 65); 2105 2106 /* 2107 * Make sure that we are still a member of a bridge interface. 2108 */ 2109 if (sc == NULL) { 2110 m_freem(m); 2111 return (0); 2112 } 2113 bifp = sc->sc_ifp; 2114 2115 /* 2116 * Acquire header 2117 */ 2118 if (m->m_len < ETHER_HDR_LEN) { 2119 m = m_pullup(m, ETHER_HDR_LEN); 2120 if (m == NULL) { 2121 IFNET_STAT_INC(bifp, oerrors, 1); 2122 return (0); 2123 } 2124 } 2125 eh = mtod(m, struct ether_header *); 2126 from_us = bridge_from_us(sc, eh); 2127 2128 /* 2129 * If bridge is down, but the original output interface is up, 2130 * go ahead and send out that interface. Otherwise, the packet 2131 * is dropped below. 2132 */ 2133 if ((bifp->if_flags & IFF_RUNNING) == 0) { 2134 dst_if = ifp; 2135 goto sendunicast; 2136 } 2137 2138 /* 2139 * If the packet is a multicast, or we don't know a better way to 2140 * get there, send to all interfaces. 2141 */ 2142 if (ETHER_IS_MULTICAST(eh->ether_dhost)) 2143 dst_if = NULL; 2144 else 2145 dst_if = bridge_rtlookup(sc, eh->ether_dhost); 2146 2147 if (dst_if == NULL) { 2148 struct mbuf *mc; 2149 int used = 0; 2150 int found = 0; 2151 2152 if (sc->sc_span) 2153 bridge_span(sc, m); 2154 2155 alt_if = NULL; 2156 alt_priority = 0; 2157 TAILQ_FOREACH_MUTABLE(bif, &sc->sc_iflists[mycpuid], 2158 bif_next, nbif) { 2159 dst_if = bif->bif_ifp; 2160 2161 if ((dst_if->if_flags & IFF_RUNNING) == 0) 2162 continue; 2163 2164 /* 2165 * If this is not the original output interface, 2166 * and the interface is participating in spanning 2167 * tree, make sure the port is in a state that 2168 * allows forwarding. 2169 * 2170 * We keep track of a possible backup IF if we are 2171 * unable to find any interfaces to forward through. 2172 * 2173 * NOTE: Currently round-robining is not implemented 2174 * across bonded interface groups (needs an 2175 * algorithm to track each group somehow). 2176 * 2177 * Similarly we track only one alternative 2178 * interface if no suitable interfaces are 2179 * found. 2180 */ 2181 if (dst_if != ifp && 2182 (bif->bif_flags & IFBIF_STP) != 0) { 2183 switch (bif->bif_state) { 2184 case BSTP_IFSTATE_BONDED: 2185 if (bif->bif_priority + 512 > 2186 alt_priority) { 2187 alt_priority = 2188 bif->bif_priority + 512; 2189 alt_if = bif->bif_ifp; 2190 } 2191 continue; 2192 case BSTP_IFSTATE_BLOCKING: 2193 if (bif->bif_priority + 256 > 2194 alt_priority) { 2195 alt_priority = 2196 bif->bif_priority + 256; 2197 alt_if = bif->bif_ifp; 2198 } 2199 continue; 2200 case BSTP_IFSTATE_LEARNING: 2201 if (bif->bif_priority > alt_priority) { 2202 alt_priority = 2203 bif->bif_priority; 2204 alt_if = bif->bif_ifp; 2205 } 2206 continue; 2207 case BSTP_IFSTATE_L1BLOCKING: 2208 case BSTP_IFSTATE_LISTENING: 2209 case BSTP_IFSTATE_DISABLED: 2210 continue; 2211 default: 2212 /* FORWARDING */ 2213 break; 2214 } 2215 } 2216 2217 KKASSERT(used == 0); 2218 if (TAILQ_NEXT(bif, bif_next) == NULL) { 2219 used = 1; 2220 mc = m; 2221 } else { 2222 mc = m_copypacket(m, MB_DONTWAIT); 2223 if (mc == NULL) { 2224 IFNET_STAT_INC(bifp, oerrors, 1); 2225 continue; 2226 } 2227 } 2228 2229 /* 2230 * If the packet is 'from' us override ether_shost. 2231 */ 2232 bridge_handoff(sc, dst_if, mc, from_us); 2233 found = 1; 2234 2235 if (nbif != NULL && !nbif->bif_onlist) { 2236 KKASSERT(bif->bif_onlist); 2237 nbif = TAILQ_NEXT(bif, bif_next); 2238 } 2239 } 2240 2241 /* 2242 * If we couldn't find anything use the backup interface 2243 * if we have one. 2244 */ 2245 if (found == 0 && alt_if) { 2246 KKASSERT(used == 0); 2247 mc = m; 2248 used = 1; 2249 bridge_handoff(sc, alt_if, mc, from_us); 2250 } 2251 2252 if (used == 0) 2253 m_freem(m); 2254 return (0); 2255 } 2256 2257 /* 2258 * Unicast 2259 */ 2260 sendunicast: 2261 dst_if = bridge_select_unicast(sc, dst_if, 0, m); 2262 2263 if (sc->sc_span) 2264 bridge_span(sc, m); 2265 if (dst_if == NULL) 2266 m_freem(m); 2267 else 2268 bridge_handoff(sc, dst_if, m, from_us); 2269 return (0); 2270 } 2271 2272 /* 2273 * Returns the bridge interface associated with an ifc. 2274 * Pass ifp->if_bridge (must not be NULL). Used by the ARP 2275 * code to supply the bridge for the is-at info, making 2276 * the bridge responsible for matching local addresses. 2277 * 2278 * Without this the ARP code will supply bridge member interfaces 2279 * for the is-at which makes it difficult the bridge to fail-over 2280 * interfaces (amoung other things). 2281 */ 2282 static struct ifnet * 2283 bridge_interface(void *if_bridge) 2284 { 2285 struct bridge_softc *sc = if_bridge; 2286 return (sc->sc_ifp); 2287 } 2288 2289 /* 2290 * bridge_start: 2291 * 2292 * Start output on a bridge. 2293 */ 2294 static void 2295 bridge_start(struct ifnet *ifp, struct ifaltq_subque *ifsq) 2296 { 2297 struct bridge_softc *sc = ifp->if_softc; 2298 2299 ASSERT_ALTQ_SQ_DEFAULT(ifp, ifsq); 2300 ASSERT_ALTQ_SQ_SERIALIZED_HW(ifsq); 2301 2302 ifsq_set_oactive(ifsq); 2303 for (;;) { 2304 struct ifnet *dst_if = NULL; 2305 struct ether_header *eh; 2306 struct mbuf *m; 2307 2308 m = ifsq_dequeue(ifsq, NULL); 2309 if (m == NULL) 2310 break; 2311 mbuftrackid(m, 75); 2312 2313 if (m->m_len < sizeof(*eh)) { 2314 m = m_pullup(m, sizeof(*eh)); 2315 if (m == NULL) { 2316 IFNET_STAT_INC(ifp, oerrors, 1); 2317 continue; 2318 } 2319 } 2320 eh = mtod(m, struct ether_header *); 2321 2322 BPF_MTAP(ifp, m); 2323 IFNET_STAT_INC(ifp, opackets, 1); 2324 2325 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) 2326 dst_if = bridge_rtlookup(sc, eh->ether_dhost); 2327 2328 /* 2329 * Multicast or broadcast 2330 */ 2331 if (dst_if == NULL) { 2332 bridge_start_bcast(sc, m); 2333 continue; 2334 } 2335 2336 /* 2337 * Unicast 2338 */ 2339 dst_if = bridge_select_unicast(sc, dst_if, 0, m); 2340 2341 if (dst_if == NULL) 2342 m_freem(m); 2343 else 2344 bridge_enqueue(dst_if, m); 2345 } 2346 ifsq_clr_oactive(ifsq); 2347 } 2348 2349 /* 2350 * bridge_forward: 2351 * 2352 * Forward packets received on a bridge interface via the input 2353 * path. 2354 * 2355 * This implements the forwarding function of the bridge. 2356 */ 2357 static void 2358 bridge_forward(struct bridge_softc *sc, struct mbuf *m) 2359 { 2360 struct bridge_iflist *bif; 2361 struct ifnet *src_if, *dst_if, *ifp; 2362 struct ether_header *eh; 2363 int from_blocking; 2364 2365 mbuftrackid(m, 66); 2366 src_if = m->m_pkthdr.rcvif; 2367 ifp = sc->sc_ifp; 2368 2369 ASSERT_IFNET_NOT_SERIALIZED_ALL(ifp); 2370 2371 IFNET_STAT_INC(ifp, ipackets, 1); 2372 IFNET_STAT_INC(ifp, ibytes, m->m_pkthdr.len); 2373 2374 /* 2375 * Look up the bridge_iflist. 2376 */ 2377 bif = bridge_lookup_member_if(sc, src_if); 2378 if (bif == NULL) { 2379 /* Interface is not a bridge member (anymore?) */ 2380 m_freem(m); 2381 return; 2382 } 2383 2384 /* 2385 * In spanning tree mode receiving a packet from an interface 2386 * in a BLOCKING state is allowed, it could be a member of last 2387 * resort from the sender's point of view, but forwarding it is 2388 * not allowed. 2389 * 2390 * The sender's spanning tree will eventually sync up and the 2391 * sender will go into a BLOCKING state too (but this still may be 2392 * an interface of last resort during state changes). 2393 */ 2394 if (bif->bif_flags & IFBIF_STP) { 2395 switch (bif->bif_state) { 2396 case BSTP_IFSTATE_L1BLOCKING: 2397 case BSTP_IFSTATE_LISTENING: 2398 case BSTP_IFSTATE_DISABLED: 2399 m_freem(m); 2400 return; 2401 default: 2402 /* learning, blocking, bonded, forwarding */ 2403 break; 2404 } 2405 from_blocking = (bif->bif_state == BSTP_IFSTATE_BLOCKING); 2406 } else { 2407 from_blocking = 0; 2408 } 2409 2410 eh = mtod(m, struct ether_header *); 2411 2412 /* 2413 * If the interface is learning, and the source 2414 * address is valid and not multicast, record 2415 * the address. 