| 1 | /* |
| 2 | * Copyright (C) 1999-2000 by Maksim Yevmenkin <m_evmenkin@yahoo.com> |
| 3 | * All rights reserved. |
| 4 | * |
| 5 | * Redistribution and use in source and binary forms, with or without |
| 6 | * modification, are permitted provided that the following conditions |
| 7 | * are met: |
| 8 | * 1. Redistributions of source code must retain the above copyright |
| 9 | * notice, this list of conditions and the following disclaimer. |
| 10 | * 2. Redistributions in binary form must reproduce the above copyright |
| 11 | * notice, this list of conditions and the following disclaimer in the |
| 12 | * documentation and/or other materials provided with the distribution. |
| 13 | * |
| 14 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND |
| 15 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 16 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 17 | * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE |
| 18 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 19 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 20 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 21 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 22 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 23 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 24 | * SUCH DAMAGE. |
| 25 | * |
| 26 | * BASED ON: |
| 27 | * ------------------------------------------------------------------------- |
| 28 | * |
| 29 | * Copyright (c) 1988, Julian Onions <jpo@cs.nott.ac.uk> |
| 30 | * Nottingham University 1987. |
| 31 | */ |
| 32 | |
| 33 | /* |
| 34 | * $FreeBSD: src/sys/net/if_tap.c,v 1.3.2.3 2002/04/14 21:41:48 luigi Exp $ |
| 35 | * $DragonFly: src/sys/net/tap/if_tap.c,v 1.41 2008/09/05 17:03:15 dillon Exp $ |
| 36 | * $Id: if_tap.c,v 0.21 2000/07/23 21:46:02 max Exp $ |
| 37 | */ |
| 38 | |
| 39 | #include "opt_inet.h" |
| 40 | |
| 41 | #include <sys/param.h> |
| 42 | #include <sys/conf.h> |
| 43 | #include <sys/device.h> |
| 44 | #include <sys/filedesc.h> |
| 45 | #include <sys/filio.h> |
| 46 | #include <sys/kernel.h> |
| 47 | #include <sys/malloc.h> |
| 48 | #include <sys/mbuf.h> |
| 49 | #include <sys/poll.h> |
| 50 | #include <sys/proc.h> |
| 51 | #include <sys/priv.h> |
| 52 | #include <sys/signalvar.h> |
| 53 | #include <sys/socket.h> |
| 54 | #include <sys/sockio.h> |
| 55 | #include <sys/sysctl.h> |
| 56 | #include <sys/systm.h> |
| 57 | #include <sys/thread2.h> |
| 58 | #include <sys/ttycom.h> |
| 59 | #include <sys/uio.h> |
| 60 | #include <sys/vnode.h> |
| 61 | #include <sys/serialize.h> |
| 62 | |
| 63 | #include <net/bpf.h> |
| 64 | #include <net/ethernet.h> |
| 65 | #include <net/if.h> |
| 66 | #include <net/ifq_var.h> |
| 67 | #include <net/if_arp.h> |
| 68 | #include <net/if_clone.h> |
| 69 | #include <net/route.h> |
| 70 | #include <sys/devfs.h> |
| 71 | |
| 72 | #include <netinet/in.h> |
| 73 | |
| 74 | #include "if_tapvar.h" |
| 75 | #include "if_tap.h" |
| 76 | |
| 77 | #define TAP_IFFLAGS (IFF_BROADCAST|IFF_SIMPLEX|IFF_MULTICAST) |
| 78 | |
| 79 | #define TAP_PREALLOCATED_UNITS 4 |
| 80 | |
| 81 | #define CDEV_NAME "tap" |
| 82 | #define CDEV_MAJOR 149 |
| 83 | #define TAPDEBUG if (tapdebug) if_printf |
| 84 | |
| 85 | #define TAP "tap" |
| 86 | #define VMNET "vmnet" |
| 87 | #define VMNET_DEV_MASK 0x00010000 |
| 88 | |
| 89 | DEVFS_DECLARE_CLONE_BITMAP(tap); |
| 90 | |
| 91 | /* module */ |
| 92 | static int tapmodevent (module_t, int, void *); |
| 93 | |
| 94 | /* device */ |
| 95 | static struct tap_softc *tapcreate(int, cdev_t); |
| 96 | static void tapdestroy(struct tap_softc *); |
| 97 | |
| 98 | /* clone */ |
| 99 | static int tap_clone_create(struct if_clone *, int); |
| 100 | static void tap_clone_destroy(struct ifnet *); |
| 101 | |
| 102 | |
| 103 | /* network interface */ |
| 104 | static void tapifstart (struct ifnet *); |
| 105 | static int tapifioctl (struct ifnet *, u_long, caddr_t, |
| 106 | struct ucred *); |
| 107 | static void tapifinit (void *); |
| 108 | static void tapifstop(struct tap_softc *, int); |
| 109 | static void tapifflags(struct tap_softc *); |
| 110 | |
| 111 | /* character device */ |
