sys/net/tap/if_tap.c

   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 1.30 2006/10/25 10:27:27 swildner 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/signalvar.h>
  52 #include <sys/socket.h>
  53 #include <sys/sockio.h>
  54 #include <sys/sysctl.h>
  55 #include <sys/systm.h>
  56 #include <sys/thread2.h>
  57 #include <sys/ttycom.h>
  58 #include <sys/uio.h>
  59 #include <sys/vnode.h>
  60 #include <sys/serialize.h>
  61
  62 #include <net/bpf.h>
  63 #include <net/ethernet.h>
  64 #include <net/if.h>
  65 #include <net/ifq_var.h>
  66 #include <net/if_arp.h>
  67 #include <net/route.h>
  68
  69 #include <netinet/in.h>
  70
  71 #include "if_tapvar.h"
  72 #include "if_tap.h"
  73
  74
  75 #define CDEV_NAME       "tap"
  76 #define CDEV_MAJOR      149
  77 #define TAPDEBUG        if (tapdebug) if_printf
  78
  79 #define TAP             "tap"
  80 #define VMNET           "vmnet"
  81 #define VMNET_DEV_MASK  0x00010000
  82
  83 /* module */
  84 static int              tapmodevent     (module_t, int, void *);
  85
  86 /* device */
  87 static void             tapcreate       (cdev_t);
  88
  89 /* network interface */
  90 static void             tapifstart      (struct ifnet *);
  91 static int              tapifioctl      (struct ifnet *, u_long, caddr_t,
  92                                          struct ucred *);
  93 static void             tapifinit       (void *);
  94
  95 /* character device */
  96 static d_open_t         tapopen;
  97 static d_close_t        tapclose;
  98 static d_read_t         tapread;
  99 static d_write_t        tapwrite;
 100 static d_ioctl_t        tapioctl;
 101 static d_poll_t         tappoll;
 102
 103 static struct dev_ops   tap_ops = {
 104         { CDEV_NAME, CDEV_MAJOR, 0 },
 105         .d_open =       tapopen,
 106         .d_close =      tapclose,
 107         .d_read =       tapread,
 108         .d_write =      tapwrite,
 109         .d_ioctl =      tapioctl,
 110         .d_poll =       tappoll,
 111 };
 112
 113 static int              taprefcnt = 0;          /* module ref. counter   */
 114 static int              taplastunit = -1;       /* max. open unit number */
 115 static int              tapdebug = 0;           /* debug flag            */
 116
 117 MALLOC_DECLARE(M_TAP);
 118 MALLOC_DEFINE(M_TAP, CDEV_NAME, "Ethernet tunnel interface");
 119 SYSCTL_INT(_debug, OID_AUTO, if_tap_debug, CTLFLAG_RW, &tapdebug, 0, "");
 120 DEV_MODULE(if_tap, tapmodevent, NULL);
 121
 122 /*
 123  * tapmodevent
 124  *
 125  * module event handler
 126  */
 127 static int
 128 tapmodevent(module_t mod, int type, void *data)
 129 {
 130         static int               attached = 0;
 131         struct ifnet            *ifp = NULL;
 132         int                      unit;
 133
 134         switch (type) {
 135         case MOD_LOAD:
 136                 if (attached)
 137                         return (EEXIST);
 138
 139                 dev_ops_add(&tap_ops, 0, 0);
 140                 attached = 1;
 141         break;
 142
 143         case MOD_UNLOAD:
 144                 if (taprefcnt > 0)
 145                         return (EBUSY);
 146
 147                 dev_ops_remove(&tap_ops, 0, 0);
 148
 149                 /* XXX: maintain tap ifs in a local list */
 150                 unit = 0;
 151                 while (unit <= taplastunit) {
 152                         TAILQ_FOREACH(ifp, &ifnet, if_link) {
 153                                 if ((strcmp(ifp->if_dname, TAP) == 0) ||
 154                                     (strcmp(ifp->if_dname, VMNET) == 0)) {
 155                                         if (ifp->if_dunit == unit)
 156                                                 break;
 157                                 }
 158                         }
 159
 160                         if (ifp != NULL) {
 161                                 struct tap_softc        *tp = ifp->if_softc;
 162
 163                                 TAPDEBUG(ifp, "detached. minor = %#x, " \
 164                                         "taplastunit = %d\n",
 165                                         minor(tp->tap_dev), taplastunit);
 166
 167                                 ether_ifdetach(ifp);
 168                                 destroy_dev(tp->tap_dev);
 169                                 kfree(tp, M_TAP);
 170                         }
 171                         else
 172                                 unit ++;
 173                 }
 174
 175                 attached = 0;
 176         break;
 177
 178         default:
 179                 return (EOPNOTSUPP);
 180         }
 181
 182         return (0);
 183 } /* tapmodevent */
 184
 185
 186 /*
 187  * tapcreate
 188  *
 189  * to create interface
 190  */
 191 static void
 192 tapcreate(cdev_t dev)
 193 {
 194         struct ifnet            *ifp = NULL;
 195         struct tap_softc        *tp = NULL;
 196         uint8_t                 ether_addr[ETHER_ADDR_LEN];
 197         int                      unit;
 198         char                    *name = NULL;
 199
 200         /* allocate driver storage and create device */