2416 */ 2417 if ((bif->bif_flags & IFBIF_LEARNING) != 0 && 2418 from_blocking == 0 && 2419 ETHER_IS_MULTICAST(eh->ether_shost) == 0 && 2420 (eh->ether_shost[0] == 0 && 2421 eh->ether_shost[1] == 0 && 2422 eh->ether_shost[2] == 0 && 2423 eh->ether_shost[3] == 0 && 2424 eh->ether_shost[4] == 0 && 2425 eh->ether_shost[5] == 0) == 0) { 2426 bridge_rtupdate(sc, eh->ether_shost, src_if, IFBAF_DYNAMIC); 2427 } 2428 2429 /* 2430 * Don't forward from an interface in the listening or learning 2431 * state. That is, in the learning state we learn information 2432 * but we throw away the packets. 2433 * 2434 * We let through packets on interfaces in the blocking state. 2435 * The blocking state is applicable to the send side, not the 2436 * receive side. 2437 */ 2438 if ((bif->bif_flags & IFBIF_STP) != 0 && 2439 (bif->bif_state == BSTP_IFSTATE_LISTENING || 2440 bif->bif_state == BSTP_IFSTATE_LEARNING)) { 2441 m_freem(m); 2442 return; 2443 } 2444 2445 /* 2446 * At this point, the port either doesn't participate 2447 * in spanning tree or it is in the forwarding state. 2448 */ 2449 2450 /* 2451 * If the packet is unicast, destined for someone on 2452 * "this" side of the bridge, drop it. 2453 * 2454 * src_if implies the entire bonding set so we have to compare MAC 2455 * addresses and not just if pointers. 2456 */ 2457 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) { 2458 dst_if = bridge_rtlookup(sc, eh->ether_dhost); 2459 if (dst_if && memcmp(IF_LLADDR(src_if), IF_LLADDR(dst_if), 2460 ETHER_ADDR_LEN) == 0) { 2461 m_freem(m); 2462 return; 2463 } 2464 } else { 2465 /* ...forward it to all interfaces. */ 2466 IFNET_STAT_INC(ifp, imcasts, 1); 2467 dst_if = NULL; 2468 } 2469 2470 /* 2471 * Brodcast if we do not have forwarding information. However, if 2472 * we received the packet on a blocking interface we do not do this 2473 * (unless you really want to blow up your network). 2474 */ 2475 if (dst_if == NULL) { 2476 if (from_blocking) 2477 m_freem(m); 2478 else 2479 bridge_broadcast(sc, src_if, m); 2480 return; 2481 } 2482 2483 dst_if = bridge_select_unicast(sc, dst_if, from_blocking, m); 2484 2485 if (dst_if == NULL) { 2486 m_freem(m); 2487 return; 2488 } 2489 2490 if (inet_pfil_hook.ph_hashooks > 0 2491 #ifdef INET6 2492 || inet6_pfil_hook.ph_hashooks > 0 2493 #endif 2494 ) { 2495 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0) 2496 return; 2497 if (m == NULL) 2498 return; 2499 2500 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0) 2501 return; 2502 if (m == NULL) 2503 return; 2504 } 2505 bridge_handoff(sc, dst_if, m, 0); 2506 } 2507 2508 /* 2509 * bridge_input: 2510 * 2511 * Receive input from a member interface. Queue the packet for 2512 * bridging if it is not for us. 2513 */ 2514 static struct mbuf * 2515 bridge_input(struct ifnet *ifp, struct mbuf *m) 2516 { 2517 struct bridge_softc *sc = ifp->if_bridge; 2518 struct bridge_iflist *bif; 2519 struct ifnet *bifp, *new_ifp; 2520 struct ether_header *eh; 2521 struct mbuf *mc, *mc2; 2522 2523 ASSERT_IFNET_NOT_SERIALIZED_ALL(ifp); 2524 mbuftrackid(m, 67); 2525 2526 /* 2527 * Make sure that we are still a member of a bridge interface. 2528 */ 2529 if (sc == NULL) 2530 return m; 2531 2532 new_ifp = NULL; 2533 bifp = sc->sc_ifp; 2534 2535 if ((bifp->if_flags & IFF_RUNNING) == 0) 2536 goto out; 2537 2538 /* 2539 * Implement support for bridge monitoring. If this flag has been 2540 * set on this interface, discard the packet once we push it through 2541 * the bpf(4) machinery, but before we do, increment various counters 2542 * associated with this bridge. 2543 */ 2544 if (bifp->if_flags & IFF_MONITOR) { 2545 /* Change input interface to this bridge */ 2546 m->m_pkthdr.rcvif = bifp; 2547 2548 BPF_MTAP(bifp, m); 2549 2550 /* Update bridge's ifnet statistics */ 2551 IFNET_STAT_INC(bifp, ipackets, 1); 2552 IFNET_STAT_INC(bifp, ibytes, m->m_pkthdr.len); 2553 if (m->m_flags & (M_MCAST | M_BCAST)) 2554 IFNET_STAT_INC(bifp, imcasts, 1); 2555 2556 m_freem(m); 2557 m = NULL; 2558 goto out; 2559 } 2560 2561 /* 2562 * Handle the ether_header 2563 * 2564 * In all cases if the packet is destined for us via our MAC 2565 * we must clear BRIDGE_MBUF_TAGGED to ensure that we don't 2566 * repeat the source MAC out the same interface. 2567 * 2568 * This first test against our bridge MAC is the fast-path. 2569 * 2570 * NOTE! The bridge interface can serve as an endpoint for 2571 * communication but normally there are no IPs associated 2572 * with it so you cannot route through it. Instead what 2573 * you do is point your default route *THROUGH* the bridge 2574 * to the actual default router for one of the bridged spaces. 2575 * 2576 * Another possibility is to put all your IP specifications 2577 * on the bridge instead of on the individual interfaces. If 2578 * you do this it should be possible to use the bridge as an 2579 * end point and route (rather than switch) through it using 2580 * the default route or ipfw forwarding rules. 2581 */ 2582 2583 /* 2584 * Acquire header 2585 */ 2586 if (m->m_len < ETHER_HDR_LEN) { 2587 m = m_pullup(m, ETHER_HDR_LEN); 2588 if (m == NULL) 2589 goto out; 2590 } 2591 eh = mtod(m, struct ether_header *); 2592 m->m_pkthdr.fw_flags |= BRIDGE_MBUF_TAGGED; 2593 bcopy(eh, &m->m_pkthdr.br.ether, sizeof(*eh)); 2594 2595 if ((bridge_debug & 1) && 2596 (ntohs(eh->ether_type) == ETHERTYPE_ARP || 2597 ntohs(eh->ether_type) == ETHERTYPE_REVARP)) { 2598 kprintf("%02x:%02x:%02x:%02x:%02x:%02x " 2599 "%02x:%02x:%02x:%02x:%02x:%02x type %04x " 2600 "lla %02x:%02x:%02x:%02x:%02x:%02x\n", 2601 eh->ether_dhost[0], 2602 eh->ether_dhost[1], 2603 eh->ether_dhost[2], 2604 eh->ether_dhost[3], 2605 eh->ether_dhost[4], 2606 eh->ether_dhost[5], 2607 eh->ether_shost[0], 2608 eh->ether_shost[1], 2609 eh->ether_shost[2], 2610 eh->ether_shost[3], 2611 eh->ether_shost[4], 2612 eh->ether_shost[5], 2613 eh->ether_type, 2614 ((u_char *)IF_LLADDR(bifp))[0], 2615 ((u_char *)IF_LLADDR(bifp))[1], 2616 ((u_char *)IF_LLADDR(bifp))[2], 2617 ((u_char *)IF_LLADDR(bifp))[3], 2618 ((u_char *)IF_LLADDR(bifp))[4], 2619 ((u_char *)IF_LLADDR(bifp))[5] 2620 ); 2621 } 2622 2623 if (memcmp(eh->ether_dhost, IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) { 2624 /* 2625 * If the packet is for us, set the packets source as the 2626 * bridge, and return the packet back to ifnet.if_input for 2627 * local processing. 2628 */ 2629 m->m_pkthdr.fw_flags &= ~BRIDGE_MBUF_TAGGED; 2630 KASSERT(bifp->if_bridge == NULL, 2631 ("loop created in bridge_input")); 2632 if (pfil_member != 0) { 2633 if (inet_pfil_hook.ph_hashooks > 0 2634 #ifdef INET6 2635 || inet6_pfil_hook.ph_hashooks > 0 2636 #endif 2637 ) { 2638 if (bridge_pfil(&m, NULL, ifp, PFIL_IN) != 0) 2639 goto out; 2640 if (m == NULL) 2641 goto out; 2642 } 2643 } 2644 new_ifp = bifp; 2645 goto out; 2646 } 2647 2648 /* 2649 * Tap all packets arriving on the bridge, no matter if 2650 * they are local destinations or not. In is in. 2651 */ 2652 BPF_MTAP(bifp, m); 2653 2654 bif = bridge_lookup_member_if(sc, ifp); 2655 if (bif == NULL) 2656 goto out; 2657 2658 if (sc->sc_span) 2659 bridge_span(sc, m); 2660 2661 if (m->m_flags & (M_BCAST | M_MCAST)) { 2662 /* 2663 * Tap off 802.1D packets; they do not get forwarded. 2664 */ 2665 if (memcmp(eh->ether_dhost, bstp_etheraddr, 2666 ETHER_ADDR_LEN) == 0) { 2667 ifnet_serialize_all(bifp); 2668 bstp_input(sc, bif, m); 2669 ifnet_deserialize_all(bifp); 2670 2671 /* m is freed by bstp_input */ 2672 m = NULL; 2673 goto out; 2674 } 2675 2676 /* 2677 * Other than 802.11d packets, ignore packets if the 2678 * interface is not in a good state. 