| 112 | static d_open_t tapopen; |
| 113 | static d_clone_t tapclone; |
| 114 | static d_close_t tapclose; |
| 115 | static d_read_t tapread; |
| 116 | static d_write_t tapwrite; |
| 117 | static d_ioctl_t tapioctl; |
| 118 | static d_poll_t tappoll; |
| 119 | static d_kqfilter_t tapkqfilter; |
| 120 | |
| 121 | static struct dev_ops tap_ops = { |
| 122 | { CDEV_NAME, CDEV_MAJOR, 0 }, |
| 123 | .d_open = tapopen, |
| 124 | .d_close = tapclose, |
| 125 | .d_read = tapread, |
| 126 | .d_write = tapwrite, |
| 127 | .d_ioctl = tapioctl, |
| 128 | .d_poll = tappoll, |
| 129 | .d_kqfilter = tapkqfilter |
| 130 | }; |
| 131 | |
| 132 | static int taprefcnt = 0; /* module ref. counter */ |
| 133 | static int taplastunit = -1; /* max. open unit number */ |
| 134 | static int tapdebug = 0; /* debug flag */ |
| 135 | static int tapuopen = 0; /* all user open() */ |
| 136 | static int tapuponopen = 0; /* IFF_UP */ |
| 137 | |
| 138 | MALLOC_DECLARE(M_TAP); |
| 139 | MALLOC_DEFINE(M_TAP, CDEV_NAME, "Ethernet tunnel interface"); |
| 140 | struct if_clone tap_cloner = IF_CLONE_INITIALIZER("tap", |
| 141 | tap_clone_create, tap_clone_destroy, |
| 142 | 0, IF_MAXUNIT); |
| 143 | static SLIST_HEAD(,tap_softc) tap_listhead = |
| 144 | SLIST_HEAD_INITIALIZER(&tap_listhead); |
| 145 | |
| 146 | SYSCTL_INT(_debug, OID_AUTO, if_tap_debug, CTLFLAG_RW, &tapdebug, 0, ""); |
| 147 | SYSCTL_DECL(_net_link); |
| 148 | SYSCTL_NODE(_net_link, OID_AUTO, tap, CTLFLAG_RW, 0, |
| 149 | "Ethernet tunnel software network interface"); |
| 150 | SYSCTL_INT(_net_link_tap, OID_AUTO, user_open, CTLFLAG_RW, &tapuopen, 0, |
| 151 | "Allow user to open /dev/tap (based on node permissions)"); |
| 152 | SYSCTL_INT(_net_link_tap, OID_AUTO, up_on_open, CTLFLAG_RW, &tapuponopen, 0, |
| 153 | "Bring interface up when /dev/tap is opened"); |
| 154 | SYSCTL_INT(_net_link_tap, OID_AUTO, debug, CTLFLAG_RW, &tapdebug, 0, ""); |
| 155 | |
| 156 | DEV_MODULE(if_tap, tapmodevent, NULL); |
| 157 | |
| 158 | /* |
| 159 | * tapmodevent |
| 160 | * |
| 161 | * module event handler |
| 162 | */ |
| 163 | static int |
| 164 | tapmodevent(module_t mod, int type, void *data) |
| 165 | { |
| 166 | static int attached = 0; |
| 167 | struct tap_softc *tp, *ntp; |
| 168 | int i; |
| 169 | |
| 170 | switch (type) { |
| 171 | case MOD_LOAD: |
| 172 | if (attached) |
| 173 | return (EEXIST); |
| 174 | |
| 175 | make_autoclone_dev(&tap_ops, &DEVFS_CLONE_BITMAP(tap), tapclone, |
| 176 | UID_ROOT, GID_WHEEL, 0600, "tap"); |
| 177 | SLIST_INIT(&tap_listhead); |
| 178 | if_clone_attach(&tap_cloner); |
| 179 | |
| 180 | for (i = 0; i < TAP_PREALLOCATED_UNITS; ++i) { |
| 181 | make_dev(&tap_ops, i, UID_ROOT, GID_WHEEL, |
| 182 | 0600, "tap%d", i); |
| 183 | devfs_clone_bitmap_set(&DEVFS_CLONE_BITMAP(tap), i); |
| 184 | } |
| 185 | |
| 186 | attached = 1; |
| 187 | break; |
| 188 | |
| 189 | case MOD_UNLOAD: |
| 190 | if (taprefcnt > 0) |
| 191 | return (EBUSY); |
| 192 | |
| 193 | if_clone_detach(&tap_cloner); |
| 194 | |
| 195 | /* Maintain tap ifs in a local list */ |
| 196 | SLIST_FOREACH_MUTABLE(tp, &tap_listhead, tap_link, ntp) |
| 197 | tapdestroy(tp); |
| 198 | |
| 199 | attached = 0; |
| 200 | |
| 201 | devfs_clone_handler_del("tap"); |
| 202 | dev_ops_remove_all(&tap_ops); |
| 203 | devfs_clone_bitmap_uninit(&DEVFS_CLONE_BITMAP(tap)); |
| 204 | break; |
| 205 | |
| 206 | default: |
| 207 | return (EOPNOTSUPP); |
| 208 | } |
| 209 | |
| 210 | return (0); |
| 211 | } /* tapmodevent */ |
| 212 | |
| 213 | |
| 214 | /* |
| 215 | * tapcreate - create or clone an interface |
| 216 | */ |
| 217 | static struct tap_softc * |
| 218 | tapcreate(int unit, cdev_t dev) |
| 219 | { |
| 220 | const char *name = TAP; |
| 221 | struct ifnet *ifp; |
| 222 | struct tap_softc *tp; |
| 223 | uint8_t ether_addr[ETHER_ADDR_LEN]; |
| 224 | |
| 225 | tp = kmalloc(sizeof(*tp), M_TAP, M_WAITOK | M_ZERO); |
| 226 | dev->si_drv1 = tp; |
| 227 | tp->tap_dev = dev; |
| 228 | tp->tap_unit = unit; |
| 229 | |
| 230 | reference_dev(dev); /* tp association */ |
| 231 | |
| 232 | /* generate fake MAC address: 00 bd xx xx xx unit_no */ |