2679 * 2680 * NOTE: Broadcast/mcast packets received on a blocking or 2681 * learning interface are allowed for local processing. 2682 * 2683 * The sending side of a blocked port will stop 2684 * transmitting when a better alternative is found. 2685 * However, later on we will disallow the forwarding 2686 * of bcast/mcsat packets over a blocking interface. 2687 */ 2688 if (bif->bif_flags & IFBIF_STP) { 2689 switch (bif->bif_state) { 2690 case BSTP_IFSTATE_L1BLOCKING: 2691 case BSTP_IFSTATE_LISTENING: 2692 case BSTP_IFSTATE_DISABLED: 2693 goto out; 2694 default: 2695 /* blocking, learning, bonded, forwarding */ 2696 break; 2697 } 2698 } 2699 2700 /* 2701 * Make a deep copy of the packet and enqueue the copy 2702 * for bridge processing; return the original packet for 2703 * local processing. 2704 */ 2705 mc = m_dup(m, MB_DONTWAIT); 2706 if (mc == NULL) 2707 goto out; 2708 2709 /* 2710 * It's just too dangerous to allow bcast/mcast over a 2711 * blocked interface, eventually the network will sort 2712 * itself out and a better path will be found. 2713 */ 2714 if ((bif->bif_flags & IFBIF_STP) == 0 || 2715 bif->bif_state != BSTP_IFSTATE_BLOCKING) { 2716 bridge_forward(sc, mc); 2717 } 2718 2719 /* 2720 * Reinject the mbuf as arriving on the bridge so we have a 2721 * chance at claiming multicast packets. We can not loop back 2722 * here from ether_input as a bridge is never a member of a 2723 * bridge. 2724 */ 2725 KASSERT(bifp->if_bridge == NULL, 2726 ("loop created in bridge_input")); 2727 mc2 = m_dup(m, MB_DONTWAIT); 2728 #ifdef notyet 2729 if (mc2 != NULL) { 2730 /* Keep the layer3 header aligned */ 2731 int i = min(mc2->m_pkthdr.len, max_protohdr); 2732 mc2 = m_copyup(mc2, i, ETHER_ALIGN); 2733 } 2734 #endif 2735 if (mc2 != NULL) { 2736 /* 2737 * Don't tap to bpf(4) again; we have already done 2738 * the tapping. 2739 * 2740 * Leave m_pkthdr.rcvif alone, so ARP replies are 2741 * processed as coming in on the correct interface. 2742 * 2743 * Clear the bridge flag for local processing in 2744 * case the packet gets routed. 2745 */ 2746 mc2->m_pkthdr.fw_flags &= ~BRIDGE_MBUF_TAGGED; 2747 ether_reinput_oncpu(bifp, mc2, 0); 2748 } 2749 2750 /* Return the original packet for local processing. */ 2751 goto out; 2752 } 2753 2754 /* 2755 * Input of a unicast packet. We have to allow unicast packets 2756 * input from links in the BLOCKING state as this might be an 2757 * interface of last resort. 2758 * 2759 * NOTE: We explicitly ignore normal packets received on a link 2760 * in the BLOCKING state. The point of being in that state 2761 * is to avoid getting duplicate packets. 2762 * 2763 * HOWEVER, if LINK2 is set the normal spanning tree code 2764 * will mark an interface BLOCKING to avoid multi-cast/broadcast 2765 * loops. Unicast packets CAN still loop if we allow the 2766 * case (hence we only do it in LINK2), but it isn't quite as 2767 * bad as a broadcast packet looping. 2768 */ 2769 if (bif->bif_flags & IFBIF_STP) { 2770 switch (bif->bif_state) { 2771 case BSTP_IFSTATE_L1BLOCKING: 2772 case BSTP_IFSTATE_LISTENING: 2773 case BSTP_IFSTATE_DISABLED: 2774 goto out; 2775 default: 2776 /* blocking, bonded, forwarding, learning */ 2777 break; 2778 } 2779 } 2780 2781 /* 2782 * Unicast. Make sure it's not for us. 2783 * 2784 * This loop is MPSAFE; the only blocking operation (bridge_rtupdate) 2785 * is followed by breaking out of the loop. 2786 */ 2787 TAILQ_FOREACH(bif, &sc->sc_iflists[mycpuid], bif_next) { 2788 if (bif->bif_ifp->if_type != IFT_ETHER) 2789 continue; 2790 2791 /* 2792 * It is destined for an interface linked to the bridge. 2793 * We want the bridge itself to take care of link level 2794 * forwarding to member interfaces so reinput on the bridge. 2795 * i.e. if you ping an IP on a target interface associated 2796 * with the bridge, the arp is-at response should indicate 2797 * the bridge MAC. 2798 * 2799 * Only update our addr list when learning if the port 2800 * is not in a blocking state. If it is we still allow 2801 * the packet but we do not try to learn from it. 2802 */ 2803 if (memcmp(IF_LLADDR(bif->bif_ifp), eh->ether_dhost, 2804 ETHER_ADDR_LEN) == 0) { 2805 if (bif->bif_ifp != ifp) { 2806 /* XXX loop prevention */ 2807 m->m_flags |= M_ETHER_BRIDGED; 2808 } 2809 if ((bif->bif_flags & IFBIF_LEARNING) && 2810 ((bif->bif_flags & IFBIF_STP) == 0 || 2811 bif->bif_state != BSTP_IFSTATE_BLOCKING)) { 2812 bridge_rtupdate(sc, eh->ether_shost, 2813 ifp, IFBAF_DYNAMIC); 2814 } 2815 new_ifp = bifp; /* not bif->bif_ifp */ 2816 m->m_pkthdr.fw_flags &= ~BRIDGE_MBUF_TAGGED; 2817 goto out; 2818 } 2819 2820 /* 2821 * Ignore received packets that were sent by us. 2822 */ 2823 if (memcmp(IF_LLADDR(bif->bif_ifp), eh->ether_shost, 2824 ETHER_ADDR_LEN) == 0) { 2825 m_freem(m); 2826 m = NULL; 2827 goto out; 2828 } 2829 } 2830 2831 /* 2832 * It isn't for us. 2833 * 2834 * Perform the bridge forwarding function, but disallow bridging 2835 * to interfaces in the blocking state if the packet came in on 2836 * an interface in the blocking state. 2837 */ 2838 bridge_forward(sc, m); 2839 m = NULL; 2840 2841 /* 2842 * ether_reinput_oncpu() will reprocess rcvif as 2843 * coming from new_ifp (since we do not specify 2844 * REINPUT_KEEPRCVIF). 2845 */ 2846 out: 2847 if (new_ifp != NULL) { 2848 /* 2849 * Clear the bridge flag for local processing in 2850 * case the packet gets routed. 2851 */ 2852 ether_reinput_oncpu(new_ifp, m, REINPUT_RUNBPF); 2853 m = NULL; 2854 } 2855 return (m); 2856 } 2857 2858 /* 2859 * bridge_start_bcast: 2860 * 2861 * Broadcast the packet sent from bridge to all member 2862 * interfaces. 2863 * This is a simplified version of bridge_broadcast(), however, 2864 * this function expects caller to hold bridge's serializer. 2865 */ 2866 static void 2867 bridge_start_bcast(struct bridge_softc *sc, struct mbuf *m) 2868 { 2869 struct bridge_iflist *bif; 2870 struct mbuf *mc; 2871 struct ifnet *dst_if, *alt_if, *bifp; 2872 int used = 0; 2873 int found = 0; 2874 int alt_priority; 2875 2876 mbuftrackid(m, 68); 2877 bifp = sc->sc_ifp; 2878 ASSERT_IFNET_SERIALIZED_ALL(bifp); 2879 2880 /* 2881 * Following loop is MPSAFE; nothing is blocking 2882 * in the loop body. 2883 * 2884 * NOTE: We transmit through an member in the BLOCKING state only 2885 * as a last resort. 2886 */ 2887 alt_if = NULL; 2888 alt_priority = 0; 2889 2890 TAILQ_FOREACH(bif, &sc->sc_iflists[mycpuid], bif_next) { 2891 dst_if = bif->bif_ifp; 2892 2893 if (bif->bif_flags & IFBIF_STP) { 2894 switch (bif->bif_state) { 2895 case BSTP_IFSTATE_BLOCKING: 2896 if (bif->bif_priority > alt_priority) { 2897 alt_priority = bif->bif_priority; 2898 alt_if = bif->bif_ifp; 2899 } 2900 /* fall through */ 2901 case BSTP_IFSTATE_L1BLOCKING: 2902 case BSTP_IFSTATE_DISABLED: 2903 continue; 2904 default: 2905 /* listening, learning, bonded, forwarding */ 2906 break; 2907 } 2908 } 2909 2910 if ((bif->bif_flags & IFBIF_DISCOVER) == 0 && 2911 (m->m_flags & (M_BCAST|M_MCAST)) == 0) 2912 continue; 2913 2914 if ((dst_if->if_flags & IFF_RUNNING) == 0) 2915 continue; 2916 2917 if (TAILQ_NEXT(bif, bif_next) == NULL) { 2918 mc = m; 2919 used = 1; 2920 } else { 2921 mc = m_copypacket(m, MB_DONTWAIT); 2922 if (mc == NULL) { 2923 IFNET_STAT_INC(bifp, oerrors, 1); 2924 continue; 2925 } 2926 } 2927 found = 1; 2928 bridge_enqueue(dst_if, mc); 2929 } 2930 2931 if (found == 0 && alt_if) { 2932 KKASSERT(used == 0); 2933 mc = m; 2934 used = 1; 2935 bridge_enqueue(alt_if, mc); 2936 } 2937 2938 if (used == 0) 2939 m_freem(m); 2940 } 2941 2942 /* 2943 * bridge_broadcast: 2944 * 2945 * Send a frame to all interfaces that are members of 2946 * the bridge, except for the one on which the packet 2947 * arrived. 