| 233 | ether_addr[0] = 0x00; |
| 234 | ether_addr[1] = 0xbd; |
| 235 | bcopy(&ticks, ðer_addr[2], 3); |
| 236 | ether_addr[5] = (u_char)unit; |
| 237 | |
| 238 | /* fill the rest and attach interface */ |
| 239 | ifp = &tp->tap_if; |
| 240 | ifp->if_softc = tp; |
| 241 | |
| 242 | if_initname(ifp, name, unit); |
| 243 | if (unit > taplastunit) |
| 244 | taplastunit = unit; |
| 245 | |
| 246 | ifp->if_init = tapifinit; |
| 247 | ifp->if_start = tapifstart; |
| 248 | ifp->if_ioctl = tapifioctl; |
| 249 | ifp->if_mtu = ETHERMTU; |
| 250 | ifp->if_flags = TAP_IFFLAGS; |
| 251 | ifq_set_maxlen(&ifp->if_snd, ifqmaxlen); |
| 252 | ifq_set_ready(&ifp->if_snd); |
| 253 | |
| 254 | ether_ifattach(ifp, ether_addr, NULL); |
| 255 | |
| 256 | tp->tap_flags |= TAP_INITED; |
| 257 | tp->tap_devq.ifq_maxlen = ifqmaxlen; |
| 258 | |
| 259 | SLIST_INSERT_HEAD(&tap_listhead, tp, tap_link); |
| 260 | |
| 261 | TAPDEBUG(ifp, "created. minor = %#x\n", minor(dev)); |
| 262 | return (tp); |
| 263 | } |
| 264 | |
| 265 | static |
| 266 | struct tap_softc * |
| 267 | tapfind(int unit) |
| 268 | { |
| 269 | struct tap_softc *tp; |
| 270 | |
| 271 | SLIST_FOREACH(tp, &tap_listhead, tap_link) { |
| 272 | if (tp->tap_unit == unit) |
| 273 | return(tp); |
| 274 | } |
| 275 | return (NULL); |
| 276 | } |
| 277 | |
| 278 | /* |
| 279 | * tap_clone_create: |
| 280 | * |
| 281 | * Create a new tap instance via ifconfig. |
| 282 | */ |
| 283 | static int |
| 284 | tap_clone_create(struct if_clone *ifc __unused, int unit) |
| 285 | { |
| 286 | struct tap_softc *tp; |
| 287 | cdev_t dev; |
| 288 | |
| 289 | tp = tapfind(unit); |
| 290 | if (tp == NULL) { |
| 291 | devfs_clone_bitmap_set(&DEVFS_CLONE_BITMAP(tap), unit); |
| 292 | dev = make_dev(&tap_ops, unit, UID_ROOT, GID_WHEEL, |
| 293 | 0600, "%s%d", TAP, unit); |
| 294 | tp = tapcreate(unit, dev); |
| 295 | } |
| 296 | tp->tap_flags |= TAP_CLONE; |
| 297 | TAPDEBUG(&tp->tap_if, "clone created. minor = %#x tap_flags = 0x%x\n", |
| 298 | minor(tp->tap_dev), tp->tap_flags); |
| 299 | |
| 300 | return (0); |
| 301 | } |
| 302 | |
| 303 | /* |
| 304 | * tapopen |
| 305 | * |
| 306 | * to open tunnel. must be superuser |
| 307 | */ |
| 308 | static int |
| 309 | tapopen(struct dev_open_args *ap) |
| 310 | { |
| 311 | cdev_t dev = NULL; |
| 312 | struct tap_softc *tp = NULL; |
| 313 | struct ifnet *ifp = NULL; |
| 314 | int error; |
| 315 | |
| 316 | if (tapuopen == 0 && |
| 317 | (error = priv_check_cred(ap->a_cred, PRIV_ROOT, 0)) != 0) |
| 318 | return (error); |
| 319 | |
| 320 | get_mplock(); |
| 321 | dev = ap->a_head.a_dev; |
| 322 | tp = dev->si_drv1; |
| 323 | if (tp == NULL) |
| 324 | tp = tapcreate(minor(dev), dev); |
| 325 | if (tp->tap_flags & TAP_OPEN) { |
| 326 | rel_mplock(); |
| 327 | return (EBUSY); |
| 328 | } |
| 329 | ifp = &tp->arpcom.ac_if; |
| 330 | |
| 331 | if ((tp->tap_flags & TAP_CLONE) == 0) { |
| 332 | EVENTHANDLER_INVOKE(ifnet_attach_event, ifp); |
| 333 | |
| 334 | /* Announce the return of the interface. */ |
| 335 | rt_ifannouncemsg(ifp, IFAN_ARRIVAL); |
| 336 | } |
| 337 | |
| 338 | bcopy(tp->arpcom.ac_enaddr, tp->ether_addr, sizeof(tp->ether_addr)); |
| 339 | |
| 340 | if (curthread->td_proc) |
| 341 | fsetown(curthread->td_proc->p_pid, &tp->tap_sigtd); |
| 342 | tp->tap_flags |= TAP_OPEN; |
| 343 | taprefcnt ++; |
| 344 | |
| 345 | if (tapuponopen && (ifp->if_flags & IFF_UP) == 0) { |
| 346 | crit_enter(); |
| 347 | if_up(ifp); |
| 348 | crit_exit(); |
| 349 | |
| 350 | ifnet_serialize_all(ifp); |
| 351 | tapifflags(tp); |
| 352 | ifnet_deserialize_all(ifp); |
| 353 | |
| 354 | tp->tap_flags |= TAP_CLOSEDOWN; |
| 355 | } |
| 356 | |
| 357 | TAPDEBUG(ifp, "opened. minor = %#x, refcnt = %d, taplastunit = %d\n", |
| 358 | minor(tp->tap_dev), taprefcnt, taplastunit); |
| 359 | |
| 360 | rel_mplock(); |
| 361 | return (0); |
| 362 | } |
| 363 | |
| 364 | static int |
| 365 | tapclone(struct dev_clone_args *ap) |
| 366 | { |
| 367 | int unit; |
| 368 | |
| 369 | unit = devfs_clone_bitmap_get(&DEVFS_CLONE_BITMAP(tap), 0); |
| 370 | ap->a_dev = make_only_dev(&tap_ops, unit, UID_ROOT, GID_WHEEL, |
| 371 | 0600, "%s%d", TAP, unit); |
| 372 | tapcreate(unit, ap->a_dev); |
| 373 | return (0); |
| 374 | } |
| 375 | |
| 376 | /* |
| 377 | * tapclose |
| 378 | * |
| 379 | * close the device - mark i/f down & delete routing info |
| 380 | */ |
| 381 | static int |
| 382 | tapclose(struct dev_close_args *ap) |
| 383 | { |
| 384 | cdev_t dev = ap->a_head.a_dev; |
| 385 | struct tap_softc *tp = dev->si_drv1; |
| 386 | struct ifnet *ifp = &tp->tap_if; |
| 387 | int clear_flags = 0; |
| 388 | |
| 389 | get_mplock(); |
| 390 | |
| 391 | /* Junk all pending output */ |
| 392 | ifq_purge(&ifp->if_snd); |
| 393 | |
| 394 | /* |
| 395 | * Do not bring the interface down, and do not anything with |
| 396 | * interface, if we are in VMnet mode. just close the device. |
| 397 | * |
| 398 | * If the interface is not cloned, we always bring it down. |
| 399 | * |
| 400 | * If the interface is cloned, then we bring it down during |
| 401 | * closing only if it was brought up during opening. |
| 402 | */ |
| 403 | if ((tp->tap_flags & TAP_VMNET) == 0 && |
| 404 | ((tp->tap_flags & TAP_CLONE) == 0 || |
| 405 | (tp->tap_flags & TAP_CLOSEDOWN))) { |
| 406 | if (ifp->if_flags & IFF_UP) |
| 407 | if_down(ifp); |
| 408 | clear_flags = 1; |
| 409 | } |
| 410 | ifnet_serialize_all(ifp); |
| 411 | tapifstop(tp, clear_flags); |
| 412 | ifnet_deserialize_all(ifp); |
| 413 | |
| 414 | if ((tp->tap_flags & TAP_CLONE) == 0) { |
| 415 | if_purgeaddrs_nolink(ifp); |
| 416 | |
| 417 | EVENTHANDLER_INVOKE(ifnet_detach_event, ifp); |
| 418 | |
| 419 | /* Announce the departure of the interface. */ |
| 420 | rt_ifannouncemsg(ifp, IFAN_DEPARTURE); |
| 421 | } |
| 422 | |
| 423 | funsetown(tp->tap_sigio); |
| 424 | tp->tap_sigio = NULL; |
| 425 | selwakeup(&tp->tap_rsel); |
| 426 | |
| 427 | tp->tap_flags &= ~TAP_OPEN; |
| 428 | funsetown(tp->tap_sigtd); |
| 429 | tp->tap_sigtd = NULL; |
| 430 | |
| 431 | taprefcnt --; |
| 432 | if (taprefcnt < 0) { |
| 433 | taprefcnt = 0; |
| 434 | if_printf(ifp, "minor = %#x, refcnt = %d is out of sync. " |
| 435 | "set refcnt to 0\n", minor(tp->tap_dev), taprefcnt); |
| 436 | } |
| 437 | |
| 438 | TAPDEBUG(ifp, "closed. minor = %#x, refcnt = %d, taplastunit = %d\n", |
| 439 | minor(tp->tap_dev), taprefcnt, taplastunit); |
| 440 | |
| 441 | if (tp->tap_unit >= TAP_PREALLOCATED_UNITS) |
| 442 | tapdestroy(tp); |
| 443 | |
| 444 | rel_mplock(); |
| 445 | return (0); |
| 446 | } |
| 447 | |
| 448 | /* |
| 449 | * tapdestroy: |
| 450 | * |
| 451 | * Destroy a tap instance. |
| 452 | */ |
| 453 | static void |
| 454 | tapdestroy(struct tap_softc *tp) |
| 455 | { |
| 456 | struct ifnet *ifp = &tp->arpcom.ac_if; |
| 457 | cdev_t dev; |
| 458 | |
| 459 | TAPDEBUG(ifp, "destroyed. minor = %#x, refcnt = %d, taplastunit = %d\n", |
| 460 | minor(tp->tap_dev), taprefcnt, taplastunit); |
| 461 | |
| 462 | ifnet_serialize_all(ifp); |
| 463 | tapifstop(tp, 1); |
| 464 | ifnet_deserialize_all(ifp); |
| 465 | |
| 466 | ether_ifdetach(ifp); |
| 467 | SLIST_REMOVE(&tap_listhead, tp, tap_softc, tap_link); |
| 468 | |
| 469 | dev = tp->tap_dev; |
| 470 | tp->tap_dev = NULL; |
| 471 | dev->si_drv1 = NULL; |
| 472 | |
| 473 | release_dev(dev); /* tp association */ |
| 474 | |
| 475 | /* |
| 476 | * Also destroy the cloned device |
| 477 | */ |
| 478 | if (tp->tap_unit >= TAP_PREALLOCATED_UNITS) { |
| 479 | destroy_dev(dev); |
| 480 | devfs_clone_bitmap_put(&DEVFS_CLONE_BITMAP(tap), tp->tap_unit); |
| 481 | } |
| 482 | |
| 483 | kfree(tp, M_TAP); |
| 484 | |
| 485 | taplastunit--; |
| 486 | } |
| 487 | |
| 488 | /* |
| 489 | * tap_clone_destroy: |
| 490 | * |
| 491 | * Destroy a tap instance. |
| 492 | */ |
| 493 | static void |
| 494 | tap_clone_destroy(struct ifnet *ifp) |
| 495 | { |
| 496 | struct tap_softc *tp = ifp->if_softc; |
| 497 | |
| 498 | TAPDEBUG(&tp->tap_if, "clone destroyed. minor = %#x tap_flags = 0x%x\n", |
| 499 | minor(tp->tap_dev), tp->tap_flags); |
| 500 | if (tp->tap_flags & TAP_CLONE) |
| 501 | tapdestroy(tp); |
| 502 | } |
| 503 | |
| 504 | /* |
| 505 | * tapifinit |
| 506 | * |
| 507 | * Network interface initialization function (called with if serializer held) |