2948 */ 2949 static void 2950 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if, 2951 struct mbuf *m) 2952 { 2953 struct bridge_iflist *bif, *nbif; 2954 struct ether_header *eh; 2955 struct mbuf *mc; 2956 struct ifnet *dst_if, *alt_if, *bifp; 2957 int used; 2958 int found; 2959 int alt_priority; 2960 int from_us; 2961 2962 mbuftrackid(m, 69); 2963 bifp = sc->sc_ifp; 2964 ASSERT_IFNET_NOT_SERIALIZED_ALL(bifp); 2965 2966 eh = mtod(m, struct ether_header *); 2967 from_us = bridge_from_us(sc, eh); 2968 2969 if (inet_pfil_hook.ph_hashooks > 0 2970 #ifdef INET6 2971 || inet6_pfil_hook.ph_hashooks > 0 2972 #endif 2973 ) { 2974 if (bridge_pfil(&m, bifp, src_if, PFIL_IN) != 0) 2975 return; 2976 if (m == NULL) 2977 return; 2978 2979 /* Filter on the bridge interface before broadcasting */ 2980 if (bridge_pfil(&m, bifp, NULL, PFIL_OUT) != 0) 2981 return; 2982 if (m == NULL) 2983 return; 2984 } 2985 2986 alt_if = NULL; 2987 alt_priority = 0; 2988 found = 0; 2989 used = 0; 2990 2991 TAILQ_FOREACH_MUTABLE(bif, &sc->sc_iflists[mycpuid], bif_next, nbif) { 2992 dst_if = bif->bif_ifp; 2993 2994 if ((dst_if->if_flags & IFF_RUNNING) == 0) 2995 continue; 2996 2997 /* 2998 * Don't bounce the packet out the same interface it came 2999 * in on. We have to test MAC addresses because a packet 3000 * can come in a bonded interface and we don't want it to 3001 * be echod out the forwarding interface for the same bonding 3002 * set. 3003 */ 3004 if (src_if && memcmp(IF_LLADDR(src_if), IF_LLADDR(dst_if), 3005 ETHER_ADDR_LEN) == 0) { 3006 continue; 3007 } 3008 3009 /* 3010 * Generally speaking we only broadcast through forwarding 3011 * interfaces. If no interfaces are available we select 3012 * a BONDED, BLOCKING, or LEARNING interface to forward 3013 * through. 3014 */ 3015 if (bif->bif_flags & IFBIF_STP) { 3016 switch (bif->bif_state) { 3017 case BSTP_IFSTATE_BONDED: 3018 if (bif->bif_priority + 512 > alt_priority) { 3019 alt_priority = bif->bif_priority + 512; 3020 alt_if = bif->bif_ifp; 3021 } 3022 continue; 3023 case BSTP_IFSTATE_BLOCKING: 3024 if (bif->bif_priority + 256 > alt_priority) { 3025 alt_priority = bif->bif_priority + 256; 3026 alt_if = bif->bif_ifp; 3027 } 3028 continue; 3029 case BSTP_IFSTATE_LEARNING: 3030 if (bif->bif_priority > alt_priority) { 3031 alt_priority = bif->bif_priority; 3032 alt_if = bif->bif_ifp; 3033 } 3034 continue; 3035 case BSTP_IFSTATE_L1BLOCKING: 3036 case BSTP_IFSTATE_DISABLED: 3037 case BSTP_IFSTATE_LISTENING: 3038 continue; 3039 default: 3040 /* forwarding */ 3041 break; 3042 } 3043 } 3044 3045 if ((bif->bif_flags & IFBIF_DISCOVER) == 0 && 3046 (m->m_flags & (M_BCAST|M_MCAST)) == 0) { 3047 continue; 3048 } 3049 3050 if (TAILQ_NEXT(bif, bif_next) == NULL) { 3051 mc = m; 3052 used = 1; 3053 } else { 3054 mc = m_copypacket(m, MB_DONTWAIT); 3055 if (mc == NULL) { 3056 IFNET_STAT_INC(sc->sc_ifp, oerrors, 1); 3057 continue; 3058 } 3059 } 3060 found = 1; 3061 3062 /* 3063 * Filter on the output interface. Pass a NULL bridge 3064 * interface pointer so we do not redundantly filter on 3065 * the bridge for each interface we broadcast on. 3066 */ 3067 if (inet_pfil_hook.ph_hashooks > 0 3068 #ifdef INET6 3069 || inet6_pfil_hook.ph_hashooks > 0 3070 #endif 3071 ) { 3072 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0) 3073 continue; 3074 if (mc == NULL) 3075 continue; 3076 } 3077 bridge_handoff(sc, dst_if, mc, from_us); 3078 3079 if (nbif != NULL && !nbif->bif_onlist) { 3080 KKASSERT(bif->bif_onlist); 3081 nbif = TAILQ_NEXT(bif, bif_next); 3082 } 3083 } 3084 3085 if (found == 0 && alt_if) { 3086 KKASSERT(used == 0); 3087 mc = m; 3088 used = 1; 3089 bridge_enqueue(alt_if, mc); 3090 } 3091 3092 if (used == 0) 3093 m_freem(m); 3094 } 3095 3096 /* 3097 * bridge_span: 3098 * 3099 * Duplicate a packet out one or more interfaces that are in span mode, 3100 * the original mbuf is unmodified. 3101 */ 3102 static void 3103 bridge_span(struct bridge_softc *sc, struct mbuf *m) 3104 { 3105 struct bridge_iflist *bif; 3106 struct ifnet *dst_if, *bifp; 3107 struct mbuf *mc; 3108 3109 mbuftrackid(m, 70); 3110 bifp = sc->sc_ifp; 3111 ifnet_serialize_all(bifp); 3112 3113 TAILQ_FOREACH(bif, &sc->sc_spanlist, bif_next) { 3114 dst_if = bif->bif_ifp; 3115 3116 if ((dst_if->if_flags & IFF_RUNNING) == 0) 3117 continue; 3118 3119 mc = m_copypacket(m, MB_DONTWAIT); 3120 if (mc == NULL) { 3121 IFNET_STAT_INC(sc->sc_ifp, oerrors, 1); 3122 continue; 3123 } 3124 bridge_enqueue(dst_if, mc); 3125 } 3126 3127 ifnet_deserialize_all(bifp); 3128 } 3129 3130 static void 3131 bridge_rtmsg_sync_handler(netmsg_t msg) 3132 { 3133 ifnet_forwardmsg(&msg->lmsg, mycpuid + 1); 3134 } 3135 3136 static void 3137 bridge_rtmsg_sync(struct bridge_softc *sc) 3138 { 3139 struct netmsg_base msg; 3140 3141 ASSERT_IFNET_NOT_SERIALIZED_ALL(sc->sc_ifp); 3142 3143 netmsg_init(&msg, NULL, &curthread->td_msgport, 3144 0, bridge_rtmsg_sync_handler); 3145 ifnet_domsg(&msg.lmsg, 0); 3146 } 3147 3148 static __inline void 3149 bridge_rtinfo_update(struct bridge_rtinfo *bri, struct ifnet *dst_if, 3150 int setflags, uint8_t flags, uint32_t timeo) 3151 { 3152 if ((bri->bri_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 3153 bri->bri_ifp != dst_if) 3154 bri->bri_ifp = dst_if; 3155 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 3156 bri->bri_expire != time_second + timeo) 3157 bri->bri_expire = time_second + timeo; 3158 if (setflags) 3159 bri->bri_flags = flags; 3160 } 3161 3162 static int 3163 bridge_rtinstall_oncpu(struct bridge_softc *sc, const uint8_t *dst, 3164 struct ifnet *dst_if, int setflags, uint8_t flags, 3165 struct bridge_rtinfo **bri0) 3166 { 3167 struct bridge_rtnode *brt; 3168 struct bridge_rtinfo *bri; 3169 3170 if (mycpuid == 0) { 3171 brt = bridge_rtnode_lookup(sc, dst); 3172 if (brt != NULL) { 3173 /* 3174 * rtnode for 'dst' already exists. We inform the 3175 * caller about this by leaving bri0 as NULL. The 3176 * caller will terminate the intallation upon getting 3177 * NULL bri0. However, we still need to update the 3178 * rtinfo. 3179 */ 3180 KKASSERT(*bri0 == NULL); 3181 3182 /* Update rtinfo */ 3183 bridge_rtinfo_update(brt->brt_info, dst_if, setflags, 3184 flags, sc->sc_brttimeout); 3185 return 0; 3186 } 3187 3188 /* 3189 * We only need to check brtcnt on CPU0, since if limit 3190 * is to be exceeded, ENOSPC is returned. Caller knows 3191 * this and will terminate the installation. 3192 */ 3193 if (sc->sc_brtcnt >= sc->sc_brtmax) 3194 return ENOSPC; 3195 3196 KKASSERT(*bri0 == NULL); 3197 bri = kmalloc(sizeof(struct bridge_rtinfo), M_DEVBUF, 3198 M_WAITOK | M_ZERO); 3199 *bri0 = bri; 3200 3201 /* Setup rtinfo */ 3202 bri->bri_flags = IFBAF_DYNAMIC; 3203 bridge_rtinfo_update(bri, dst_if, setflags, flags, 3204 sc->sc_brttimeout); 3205 } else { 3206 bri = *bri0; 3207 KKASSERT(bri != NULL); 3208 } 3209 3210 brt = kmalloc(sizeof(struct bridge_rtnode), M_DEVBUF, 3211 M_WAITOK | M_ZERO); 3212 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN); 3213 brt->brt_info = bri; 3214 3215 bridge_rtnode_insert(sc, brt); 3216 return 0; 3217 } 3218 3219 static void 3220 bridge_rtinstall_handler(netmsg_t msg) 3221 { 3222 struct netmsg_brsaddr *brmsg = (struct netmsg_brsaddr *)msg; 3223 int error; 3224 3225 error = bridge_rtinstall_oncpu(brmsg->br_softc, 3226 brmsg->br_dst, brmsg->br_dst_if, 3227 brmsg->br_setflags, brmsg->br_flags, 3228 &brmsg->br_rtinfo); 3229 if (error) { 3230 KKASSERT(mycpuid == 0 && brmsg->br_rtinfo == NULL); 3231 lwkt_replymsg(&brmsg->base.lmsg, error); 3232 return; 3233 } else if (brmsg->br_rtinfo == NULL) { 3234 /* rtnode already exists for 'dst' */ 3235 KKASSERT(mycpuid == 0); 3236 lwkt_replymsg(&brmsg->base.lmsg, 0); 3237 return; 3238 } 3239 ifnet_forwardmsg(&brmsg->base.lmsg, mycpuid + 1); 3240 } 3241 3242 /* 3243 * bridge_rtupdate: 3244 * 3245 * Add/Update a bridge routing entry. 