| 508 | * |
| 509 | * MPSAFE |
| 510 | */ |
| 511 | static void |
| 512 | tapifinit(void *xtp) |
| 513 | { |
| 514 | struct tap_softc *tp = xtp; |
| 515 | struct ifnet *ifp = &tp->tap_if; |
| 516 | |
| 517 | TAPDEBUG(ifp, "initializing, minor = %#x tap_flags = 0x%x\n", |
| 518 | minor(tp->tap_dev), tp->tap_flags); |
| 519 | |
| 520 | ASSERT_IFNET_SERIALIZED_ALL(ifp); |
| 521 | |
| 522 | tapifstop(tp, 1); |
| 523 | |
| 524 | ifp->if_flags |= IFF_RUNNING; |
| 525 | ifp->if_flags &= ~IFF_OACTIVE; |
| 526 | |
| 527 | /* attempt to start output */ |
| 528 | tapifstart(ifp); |
| 529 | } |
| 530 | |
| 531 | |
| 532 | /* |
| 533 | * tapifioctl |
| 534 | * |
| 535 | * Process an ioctl request on network interface (called with if serializer |
| 536 | * held). |
| 537 | * |
| 538 | * MPSAFE |
| 539 | */ |
| 540 | static int |
| 541 | tapifioctl(struct ifnet *ifp, u_long cmd, caddr_t data, struct ucred *cr) |
| 542 | { |
| 543 | struct tap_softc *tp = (struct tap_softc *)(ifp->if_softc); |
| 544 | struct ifstat *ifs = NULL; |
| 545 | int dummy; |
| 546 | |
| 547 | switch (cmd) { |
| 548 | case SIOCSIFADDR: |
| 549 | case SIOCGIFADDR: |
| 550 | case SIOCSIFMTU: |
| 551 | dummy = ether_ioctl(ifp, cmd, data); |
| 552 | return (dummy); |
| 553 | |
| 554 | case SIOCSIFFLAGS: |
| 555 | tapifflags(tp); |
| 556 | break; |
| 557 | |
| 558 | case SIOCADDMULTI: /* XXX -- just like vmnet does */ |
| 559 | case SIOCDELMULTI: |
| 560 | break; |
| 561 | |
| 562 | case SIOCGIFSTATUS: |
| 563 | ifs = (struct ifstat *)data; |
| 564 | dummy = strlen(ifs->ascii); |
| 565 | if ((tp->tap_flags & TAP_OPEN) && |
| 566 | dummy < sizeof(ifs->ascii)) { |
| 567 | if (tp->tap_sigtd && tp->tap_sigtd->sio_proc) { |
| 568 | ksnprintf(ifs->ascii + dummy, |
| 569 | sizeof(ifs->ascii) - dummy, |
| 570 | "\tOpened by pid %d\n", |
| 571 | (int)tp->tap_sigtd->sio_proc->p_pid); |
| 572 | } else { |
| 573 | ksnprintf(ifs->ascii + dummy, |
| 574 | sizeof(ifs->ascii) - dummy, |
| 575 | "\tOpened by <unknown>\n"); |
| 576 | } |
| 577 | } |
| 578 | break; |
| 579 | |
| 580 | default: |
| 581 | return (EINVAL); |
| 582 | } |
| 583 | |
| 584 | return (0); |
| 585 | } |
| 586 | |
| 587 | |
| 588 | /* |
| 589 | * tapifstart |
| 590 | * |
| 591 | * Queue packets from higher level ready to put out (called with if serializer |
| 592 | * held) |
| 593 | * |
| 594 | * MPSAFE |
| 595 | */ |
| 596 | static void |
| 597 | tapifstart(struct ifnet *ifp) |
| 598 | { |
| 599 | struct tap_softc *tp = ifp->if_softc; |
| 600 | struct ifqueue *ifq; |
| 601 | struct mbuf *m; |
| 602 | int has_data = 0; |
| 603 | |
| 604 | TAPDEBUG(ifp, "starting, minor = %#x\n", minor(tp->tap_dev)); |
| 605 | |
| 606 | /* |
| 607 | * do not junk pending output if we are in VMnet mode. |
| 608 | * XXX: can this do any harm because of queue overflow? |
| 609 | */ |
| 610 | |
| 611 | if (((tp->tap_flags & TAP_VMNET) == 0) && |
| 612 | ((tp->tap_flags & TAP_READY) != TAP_READY)) { |
| 613 | TAPDEBUG(ifp, "not ready. minor = %#x, tap_flags = 0x%x\n", |
| 614 | minor(tp->tap_dev), tp->tap_flags); |
| 615 | ifq_purge(&ifp->if_snd); |
| 616 | return; |
| 617 | } |
| 618 | |
| 619 | ifp->if_flags |= IFF_OACTIVE; |
| 620 | |
| 621 | ifq = &tp->tap_devq; |
| 622 | while ((m = ifq_dequeue(&ifp->if_snd, NULL)) != NULL) { |
| 623 | if (IF_QFULL(ifq)) { |
| 624 | IF_DROP(ifq); |
| 625 | ifp->if_oerrors++; |
| 626 | m_freem(m); |
| 627 | } else { |
| 628 | IF_ENQUEUE(ifq, m); |
| 629 | ifp->if_opackets++; |
| 630 | has_data = 1; |
| 631 | } |
| 632 | } |
| 633 | |
| 634 | if (has_data) { |
| 635 | if (tp->tap_flags & TAP_RWAIT) { |
| 636 | tp->tap_flags &= ~TAP_RWAIT; |
| 637 | wakeup((caddr_t)tp); |
| 638 | } |
| 639 | |
| 640 | get_mplock(); |
| 641 | KNOTE(&tp->tap_rsel.si_note, 0); |
| 642 | rel_mplock(); |
| 643 | |
| 644 | if ((tp->tap_flags & TAP_ASYNC) && (tp->tap_sigio != NULL)) { |
| 645 | get_mplock(); |
| 646 | pgsigio(tp->tap_sigio, SIGIO, 0); |
| 647 | rel_mplock(); |
| 648 | } |
| 649 | |
| 650 | /* |
| 651 | * selwakeup is not MPSAFE. tapifstart is. |
| 652 | */ |
| 653 | get_mplock(); |