3246 */ 3247 static int 3248 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, 3249 struct ifnet *dst_if, uint8_t flags) 3250 { 3251 struct bridge_rtnode *brt; 3252 3253 /* 3254 * A route for this destination might already exist. If so, 3255 * update it, otherwise create a new one. 3256 */ 3257 if ((brt = bridge_rtnode_lookup(sc, dst)) == NULL) { 3258 struct netmsg_brsaddr *brmsg; 3259 3260 if (sc->sc_brtcnt >= sc->sc_brtmax) 3261 return ENOSPC; 3262 3263 brmsg = kmalloc(sizeof(*brmsg), M_LWKTMSG, M_WAITOK | M_NULLOK); 3264 if (brmsg == NULL) 3265 return ENOMEM; 3266 3267 netmsg_init(&brmsg->base, NULL, &netisr_afree_rport, 3268 0, bridge_rtinstall_handler); 3269 memcpy(brmsg->br_dst, dst, ETHER_ADDR_LEN); 3270 brmsg->br_dst_if = dst_if; 3271 brmsg->br_flags = flags; 3272 brmsg->br_setflags = 0; 3273 brmsg->br_softc = sc; 3274 brmsg->br_rtinfo = NULL; 3275 3276 ifnet_sendmsg(&brmsg->base.lmsg, 0); 3277 return 0; 3278 } 3279 bridge_rtinfo_update(brt->brt_info, dst_if, 0, flags, 3280 sc->sc_brttimeout); 3281 return 0; 3282 } 3283 3284 static int 3285 bridge_rtsaddr(struct bridge_softc *sc, const uint8_t *dst, 3286 struct ifnet *dst_if, uint8_t flags) 3287 { 3288 struct netmsg_brsaddr brmsg; 3289 3290 ASSERT_IFNET_NOT_SERIALIZED_ALL(sc->sc_ifp); 3291 3292 netmsg_init(&brmsg.base, NULL, &curthread->td_msgport, 3293 0, bridge_rtinstall_handler); 3294 memcpy(brmsg.br_dst, dst, ETHER_ADDR_LEN); 3295 brmsg.br_dst_if = dst_if; 3296 brmsg.br_flags = flags; 3297 brmsg.br_setflags = 1; 3298 brmsg.br_softc = sc; 3299 brmsg.br_rtinfo = NULL; 3300 3301 return ifnet_domsg(&brmsg.base.lmsg, 0); 3302 } 3303 3304 /* 3305 * bridge_rtlookup: 3306 * 3307 * Lookup the destination interface for an address. 3308 */ 3309 static struct ifnet * 3310 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr) 3311 { 3312 struct bridge_rtnode *brt; 3313 3314 if ((brt = bridge_rtnode_lookup(sc, addr)) == NULL) 3315 return NULL; 3316 return brt->brt_info->bri_ifp; 3317 } 3318 3319 static void 3320 bridge_rtreap_handler(netmsg_t msg) 3321 { 3322 struct bridge_softc *sc = msg->lmsg.u.ms_resultp; 3323 struct bridge_rtnode *brt, *nbrt; 3324 3325 LIST_FOREACH_MUTABLE(brt, &sc->sc_rtlists[mycpuid], brt_list, nbrt) { 3326 if (brt->brt_info->bri_dead) 3327 bridge_rtnode_destroy(sc, brt); 3328 } 3329 ifnet_forwardmsg(&msg->lmsg, mycpuid + 1); 3330 } 3331 3332 static void 3333 bridge_rtreap(struct bridge_softc *sc) 3334 { 3335 struct netmsg_base msg; 3336 3337 ASSERT_IFNET_NOT_SERIALIZED_ALL(sc->sc_ifp); 3338 3339 netmsg_init(&msg, NULL, &curthread->td_msgport, 3340 0, bridge_rtreap_handler); 3341 msg.lmsg.u.ms_resultp = sc; 3342 3343 ifnet_domsg(&msg.lmsg, 0); 3344 } 3345 3346 static void 3347 bridge_rtreap_async(struct bridge_softc *sc) 3348 { 3349 struct netmsg_base *msg; 3350 3351 msg = kmalloc(sizeof(*msg), M_LWKTMSG, M_WAITOK); 3352 3353 netmsg_init(msg, NULL, &netisr_afree_rport, 3354 0, bridge_rtreap_handler); 3355 msg->lmsg.u.ms_resultp = sc; 3356 3357 ifnet_sendmsg(&msg->lmsg, 0); 3358 } 3359 3360 /* 3361 * bridge_rttrim: 3362 * 3363 * Trim the routine table so that we have a number 3364 * of routing entries less than or equal to the 3365 * maximum number. 3366 */ 3367 static void 3368 bridge_rttrim(struct bridge_softc *sc) 3369 { 3370 struct bridge_rtnode *brt; 3371 int dead; 3372 3373 ASSERT_IFNET_NOT_SERIALIZED_ALL(sc->sc_ifp); 3374 3375 /* Make sure we actually need to do this. */ 3376 if (sc->sc_brtcnt <= sc->sc_brtmax) 3377 return; 3378 3379 /* 3380 * Find out how many rtnodes are dead 3381 */ 3382 dead = bridge_rtage_finddead(sc); 3383 KKASSERT(dead <= sc->sc_brtcnt); 3384 3385 if (sc->sc_brtcnt - dead <= sc->sc_brtmax) { 3386 /* Enough dead rtnodes are found */ 3387 bridge_rtreap(sc); 3388 return; 3389 } 3390 3391 /* 3392 * Kill some dynamic rtnodes to meet the brtmax 3393 */ 3394 LIST_FOREACH(brt, &sc->sc_rtlists[mycpuid], brt_list) { 3395 struct bridge_rtinfo *bri = brt->brt_info; 3396 3397 if (bri->bri_dead) { 3398 /* 3399 * We have counted this rtnode in 3400 * bridge_rtage_finddead() 3401 */ 3402 continue; 3403 } 3404 3405 if ((bri->bri_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { 3406 bri->bri_dead = 1; 3407 ++dead; 3408 KKASSERT(dead <= sc->sc_brtcnt); 3409 3410 if (sc->sc_brtcnt - dead <= sc->sc_brtmax) { 3411 /* Enough rtnodes are collected */ 3412 break; 3413 } 3414 } 3415 } 3416 if (dead) 3417 bridge_rtreap(sc); 3418 } 3419 3420 /* 3421 * bridge_timer: 3422 * 3423 * Aging timer for the bridge. 3424 */ 3425 static void 3426 bridge_timer(void *arg) 3427 { 3428 struct bridge_softc *sc = arg; 3429 struct netmsg_base *msg; 3430 3431 KKASSERT(mycpuid == BRIDGE_CFGCPU); 3432 3433 crit_enter(); 3434 3435 if (callout_pending(&sc->sc_brcallout) || 3436 !callout_active(&sc->sc_brcallout)) { 3437 crit_exit(); 3438 return; 3439 } 3440 callout_deactivate(&sc->sc_brcallout); 3441 3442 msg = &sc->sc_brtimemsg; 3443 KKASSERT(msg->lmsg.ms_flags & MSGF_DONE); 3444 lwkt_sendmsg(BRIDGE_CFGPORT, &msg->lmsg); 3445 3446 crit_exit(); 3447 } 3448 3449 static void 3450 bridge_timer_handler(netmsg_t msg) 3451 { 3452 struct bridge_softc *sc = msg->lmsg.u.ms_resultp; 3453 3454 KKASSERT(&curthread->td_msgport == BRIDGE_CFGPORT); 3455 3456 crit_enter(); 3457 /* Reply ASAP */ 3458 lwkt_replymsg(&msg->lmsg, 0); 3459 crit_exit(); 3460 3461 bridge_rtage(sc); 3462 if (sc->sc_ifp->if_flags & IFF_RUNNING) { 3463 callout_reset(&sc->sc_brcallout, 3464 bridge_rtable_prune_period * hz, bridge_timer, sc); 3465 } 3466 } 3467 3468 static int 3469 bridge_rtage_finddead(struct bridge_softc *sc) 3470 { 3471 struct bridge_rtnode *brt; 3472 int dead = 0; 3473 3474 LIST_FOREACH(brt, &sc->sc_rtlists[mycpuid], brt_list) { 3475 struct bridge_rtinfo *bri = brt->brt_info; 3476 3477 if ((bri->bri_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && 3478 time_second >= bri->bri_expire) { 3479 bri->bri_dead = 1; 3480 ++dead; 3481 KKASSERT(dead <= sc->sc_brtcnt); 3482 } 3483 } 3484 return dead; 3485 } 3486 3487 /* 3488 * bridge_rtage: 3489 * 3490 * Perform an aging cycle. 3491 */ 3492 static void 3493 bridge_rtage(struct bridge_softc *sc) 3494 { 3495 ASSERT_IFNET_NOT_SERIALIZED_ALL(sc->sc_ifp); 3496 3497 if (bridge_rtage_finddead(sc)) 3498 bridge_rtreap(sc); 3499 } 3500 3501 /* 3502 * bridge_rtflush: 3503 * 3504 * Remove all dynamic addresses from the bridge. 3505 */ 3506 static void 3507 bridge_rtflush(struct bridge_softc *sc, int bf) 3508 { 3509 struct bridge_rtnode *brt; 3510 int reap; 3511 3512 reap = 0; 3513 LIST_FOREACH(brt, &sc->sc_rtlists[mycpuid], brt_list) { 3514 struct bridge_rtinfo *bri = brt->brt_info; 3515 3516 if ((bf & IFBF_FLUSHALL) || 3517 (bri->bri_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { 3518 bri->bri_dead = 1; 3519 reap = 1; 3520 } 3521 } 3522 if (reap) { 3523 if (bf & IFBF_FLUSHSYNC) 3524 bridge_rtreap(sc); 3525 else 3526 bridge_rtreap_async(sc); 3527 } 3528 } 3529 3530 /* 3531 * bridge_rtdaddr: 3532 * 3533 * Remove an address from the table. 3534 */ 3535 static int 3536 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr) 3537 { 3538 struct bridge_rtnode *brt; 3539 3540 ASSERT_IFNET_NOT_SERIALIZED_ALL(sc->sc_ifp); 3541 3542 if ((brt = bridge_rtnode_lookup(sc, addr)) == NULL) 3543 return (ENOENT); 3544 3545 /* TODO: add a cheaper delete operation */ 3546 brt->brt_info->bri_dead = 1; 3547 bridge_rtreap(sc); 3548 return (0); 3549 } 3550 3551 /* 3552 * bridge_rtdelete: 3553 * 3554 * Delete routes to a speicifc member interface. 