| 654 | selwakeup(&tp->tap_rsel); |
| 655 | rel_mplock(); |
| 656 | } |
| 657 | |
| 658 | ifp->if_flags &= ~IFF_OACTIVE; |
| 659 | } |
| 660 | |
| 661 | |
| 662 | /* |
| 663 | * tapioctl |
| 664 | * |
| 665 | * The ops interface is now pretty minimal. Called via fileops with nothing |
| 666 | * held. |
| 667 | * |
| 668 | * MPSAFE |
| 669 | */ |
| 670 | static int |
| 671 | tapioctl(struct dev_ioctl_args *ap) |
| 672 | { |
| 673 | cdev_t dev = ap->a_head.a_dev; |
| 674 | caddr_t data = ap->a_data; |
| 675 | struct tap_softc *tp = dev->si_drv1; |
| 676 | struct ifnet *ifp = &tp->tap_if; |
| 677 | struct tapinfo *tapp = NULL; |
| 678 | struct mbuf *mb; |
| 679 | short f; |
| 680 | int error; |
| 681 | |
| 682 | ifnet_serialize_all(ifp); |
| 683 | error = 0; |
| 684 | |
| 685 | switch (ap->a_cmd) { |
| 686 | case TAPSIFINFO: |
| 687 | tapp = (struct tapinfo *)data; |
| 688 | ifp->if_mtu = tapp->mtu; |
| 689 | ifp->if_type = tapp->type; |
| 690 | ifp->if_baudrate = tapp->baudrate; |
| 691 | break; |
| 692 | |
| 693 | case TAPGIFINFO: |
| 694 | tapp = (struct tapinfo *)data; |
| 695 | tapp->mtu = ifp->if_mtu; |
| 696 | tapp->type = ifp->if_type; |
| 697 | tapp->baudrate = ifp->if_baudrate; |
| 698 | break; |
| 699 | |
| 700 | case TAPSDEBUG: |
| 701 | tapdebug = *(int *)data; |
| 702 | break; |
| 703 | |
| 704 | case TAPGDEBUG: |
| 705 | *(int *)data = tapdebug; |
| 706 | break; |
| 707 | |
| 708 | case FIOASYNC: |
| 709 | if (*(int *)data) |
| 710 | tp->tap_flags |= TAP_ASYNC; |
| 711 | else |
| 712 | tp->tap_flags &= ~TAP_ASYNC; |
| 713 | break; |
| 714 | |
| 715 | case FIONREAD: |
| 716 | *(int *)data = 0; |
| 717 | |
| 718 | /* Take a look at devq first */ |
| 719 | IF_POLL(&tp->tap_devq, mb); |
| 720 | if (mb == NULL) |
| 721 | mb = ifq_poll(&ifp->if_snd); |
| 722 | |
| 723 | if (mb != NULL) { |
| 724 | for(; mb != NULL; mb = mb->m_next) |
| 725 | *(int *)data += mb->m_len; |
| 726 | } |
| 727 | break; |
| 728 | |
| 729 | case FIOSETOWN: |
| 730 | error = fsetown(*(int *)data, &tp->tap_sigio); |
| 731 | break; |
| 732 | |
| 733 | case FIOGETOWN: |
| 734 | *(int *)data = fgetown(tp->tap_sigio); |
| 735 | break; |
| 736 | |
| 737 | /* this is deprecated, FIOSETOWN should be used instead */ |
| 738 | case TIOCSPGRP: |
| 739 | error = fsetown(-(*(int *)data), &tp->tap_sigio); |
| 740 | break; |
| 741 | |
| 742 | /* this is deprecated, FIOGETOWN should be used instead */ |
| 743 | case TIOCGPGRP: |
| 744 | *(int *)data = -fgetown(tp->tap_sigio); |
| 745 | break; |
| 746 | |
| 747 | /* VMware/VMnet port ioctl's */ |
| 748 | |
| 749 | case SIOCGIFFLAGS: /* get ifnet flags */ |
| 750 | bcopy(&ifp->if_flags, data, sizeof(ifp->if_flags)); |
| 751 | break; |
| 752 | |
| 753 | case VMIO_SIOCSIFFLAGS: /* VMware/VMnet SIOCSIFFLAGS */ |
| 754 | f = *(short *)data; |
| 755 | f &= 0x0fff; |
| 756 | f &= ~IFF_CANTCHANGE; |
| 757 | f |= IFF_UP; |
| 758 | ifp->if_flags = f | (ifp->if_flags & IFF_CANTCHANGE); |
| 759 | break; |
| 760 | |
| 761 | case OSIOCGIFADDR: /* get MAC address of the remote side */ |
| 762 | case SIOCGIFADDR: |
| 763 | bcopy(tp->ether_addr, data, sizeof(tp->ether_addr)); |
| 764 | break; |
| 765 | |
| 766 | case SIOCSIFADDR: /* set MAC address of the remote side */ |
| 767 | bcopy(data, tp->ether_addr, sizeof(tp->ether_addr)); |
| 768 | break; |
| 769 | |
| 770 | default: |
| 771 | error = ENOTTY; |
| 772 | break; |
| 773 | } |
| 774 | ifnet_deserialize_all(ifp); |
| 775 | return (error); |
| 776 | } |
| 777 | |
| 778 | |
| 779 | /* |
| 780 | * tapread |
| 781 | * |
| 782 | * The ops read interface - reads a packet at a time, or at |
| 783 | * least as much of a packet as can be read. |
| 784 | * |
| 785 | * Called from the fileops interface with nothing held. |
| 786 | * |
| 787 | * MPSAFE |
| 788 | */ |
| 789 | static int |
| 790 | tapread(struct dev_read_args *ap) |
| 791 | { |
| 792 | cdev_t dev = ap->a_head.a_dev; |
| 793 | struct uio *uio = ap->a_uio; |
| 794 | struct tap_softc *tp = dev->si_drv1; |
| 795 | struct ifnet *ifp = &tp->tap_if; |
| 796 | struct mbuf *m0 = NULL; |
| 797 | int error = 0, len; |
| 798 | |