3555 */ 3556 void 3557 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int bf) 3558 { 3559 struct bridge_rtnode *brt; 3560 int reap; 3561 3562 reap = 0; 3563 LIST_FOREACH(brt, &sc->sc_rtlists[mycpuid], brt_list) { 3564 struct bridge_rtinfo *bri = brt->brt_info; 3565 3566 if (bri->bri_ifp == ifp && 3567 ((bf & IFBF_FLUSHALL) || 3568 (bri->bri_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)) { 3569 bri->bri_dead = 1; 3570 reap = 1; 3571 } 3572 } 3573 if (reap) { 3574 if (bf & IFBF_FLUSHSYNC) 3575 bridge_rtreap(sc); 3576 else 3577 bridge_rtreap_async(sc); 3578 } 3579 } 3580 3581 /* 3582 * bridge_rtable_init: 3583 * 3584 * Initialize the route table for this bridge. 3585 */ 3586 static void 3587 bridge_rtable_init(struct bridge_softc *sc) 3588 { 3589 int cpu; 3590 3591 /* 3592 * Initialize per-cpu hash tables 3593 */ 3594 sc->sc_rthashs = kmalloc(sizeof(*sc->sc_rthashs) * ncpus, 3595 M_DEVBUF, M_WAITOK); 3596 for (cpu = 0; cpu < ncpus; ++cpu) { 3597 int i; 3598 3599 sc->sc_rthashs[cpu] = 3600 kmalloc(sizeof(struct bridge_rtnode_head) * BRIDGE_RTHASH_SIZE, 3601 M_DEVBUF, M_WAITOK); 3602 3603 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++) 3604 LIST_INIT(&sc->sc_rthashs[cpu][i]); 3605 } 3606 sc->sc_rthash_key = karc4random(); 3607 3608 /* 3609 * Initialize per-cpu lists 3610 */ 3611 sc->sc_rtlists = kmalloc(sizeof(struct bridge_rtnode_head) * ncpus, 3612 M_DEVBUF, M_WAITOK); 3613 for (cpu = 0; cpu < ncpus; ++cpu) 3614 LIST_INIT(&sc->sc_rtlists[cpu]); 3615 } 3616 3617 /* 3618 * bridge_rtable_fini: 3619 * 3620 * Deconstruct the route table for this bridge. 3621 */ 3622 static void 3623 bridge_rtable_fini(struct bridge_softc *sc) 3624 { 3625 int cpu; 3626 3627 /* 3628 * Free per-cpu hash tables 3629 */ 3630 for (cpu = 0; cpu < ncpus; ++cpu) 3631 kfree(sc->sc_rthashs[cpu], M_DEVBUF); 3632 kfree(sc->sc_rthashs, M_DEVBUF); 3633 3634 /* 3635 * Free per-cpu lists 3636 */ 3637 kfree(sc->sc_rtlists, M_DEVBUF); 3638 } 3639 3640 /* 3641 * The following hash function is adapted from "Hash Functions" by Bob Jenkins 3642 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997). 3643 */ 3644 #define mix(a, b, c) \ 3645 do { \ 3646 a -= b; a -= c; a ^= (c >> 13); \ 3647 b -= c; b -= a; b ^= (a << 8); \ 3648 c -= a; c -= b; c ^= (b >> 13); \ 3649 a -= b; a -= c; a ^= (c >> 12); \ 3650 b -= c; b -= a; b ^= (a << 16); \ 3651 c -= a; c -= b; c ^= (b >> 5); \ 3652 a -= b; a -= c; a ^= (c >> 3); \ 3653 b -= c; b -= a; b ^= (a << 10); \ 3654 c -= a; c -= b; c ^= (b >> 15); \ 3655 } while (/*CONSTCOND*/0) 3656 3657 static __inline uint32_t 3658 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr) 3659 { 3660 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key; 3661 3662 b += addr[5] << 8; 3663 b += addr[4]; 3664 a += addr[3] << 24; 3665 a += addr[2] << 16; 3666 a += addr[1] << 8; 3667 a += addr[0]; 3668 3669 mix(a, b, c); 3670 3671 return (c & BRIDGE_RTHASH_MASK); 3672 } 3673 3674 #undef mix 3675 3676 static int 3677 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b) 3678 { 3679 int i, d; 3680 3681 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) { 3682 d = ((int)a[i]) - ((int)b[i]); 3683 } 3684 3685 return (d); 3686 } 3687 3688 /* 3689 * bridge_rtnode_lookup: 3690 * 3691 * Look up a bridge route node for the specified destination. 3692 */ 3693 static struct bridge_rtnode * 3694 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr) 3695 { 3696 struct bridge_rtnode *brt; 3697 uint32_t hash; 3698 int dir; 3699 3700 hash = bridge_rthash(sc, addr); 3701 LIST_FOREACH(brt, &sc->sc_rthashs[mycpuid][hash], brt_hash) { 3702 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr); 3703 if (dir == 0) 3704 return (brt); 3705 if (dir > 0) 3706 return (NULL); 3707 } 3708 3709 return (NULL); 3710 } 3711 3712 /* 3713 * bridge_rtnode_insert: 3714 * 3715 * Insert the specified bridge node into the route table. 3716 * Caller has to make sure that rtnode does not exist. 3717 */ 3718 static void 3719 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt) 3720 { 3721 struct bridge_rtnode *lbrt; 3722 uint32_t hash; 3723 int dir; 3724 3725 hash = bridge_rthash(sc, brt->brt_addr); 3726 3727 lbrt = LIST_FIRST(&sc->sc_rthashs[mycpuid][hash]); 3728 if (lbrt == NULL) { 3729 LIST_INSERT_HEAD(&sc->sc_rthashs[mycpuid][hash], 3730 brt, brt_hash); 3731 goto out; 3732 } 3733 3734 do { 3735 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr); 3736 KASSERT(dir != 0, ("rtnode already exist")); 3737 3738 if (dir > 0) { 3739 LIST_INSERT_BEFORE(lbrt, brt, brt_hash); 3740 goto out; 3741 } 3742 if (LIST_NEXT(lbrt, brt_hash) == NULL) { 3743 LIST_INSERT_AFTER(lbrt, brt, brt_hash); 3744 goto out; 3745 } 3746 lbrt = LIST_NEXT(lbrt, brt_hash); 3747 } while (lbrt != NULL); 3748 3749 panic("no suitable position found for rtnode"); 3750 out: 3751 LIST_INSERT_HEAD(&sc->sc_rtlists[mycpuid], brt, brt_list); 3752 if (mycpuid == 0) { 3753 /* 3754 * Update the brtcnt. 3755 * We only need to do it once and we do it on CPU0. 3756 */ 3757 sc->sc_brtcnt++; 3758 } 3759 } 3760 3761 /* 3762 * bridge_rtnode_destroy: 3763 * 3764 * Destroy a bridge rtnode. 3765 */ 3766 static void 3767 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt) 3768 { 3769 LIST_REMOVE(brt, brt_hash); 3770 LIST_REMOVE(brt, brt_list); 3771 3772 if (mycpuid + 1 == ncpus) { 3773 /* Free rtinfo associated with rtnode on the last cpu */ 3774 kfree(brt->brt_info, M_DEVBUF); 3775 } 3776 kfree(brt, M_DEVBUF); 3777 3778 if (mycpuid == 0) { 3779 /* Update brtcnt only on CPU0 */ 3780 sc->sc_brtcnt--; 3781 } 3782 } 3783 3784 static __inline int 3785 bridge_post_pfil(struct mbuf *m) 3786 { 3787 if (m->m_pkthdr.fw_flags & IPFORWARD_MBUF_TAGGED) 3788 return EOPNOTSUPP; 3789 3790 /* Not yet */ 3791 if (m->m_pkthdr.fw_flags & DUMMYNET_MBUF_TAGGED) 3792 return EOPNOTSUPP; 3793 3794 return 0; 3795 } 3796 3797 /* 3798 * Send bridge packets through pfil if they are one of the types pfil can deal 3799 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without 3800 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for 3801 * that interface. 3802 */ 3803 static int 3804 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir) 3805 { 3806 int snap, error, i, hlen; 3807 struct ether_header *eh1, eh2; 3808 struct ip *ip; 3809 struct llc llc1; 3810 u_int16_t ether_type; 3811 3812 snap = 0; 3813 error = -1; /* Default error if not error == 0 */ 3814 3815 if (pfil_bridge == 0 && pfil_member == 0) 3816 return (0); /* filtering is disabled */ 3817 3818 i = min((*mp)->m_pkthdr.len, max_protohdr); 3819 if ((*mp)->m_len < i) { 3820 *mp = m_pullup(*mp, i); 3821 if (*mp == NULL) { 3822 kprintf("%s: m_pullup failed\n", __func__); 3823 return (-1); 3824 } 3825 } 3826 3827 eh1 = mtod(*mp, struct ether_header *); 3828 ether_type = ntohs(eh1->ether_type); 3829 3830 /* 3831 * Check for SNAP/LLC. 3832 */ 3833 if (ether_type < ETHERMTU) { 3834 struct llc *llc2 = (struct llc *)(eh1 + 1); 3835 3836 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 && 3837 llc2->llc_dsap == LLC_SNAP_LSAP && 3838 llc2->llc_ssap == LLC_SNAP_LSAP && 3839 llc2->llc_control == LLC_UI) { 3840 ether_type = htons(llc2->llc_un.type_snap.ether_type); 3841 snap = 1; 3842 } 3843 } 3844 3845 /* 3846 * If we're trying to filter bridge traffic, don't look at anything 3847 * other than IP and ARP traffic. If the filter doesn't understand 3848 * IPv6, don't allow IPv6 through the bridge either. This is lame 3849 * since if we really wanted, say, an AppleTalk filter, we are hosed, 3850 * but of course we don't have an AppleTalk filter to begin with. 3851 * (Note that since pfil doesn't understand ARP it will pass *ALL* 3852 * ARP traffic.) 