| 799 | TAPDEBUG(ifp, "reading, minor = %#x\n", minor(tp->tap_dev)); |
| 800 | |
| 801 | if ((tp->tap_flags & TAP_READY) != TAP_READY) { |
| 802 | TAPDEBUG(ifp, "not ready. minor = %#x, tap_flags = 0x%x\n", |
| 803 | minor(tp->tap_dev), tp->tap_flags); |
| 804 | |
| 805 | return (EHOSTDOWN); |
| 806 | } |
| 807 | |
| 808 | tp->tap_flags &= ~TAP_RWAIT; |
| 809 | |
| 810 | /* sleep until we get a packet */ |
| 811 | do { |
| 812 | ifnet_serialize_all(ifp); |
| 813 | IF_DEQUEUE(&tp->tap_devq, m0); |
| 814 | if (m0 == NULL) { |
| 815 | if (ap->a_ioflag & IO_NDELAY) { |
| 816 | ifnet_deserialize_all(ifp); |
| 817 | return (EWOULDBLOCK); |
| 818 | } |
| 819 | tp->tap_flags |= TAP_RWAIT; |
| 820 | tsleep_interlock(tp, PCATCH); |
| 821 | ifnet_deserialize_all(ifp); |
| 822 | error = tsleep(tp, PCATCH | PINTERLOCKED, "taprd", 0); |
| 823 | if (error) |
| 824 | return (error); |
| 825 | } else { |
| 826 | ifnet_deserialize_all(ifp); |
| 827 | } |
| 828 | } while (m0 == NULL); |
| 829 | |
| 830 | BPF_MTAP(ifp, m0); |
| 831 | |
| 832 | /* xfer packet to user space */ |
| 833 | while ((m0 != NULL) && (uio->uio_resid > 0) && (error == 0)) { |
| 834 | len = (int)szmin(uio->uio_resid, m0->m_len); |
| 835 | if (len == 0) |
| 836 | break; |
| 837 | |
| 838 | error = uiomove(mtod(m0, caddr_t), (size_t)len, uio); |
| 839 | m0 = m_free(m0); |
| 840 | } |
| 841 | |
| 842 | if (m0 != NULL) { |
| 843 | TAPDEBUG(ifp, "dropping mbuf, minor = %#x\n", |
| 844 | minor(tp->tap_dev)); |
| 845 | m_freem(m0); |
| 846 | } |
| 847 | |
| 848 | return (error); |
| 849 | } |
| 850 | |
| 851 | /* |
| 852 | * tapwrite |
| 853 | * |
| 854 | * The ops write interface - an atomic write is a packet - or else! |
| 855 | * |
| 856 | * Called from the fileops interface with nothing held. |
| 857 | * |
| 858 | * MPSAFE |
| 859 | */ |
| 860 | static int |
| 861 | tapwrite(struct dev_write_args *ap) |
| 862 | { |
| 863 | cdev_t dev = ap->a_head.a_dev; |
| 864 | struct uio *uio = ap->a_uio; |
| 865 | struct tap_softc *tp = dev->si_drv1; |
| 866 | struct ifnet *ifp = &tp->tap_if; |
| 867 | struct mbuf *top = NULL, **mp = NULL, *m = NULL; |
| 868 | int error = 0; |
| 869 | size_t tlen, mlen; |
| 870 | |
| 871 | TAPDEBUG(ifp, "writing, minor = %#x\n", minor(tp->tap_dev)); |
| 872 | |
| 873 | if ((tp->tap_flags & TAP_READY) != TAP_READY) { |
| 874 | TAPDEBUG(ifp, "not ready. minor = %#x, tap_flags = 0x%x\n", |
| 875 | minor(tp->tap_dev), tp->tap_flags); |
| 876 | return (EHOSTDOWN); |
| 877 | } |
| 878 | |
| 879 | if (uio->uio_resid == 0) |
| 880 | return (0); |
| 881 | |
| 882 | if (uio->uio_resid > TAPMRU) { |
| 883 | TAPDEBUG(ifp, "invalid packet len = %ld, minor = %#x\n", |
| 884 | uio->uio_resid, minor(tp->tap_dev)); |
| 885 | |
| 886 | return (EIO); |
| 887 | } |
| 888 | tlen = uio->uio_resid; |
| 889 | |
| 890 | /* get a header mbuf */ |
| 891 | MGETHDR(m, MB_DONTWAIT, MT_DATA); |
| 892 | if (m == NULL) |
| 893 | return (ENOBUFS); |
| 894 | mlen = MHLEN; |
| 895 | |
| 896 | top = 0; |
| 897 | mp = ⊤ |
| 898 | while ((error == 0) && (uio->uio_resid > 0)) { |
| 899 | m->m_len = (int)szmin(mlen, uio->uio_resid); |
| 900 | error = uiomove(mtod(m, caddr_t), (size_t)m->m_len, uio); |
| 901 | *mp = m; |
| 902 | mp = &m->m_next; |
| 903 | if (uio->uio_resid > 0) { |
| 904 | MGET(m, MB_DONTWAIT, MT_DATA); |
| 905 | if (m == NULL) { |
| 906 | error = ENOBUFS; |
| 907 | break; |
| 908 | } |
| 909 | mlen = MLEN; |
| 910 | } |
| 911 | } |
| 912 | if (error) { |
| 913 | ifp->if_ierrors ++; |
| 914 | if (top) |
| 915 | m_freem(top); |
| 916 | return (error); |
| 917 | } |
| 918 | |
| 919 | top->m_pkthdr.len = (int)tlen; |
| 920 | top->m_pkthdr.rcvif = ifp; |
| 921 | |
| 922 | /* |
| 923 | * Ethernet bridge and bpf are handled in ether_input |
| 924 | * |
| 925 | * adjust mbuf and give packet to the ether_input |
| 926 | */ |
| 927 | ifnet_serialize_all(ifp); |
| 928 | ifp->if_input(ifp, top); |
| 929 | ifp->if_ipackets ++; /* ibytes are counted in ether_input */ |
| 930 | ifnet_deserialize_all(ifp); |
| 931 | |
| 932 | return (0); |
| 933 | } |
| 934 | |
| 935 | /* |
| 936 | * tappoll |
| 937 | * |