3853 */ 3854 switch (ether_type) { 3855 case ETHERTYPE_ARP: 3856 case ETHERTYPE_REVARP: 3857 return (0); /* Automatically pass */ 3858 3859 case ETHERTYPE_IP: 3860 #ifdef INET6 3861 case ETHERTYPE_IPV6: 3862 #endif /* INET6 */ 3863 break; 3864 3865 default: 3866 /* 3867 * Check to see if the user wants to pass non-ip 3868 * packets, these will not be checked by pfil(9) 3869 * and passed unconditionally so the default is to drop. 3870 */ 3871 if (pfil_onlyip) 3872 goto bad; 3873 } 3874 3875 /* Strip off the Ethernet header and keep a copy. */ 3876 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2); 3877 m_adj(*mp, ETHER_HDR_LEN); 3878 3879 /* Strip off snap header, if present */ 3880 if (snap) { 3881 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1); 3882 m_adj(*mp, sizeof(struct llc)); 3883 } 3884 3885 /* 3886 * Check the IP header for alignment and errors 3887 */ 3888 if (dir == PFIL_IN) { 3889 switch (ether_type) { 3890 case ETHERTYPE_IP: 3891 error = bridge_ip_checkbasic(mp); 3892 break; 3893 #ifdef INET6 3894 case ETHERTYPE_IPV6: 3895 error = bridge_ip6_checkbasic(mp); 3896 break; 3897 #endif /* INET6 */ 3898 default: 3899 error = 0; 3900 } 3901 if (error) 3902 goto bad; 3903 } 3904 3905 error = 0; 3906 3907 /* 3908 * Run the packet through pfil 3909 */ 3910 switch (ether_type) { 3911 case ETHERTYPE_IP: 3912 /* 3913 * before calling the firewall, swap fields the same as 3914 * IP does. here we assume the header is contiguous 3915 */ 3916 ip = mtod(*mp, struct ip *); 3917 3918 ip->ip_len = ntohs(ip->ip_len); 3919 ip->ip_off = ntohs(ip->ip_off); 3920 3921 /* 3922 * Run pfil on the member interface and the bridge, both can 3923 * be skipped by clearing pfil_member or pfil_bridge. 3924 * 3925 * Keep the order: 3926 * in_if -> bridge_if -> out_if 3927 */ 3928 if (pfil_bridge && dir == PFIL_OUT && bifp != NULL) { 3929 error = pfil_run_hooks(&inet_pfil_hook, mp, bifp, dir); 3930 if (*mp == NULL || error != 0) /* filter may consume */ 3931 break; 3932 error = bridge_post_pfil(*mp); 3933 if (error) 3934 break; 3935 } 3936 3937 if (pfil_member && ifp != NULL) { 3938 error = pfil_run_hooks(&inet_pfil_hook, mp, ifp, dir); 3939 if (*mp == NULL || error != 0) /* filter may consume */ 3940 break; 3941 error = bridge_post_pfil(*mp); 3942 if (error) 3943 break; 3944 } 3945 3946 if (pfil_bridge && dir == PFIL_IN && bifp != NULL) { 3947 error = pfil_run_hooks(&inet_pfil_hook, mp, bifp, dir); 3948 if (*mp == NULL || error != 0) /* filter may consume */ 3949 break; 3950 error = bridge_post_pfil(*mp); 3951 if (error) 3952 break; 3953 } 3954 3955 /* check if we need to fragment the packet */ 3956 if (pfil_member && ifp != NULL && dir == PFIL_OUT) { 3957 i = (*mp)->m_pkthdr.len; 3958 if (i > ifp->if_mtu) { 3959 error = bridge_fragment(ifp, *mp, &eh2, snap, 3960 &llc1); 3961 return (error); 3962 } 3963 } 3964 3965 /* Recalculate the ip checksum and restore byte ordering */ 3966 ip = mtod(*mp, struct ip *); 3967 hlen = ip->ip_hl << 2; 3968 if (hlen < sizeof(struct ip)) 3969 goto bad; 3970 if (hlen > (*mp)->m_len) { 3971 if ((*mp = m_pullup(*mp, hlen)) == NULL) 3972 goto bad; 3973 ip = mtod(*mp, struct ip *); 3974 if (ip == NULL) 3975 goto bad; 3976 } 3977 ip->ip_len = htons(ip->ip_len); 3978 ip->ip_off = htons(ip->ip_off); 3979 ip->ip_sum = 0; 3980 if (hlen == sizeof(struct ip)) 3981 ip->ip_sum = in_cksum_hdr(ip); 3982 else 3983 ip->ip_sum = in_cksum(*mp, hlen); 3984 3985 break; 3986 #ifdef INET6 3987 case ETHERTYPE_IPV6: 3988 if (pfil_bridge && dir == PFIL_OUT && bifp != NULL) 3989 error = pfil_run_hooks(&inet6_pfil_hook, mp, bifp, 3990 dir); 3991 3992 if (*mp == NULL || error != 0) /* filter may consume */ 3993 break; 3994 3995 if (pfil_member && ifp != NULL) 3996 error = pfil_run_hooks(&inet6_pfil_hook, mp, ifp, 3997 dir); 3998 3999 if (*mp == NULL || error != 0) /* filter may consume */ 4000 break; 4001 4002 if (pfil_bridge && dir == PFIL_IN && bifp != NULL) 4003 error = pfil_run_hooks(&inet6_pfil_hook, mp, bifp, 4004 dir); 4005 break; 4006 #endif 4007 default: 4008 error = 0; 4009 break; 4010 } 4011 4012 if (*mp == NULL) 4013 return (error); 4014 if (error != 0) 4015 goto bad; 4016 4017 error = -1; 4018 4019 /* 4020 * Finally, put everything back the way it was and return 4021 */ 4022 if (snap) { 4023 M_PREPEND(*mp, sizeof(struct llc), MB_DONTWAIT); 4024 if (*mp == NULL) 4025 return (error); 4026 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc)); 4027 } 4028 4029 M_PREPEND(*mp, ETHER_HDR_LEN, MB_DONTWAIT); 4030 if (*mp == NULL) 4031 return (error); 4032 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN); 4033 4034 return (0); 4035 4036 bad: 4037 m_freem(*mp); 4038 *mp = NULL; 4039 return (error); 4040 } 4041 4042 /* 4043 * Perform basic checks on header size since 4044 * pfil assumes ip_input has already processed 4045 * it for it. Cut-and-pasted from ip_input.c. 4046 * Given how simple the IPv6 version is, 4047 * does the IPv4 version really need to be 4048 * this complicated? 4049 * 4050 * XXX Should we update ipstat here, or not? 4051 * XXX Right now we update ipstat but not 4052 * XXX csum_counter. 4053 */ 4054 static int 4055 bridge_ip_checkbasic(struct mbuf **mp) 4056 { 4057 struct mbuf *m = *mp; 4058 struct ip *ip; 4059 int len, hlen; 4060 u_short sum; 4061 4062 if (*mp == NULL) 4063 return (-1); 4064 #if 0 /* notyet */ 4065 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 4066 if ((m = m_copyup(m, sizeof(struct ip), 4067 (max_linkhdr + 3) & ~3)) == NULL) { 4068 /* XXXJRT new stat, please */ 4069 ipstat.ips_toosmall++; 4070 goto bad; 4071 } 4072 } else 4073 #endif 4074 #ifndef __predict_false 4075 #define __predict_false(x) x 4076 #endif 4077 if (__predict_false(m->m_len < sizeof (struct ip))) { 4078 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 4079 ipstat.ips_toosmall++; 4080 goto bad; 4081 } 4082 } 4083 ip = mtod(m, struct ip *); 4084 if (ip == NULL) goto bad; 4085 4086 if (ip->ip_v != IPVERSION) { 4087 ipstat.ips_badvers++; 4088 goto bad; 4089 } 4090 hlen = ip->ip_hl << 2; 4091 if (hlen < sizeof(struct ip)) { /* minimum header length */ 4092 ipstat.ips_badhlen++; 4093 goto bad; 4094 } 4095 if (hlen > m->m_len) { 4096 if ((m = m_pullup(m, hlen)) == NULL) { 4097 ipstat.ips_badhlen++; 4098 goto bad; 4099 } 4100 ip = mtod(m, struct ip *); 4101 if (ip == NULL) goto bad; 4102 } 4103 4104 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) { 4105 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID); 4106 } else { 4107 if (hlen == sizeof(struct ip)) { 4108 sum = in_cksum_hdr(ip); 4109 } else { 4110 sum = in_cksum(m, hlen); 4111 } 4112 } 4113 if (sum) { 4114 ipstat.ips_badsum++; 4115 goto bad; 4116 } 4117 4118 /* Retrieve the packet length. */ 4119 len = ntohs(ip->ip_len); 4120 4121 /* 4122 * Check for additional length bogosity 4123 */ 4124 if (len < hlen) { 4125 ipstat.ips_badlen++; 4126 goto bad; 4127 } 4128 4129 /* 4130 * Check that the amount of data in the buffers 4131 * is as at least much as the IP header would have us expect. 4132 * Drop packet if shorter than we expect. 4133 */ 4134 if (m->m_pkthdr.len < len) { 4135 ipstat.ips_tooshort++; 4136 goto bad; 4137 } 4138 4139 /* Checks out, proceed */ 4140 *mp = m; 4141 return (0); 4142 4143 bad: 4144 *mp = m; 4145 return (-1); 4146 } 4147 4148 #ifdef INET6 4149 /* 4150 * Same as above, but for IPv6. 4151 * Cut-and-pasted from ip6_input.c. 4152 * XXX Should we update ip6stat, or not? 4153 */ 4154 static int 4155 bridge_ip6_checkbasic(struct mbuf **mp) 4156 { 4157 struct mbuf *m = *mp; 4158 struct ip6_hdr *ip6; 4159 4160 /* 4161 * If the IPv6 header is not aligned, slurp it up into a new 4162 * mbuf with space for link headers, in the event we forward 4163 * it. Otherwise, if it is aligned, make sure the entire base 4164 * IPv6 header is in the first mbuf of the chain. 4165 */ 4166 #if 0 /* notyet */ 4167 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 4168 struct ifnet *inifp = m->m_pkthdr.rcvif; 4169 if ((m = m_copyup(m, sizeof(struct ip6_hdr), 4170 (max_linkhdr + 3) & ~3)) == NULL) { 4171 /* XXXJRT new stat, please */ 4172 ip6stat.ip6s_toosmall++; 4173 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 4174 goto bad; 4175 } 4176 } else 4177 #endif 4178 if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) { 4179 struct ifnet *inifp = m->m_pkthdr.rcvif; 4180 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 4181 ip6stat.ip6s_toosmall++; 4182 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 4183 goto bad; 4184 } 4185 } 4186 4187 ip6 = mtod(m, struct ip6_hdr *); 4188 4189 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 4190 ip6stat.ip6s_badvers++; 4191 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 4192 goto bad; 4193 } 4194 4195 /* Checks out, proceed */ 4196 *mp = m; 4197 return (0); 4198 4199 bad: 4200 *mp = m; 4201 return (-1); 4202 } 4203 #endif /* INET6 */ 4204 4205 /* 4206 * bridge_fragment: 4207 * 4208 * Return a fragmented mbuf chain. 4209 */ 4210 static int 4211 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh, 4212 int snap, struct llc *llc) 4213 { 4214 struct mbuf *m0; 4215 struct ip *ip; 4216 int error = -1; 4217 4218 if (m->m_len < sizeof(struct ip) && 4219 (m = m_pullup(m, sizeof(struct ip))) == NULL) 4220 goto out; 4221 ip = mtod(m, struct ip *); 4222 4223 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist, 4224 CSUM_DELAY_IP); 4225 if (error) 4226 goto out; 4227 4228 /* walk the chain and re-add the Ethernet header */ 4229 for (m0 = m; m0; m0 = m0->m_nextpkt) { 4230 if (error == 0) { 4231 if (snap) { 4232 M_PREPEND(m0, sizeof(struct llc), MB_DONTWAIT); 4233 if (m0 == NULL) { 4234 error = ENOBUFS; 4235 continue; 4236 } 4237 bcopy(llc, mtod(m0, caddr_t), 4238 sizeof(struct llc)); 4239 } 4240 M_PREPEND(m0, ETHER_HDR_LEN, MB_DONTWAIT); 4241 if (m0 == NULL) { 4242 error = ENOBUFS; 4243 continue; 4244 } 4245 bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN); 4246 } else 4247 m_freem(m); 4248 } 4249 4250 if (error == 0) 4251 ipstat.ips_fragmented++; 4252 4253 return (error); 4254 4255 out: 4256 if (m != NULL) 4257 m_freem(m); 4258 return (error); 4259 } 4260 4261 static void 4262 bridge_enqueue_handler(netmsg_t msg) 4263 { 4264 struct netmsg_packet *nmp; 4265 struct ifnet *dst_ifp; 4266 struct mbuf *m; 4267 4268 nmp = &msg->packet; 4269 m = nmp->nm_packet; 4270 dst_ifp = nmp->base.lmsg.u.ms_resultp; 4271 mbuftrackid(m, 71); 4272 4273 bridge_handoff(dst_ifp->if_bridge, dst_ifp, m, 1); 4274 } 4275 4276 static void 4277 bridge_handoff(struct bridge_softc *sc, struct ifnet *dst_ifp, 4278 struct mbuf *m, int from_us) 4279 { 4280 struct mbuf *m0; 4281 struct ifnet *bifp; 4282 4283 bifp = sc->sc_ifp; 4284 mbuftrackid(m, 72); 4285 4286 /* We may be sending a fragment so traverse the mbuf */ 4287 for (; m; m = m0) { 4288 struct altq_pktattr pktattr; 4289 4290 m0 = m->m_nextpkt; 4291 m->m_nextpkt = NULL; 4292 4293 /* 4294 * If being sent from our host override ether_shost 4295 * with the bridge MAC. This is mandatory for ARP 4296 * so things don't get confused. In particular we 4297 * don't want ARPs to get associated with link interfaces 4298 * under the bridge which might or might not stay valid. 4299 * 4300 * Also override ether_shost when relaying a packet out 4301 * the same interface it came in on, due to multi-homed 4302 * addresses & default routes, otherwise switches will 4303 * get very confused. 4304 * 4305 * Otherwise if we are in transparent mode. 4306 */ 4307 if (from_us || m->m_pkthdr.rcvif == dst_ifp) { 4308 m_copyback(m, 4309 offsetof(struct ether_header, ether_shost), 4310 ETHER_ADDR_LEN, IF_LLADDR(sc->sc_ifp)); 4311 } else if ((bifp->if_flags & IFF_LINK0) && 4312 (m->m_pkthdr.fw_flags & BRIDGE_MBUF_TAGGED)) { 4313 m_copyback(m, 4314 offsetof(struct ether_header, ether_shost), 4315 ETHER_ADDR_LEN, 4316 m->m_pkthdr.br.ether.ether_shost); 4317 } /* else retain shost */ 4318 4319 if (ifq_is_enabled(&dst_ifp->if_snd)) 4320 altq_etherclassify(&dst_ifp->if_snd, m, &pktattr); 4321 4322 ifq_dispatch(dst_ifp, m, &pktattr); 4323 } 4324 } 4325 4326 static void 4327 bridge_control_dispatch(netmsg_t msg) 4328 { 4329 struct netmsg_brctl *bc_msg = (struct netmsg_brctl *)msg; 4330 struct ifnet *bifp = bc_msg->bc_sc->sc_ifp; 4331 int error; 4332 4333 ifnet_serialize_all(bifp); 4334 error = bc_msg->bc_func(bc_msg->bc_sc, bc_msg->bc_arg); 4335 ifnet_deserialize_all(bifp); 4336 4337 lwkt_replymsg(&bc_msg->base.lmsg, error); 4338 } 4339 4340 static int 4341 bridge_control(struct bridge_softc *sc, u_long cmd, 4342 bridge_ctl_t bc_func, void *bc_arg) 4343 { 4344 struct ifnet *bifp = sc->sc_ifp; 4345 struct netmsg_brctl bc_msg; 4346 int error; 4347 4348 ASSERT_IFNET_SERIALIZED_ALL(bifp); 4349 4350 bzero(&bc_msg, sizeof(bc_msg)); 4351 4352 netmsg_init(&bc_msg.base, NULL, &curthread->td_msgport, 4353 0, bridge_control_dispatch); 4354 bc_msg.bc_func = bc_func; 4355 bc_msg.bc_sc = sc; 4356 bc_msg.bc_arg = bc_arg; 4357 4358 ifnet_deserialize_all(bifp); 4359 error = lwkt_domsg(BRIDGE_CFGPORT, &bc_msg.base.lmsg, 0); 4360 ifnet_serialize_all(bifp); 4361 return error; 4362 } 4363 4364 static void 4365 bridge_add_bif_handler(netmsg_t msg) 4366 { 4367 struct netmsg_braddbif *amsg = (struct netmsg_braddbif *)msg; 4368 struct bridge_softc *sc; 4369 struct bridge_iflist *bif; 4370 4371 sc = amsg->br_softc; 4372 4373 bif = kmalloc(sizeof(*bif), M_DEVBUF, M_WAITOK | M_ZERO); 4374 bif->bif_ifp = amsg->br_bif_ifp; 4375 bif->bif_onlist = 1; 4376 bif->bif_info = amsg->br_bif_info; 4377 4378 /* 4379 * runs through bif_info 4380 */ 4381 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER; 4382 4383 TAILQ_INSERT_HEAD(&sc->sc_iflists[mycpuid], bif, bif_next); 4384 4385 ifnet_forwardmsg(&amsg->base.lmsg, mycpuid + 1); 4386 } 4387 4388 static void 4389 bridge_add_bif(struct bridge_softc *sc, struct bridge_ifinfo *bif_info, 4390 struct ifnet *ifp) 4391 { 4392 struct netmsg_braddbif amsg; 4393 4394 ASSERT_IFNET_NOT_SERIALIZED_ALL(sc->sc_ifp); 4395 4396 netmsg_init(&amsg.base, NULL, &curthread->td_msgport, 4397 0, bridge_add_bif_handler); 4398 amsg.br_softc = sc; 4399 amsg.br_bif_info = bif_info; 4400 amsg.br_bif_ifp = ifp; 4401 4402 ifnet_domsg(&amsg.base.lmsg, 0); 4403 } 4404 4405 static void 4406 bridge_del_bif_handler(netmsg_t msg) 4407 { 4408 struct netmsg_brdelbif *dmsg = (struct netmsg_brdelbif *)msg; 4409 struct bridge_softc *sc; 4410 struct bridge_iflist *bif; 4411 4412 sc = dmsg->br_softc; 4413 4414 /* 4415 * Locate the bif associated with the br_bif_info 4416 * on the current CPU 4417 */ 4418 bif = bridge_lookup_member_ifinfo(sc, dmsg->br_bif_info); 4419 KKASSERT(bif != NULL && bif->bif_onlist); 4420 4421 /* Remove the bif from the current CPU's iflist */ 4422 bif->bif_onlist = 0; 4423 TAILQ_REMOVE(dmsg->br_bif_list, bif, bif_next); 4424 4425 /* Save the removed bif for later freeing */ 4426 TAILQ_INSERT_HEAD(dmsg->br_bif_list, bif, bif_next); 4427 4428 ifnet_forwardmsg(&dmsg->base.lmsg, mycpuid + 1); 4429 } 4430 4431 static void 4432 bridge_del_bif(struct bridge_softc *sc, struct bridge_ifinfo *bif_info, 4433 struct bridge_iflist_head *saved_bifs) 4434 { 4435 struct netmsg_brdelbif dmsg; 4436 4437 ASSERT_IFNET_NOT_SERIALIZED_ALL(sc->sc_ifp); 4438 4439 netmsg_init(&dmsg.base, NULL, &curthread->td_msgport, 4440 0, bridge_del_bif_handler); 4441 dmsg.br_softc = sc; 4442 dmsg.br_bif_info = bif_info; 4443 dmsg.br_bif_list = saved_bifs; 4444 4445 ifnet_domsg(&dmsg.base.lmsg, 0); 4446 } 4447