| 938 | * The poll interface, this is only useful on reads really. The write |
| 939 | * detect always returns true, write never blocks anyway, it either |
| 940 | * accepts the packet or drops it |
| 941 | * |
| 942 | * Called from the fileops interface with nothing held. |
| 943 | * |
| 944 | * MPSAFE |
| 945 | */ |
| 946 | static int |
| 947 | tappoll(struct dev_poll_args *ap) |
| 948 | { |
| 949 | cdev_t dev = ap->a_head.a_dev; |
| 950 | struct tap_softc *tp = dev->si_drv1; |
| 951 | struct ifnet *ifp = &tp->tap_if; |
| 952 | int revents = 0; |
| 953 | |
| 954 | TAPDEBUG(ifp, "polling, minor = %#x\n", minor(tp->tap_dev)); |
| 955 | |
| 956 | if (ap->a_events & (POLLIN | POLLRDNORM)) { |
| 957 | if (!IF_QEMPTY(&tp->tap_devq)) { |
| 958 | TAPDEBUG(ifp, |
| 959 | "has data in queue. minor = %#x\n", |
| 960 | minor(tp->tap_dev)); |
| 961 | |
| 962 | revents |= (ap->a_events & (POLLIN | POLLRDNORM)); |
| 963 | } else { |
| 964 | TAPDEBUG(ifp, "waiting for data, minor = %#x\n", |
| 965 | minor(tp->tap_dev)); |
| 966 | |
| 967 | get_mplock(); |
| 968 | selrecord(curthread, &tp->tap_rsel); |
| 969 | rel_mplock(); |
| 970 | } |
| 971 | } |
| 972 | |
| 973 | if (ap->a_events & (POLLOUT | POLLWRNORM)) |
| 974 | revents |= (ap->a_events & (POLLOUT | POLLWRNORM)); |
| 975 | ap->a_events = revents; |
| 976 | return (0); |
| 977 | } |
| 978 | |
| 979 | /* |
| 980 | * tapkqfilter - called from the fileops interface with nothing held |
| 981 | * |
| 982 | * MPSAFE |
| 983 | */ |
| 984 | static int filt_tapread(struct knote *kn, long hint); |
| 985 | static void filt_tapdetach(struct knote *kn); |
| 986 | static struct filterops tapread_filtops = |
| 987 | { 1, NULL, filt_tapdetach, filt_tapread }; |
| 988 | |
| 989 | static int |
| 990 | tapkqfilter(struct dev_kqfilter_args *ap) |
| 991 | { |
| 992 | cdev_t dev = ap->a_head.a_dev; |
| 993 | struct knote *kn = ap->a_kn; |
| 994 | struct tap_softc *tp; |
| 995 | struct klist *list; |
| 996 | struct ifnet *ifp; |
| 997 | |
| 998 | get_mplock(); |
| 999 | tp = dev->si_drv1; |
| 1000 | ifp = &tp->tap_if; |
| 1001 | ap->a_result =0; |
| 1002 | |
| 1003 | switch(kn->kn_filter) { |
| 1004 | case EVFILT_READ: |
| 1005 | list = &tp->tap_rsel.si_note; |
| 1006 | kn->kn_fop = &tapread_filtops; |
| 1007 | kn->kn_hook = (void *)tp; |
| 1008 | break; |
| 1009 | case EVFILT_WRITE: |
| 1010 | /* fall through */ |
| 1011 | default: |
| 1012 | ap->a_result = 1; |
| 1013 | rel_mplock(); |
| 1014 | return(0); |
| 1015 | } |
| 1016 | crit_enter(); |
| 1017 | SLIST_INSERT_HEAD(list, kn, kn_selnext); |
| 1018 | crit_exit(); |
| 1019 | rel_mplock(); |
| 1020 | return(0); |
| 1021 | } |
| 1022 | |
| 1023 | static int |
| 1024 | filt_tapread(struct knote *kn, long hint) |
| 1025 | { |
| 1026 | struct tap_softc *tp = (void *)kn->kn_hook; |
| 1027 | |
| 1028 | if (IF_QEMPTY(&tp->tap_devq) == 0) /* XXX serializer */ |
| 1029 | return(1); |
| 1030 | else |
| 1031 | return(0); |
| 1032 | } |
| 1033 | |
| 1034 | static void |
| 1035 | filt_tapdetach(struct knote *kn) |
| 1036 | { |
| 1037 | struct tap_softc *tp = (void *)kn->kn_hook; |
| 1038 | |
| 1039 | SLIST_REMOVE(&tp->tap_rsel.si_note, kn, knote, kn_selnext); |
| 1040 | } |
| 1041 | |
| 1042 | static void |
| 1043 | tapifstop(struct tap_softc *tp, int clear_flags) |
| 1044 | { |
| 1045 | struct ifnet *ifp = &tp->tap_if; |
| 1046 | |
| 1047 | ASSERT_IFNET_SERIALIZED_ALL(ifp); |
| 1048 | IF_DRAIN(&tp->tap_devq); |
| 1049 | tp->tap_flags &= ~TAP_CLOSEDOWN; |
| 1050 | if (clear_flags) |
| 1051 | ifp->if_flags &= ~(IFF_RUNNING | IFF_OACTIVE); |
| 1052 | } |
| 1053 | |
| 1054 | static void |
| 1055 | tapifflags(struct tap_softc *tp) |
| 1056 | { |
| 1057 | struct ifnet *ifp = &tp->arpcom.ac_if; |
| 1058 | |
| 1059 | ASSERT_IFNET_SERIALIZED_ALL(ifp); |
| 1060 | if ((tp->tap_flags & TAP_VMNET) == 0) { |
| 1061 | /* |
| 1062 | * Only for non-vmnet tap(4) |
| 1063 | */ |
| 1064 | if (ifp->if_flags & IFF_UP) { |
| 1065 | if ((ifp->if_flags & IFF_RUNNING) == 0) |
| 1066 | tapifinit(tp); |
| 1067 | } else { |
| 1068 | tapifstop(tp, 1); |
| 1069 | } |
| 1070 | } else { |
| 1071 | /* XXX */ |
| 1072 | } |
| 1073 | } |