 201         MALLOC(tp, struct tap_softc *, sizeof(*tp), M_TAP, M_WAITOK);
 202         bzero(tp, sizeof(*tp));
 203
 204         /* select device: tap or vmnet */
 205         if (minor(dev) & VMNET_DEV_MASK) {
 206                 name = VMNET;
 207                 unit = lminor(dev) & 0xff;
 208                 tp->tap_flags |= TAP_VMNET;
 209         }
 210         else {
 211                 name = TAP;
 212                 unit = lminor(dev);
 213         }
 214
 215         tp->tap_dev = make_dev(&tap_ops, minor(dev), UID_ROOT, GID_WHEEL,
 216                                                 0600, "%s%d", name, unit);
 217         tp->tap_dev->si_drv1 = dev->si_drv1 = tp;
 218         reference_dev(tp->tap_dev);     /* so we can destroy it later */
 219
 220         /* generate fake MAC address: 00 bd xx xx xx unit_no */
 221         ether_addr[0] = 0x00;
 222         ether_addr[1] = 0xbd;
 223         bcopy(&ticks, ether_addr, 4);
 224         ether_addr[5] = (u_char)unit;
 225
 226         /* fill the rest and attach interface */
 227         ifp = &tp->tap_if;
 228         ifp->if_softc = tp;
 229
 230         if_initname(ifp, name, unit);
 231         if (unit > taplastunit)
 232                 taplastunit = unit;
 233
 234         ifp->if_init = tapifinit;
 235         ifp->if_start = tapifstart;
 236         ifp->if_ioctl = tapifioctl;
 237         ifp->if_mtu = ETHERMTU;
 238         ifp->if_flags = (IFF_BROADCAST|IFF_SIMPLEX|IFF_MULTICAST);
 239         ifq_set_maxlen(&ifp->if_snd, ifqmaxlen);
 240         ifq_set_ready(&ifp->if_snd);
 241
 242         ether_ifattach(ifp, ether_addr, NULL);
 243
 244         tp->tap_flags |= TAP_INITED;
 245
 246         TAPDEBUG(ifp, "created. minor = %#x\n", minor(tp->tap_dev));
 247 } /* tapcreate */
 248
 249
 250 /*
 251  * tapopen
 252  *
 253  * to open tunnel. must be superuser
 254  */
 255 static int
 256 tapopen(struct dev_open_args *ap)
 257 {
 258         cdev_t dev = ap->a_head.a_dev;
 259         struct tap_softc        *tp = NULL;
 260         int                      error;
 261
 262         if ((error = suser_cred(ap->a_cred, 0)) != 0)
 263                 return (error);
 264
 265         tp = dev->si_drv1;
 266         if (tp == NULL) {
 267                 tapcreate(dev);
 268                 tp = dev->si_drv1;
 269         }
 270
 271         if (tp->tap_flags & TAP_OPEN)
 272                 return (EBUSY);
 273
 274         bcopy(tp->arpcom.ac_enaddr, tp->ether_addr, sizeof(tp->ether_addr));
 275
 276         tp->tap_td = curthread;
 277         tp->tap_flags |= TAP_OPEN;
 278         taprefcnt ++;
 279
 280         TAPDEBUG(&tp->arpcom.ac_if,
 281                  "opened. minor = %#x, refcnt = %d, taplastunit = %d\n",
 282                  minor(tp->tap_dev), taprefcnt, taplastunit);
 283
 284         return (0);
 285 } /* tapopen */
 286
 287
 288 /*
 289  * tapclose
 290  *
 291  * close the device - mark i/f down & delete routing info
 292  */
 293 static int
 294 tapclose(struct dev_close_args *ap)
 295 {
 296         cdev_t dev = ap->a_head.a_dev;
 297         struct tap_softc        *tp = dev->si_drv1;
 298         struct ifnet            *ifp = &tp->tap_if;
 299
 300         /* junk all pending output */
 301
 302         lwkt_serialize_enter(ifp->if_serializer);
 303         ifq_purge(&ifp->if_snd);
 304         lwkt_serialize_exit(ifp->if_serializer);
 305
 306         /*
 307          * do not bring the interface down, and do not anything with
 308          * interface, if we are in VMnet mode. just close the device.
 309          */
 310
 311         if (((tp->tap_flags & TAP_VMNET) == 0) && (ifp->if_flags & IFF_UP)) {
 312                 if_down(ifp);
 313                 lwkt_serialize_enter(ifp->if_serializer);
 314                 if (ifp->if_flags & IFF_RUNNING) {
 315                         /* find internet addresses and delete routes */
 316                         struct ifaddr   *ifa = NULL;
 317
 318                         TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
 319                                 if (ifa->ifa_addr->sa_family == AF_INET) {
 320                                         rtinit(ifa, (int)RTM_DELETE, 0);
 321
 322                                         /* remove address from interface */
 323                                         bzero(ifa->ifa_addr,
 324                                                    sizeof(*(ifa->ifa_addr)));
 325                                         bzero(ifa->ifa_dstaddr,
 326                                                    sizeof(*(ifa->ifa_dstaddr)));
 327                                         bzero(ifa->ifa_netmask,
 328                                                    sizeof(*(ifa->ifa_netmask)));
 329                                 }
 330                         }
 331
 332                         ifp->if_flags &= ~IFF_RUNNING;
 333                 }
 334                 lwkt_serialize_exit(ifp->if_serializer);
 335         }
 336
 337         funsetown(tp->tap_sigio);
 338         selwakeup(&tp->tap_rsel);
 339
 340         tp->tap_flags &= ~TAP_OPEN;
 341         tp->tap_td = NULL;
 342
 343         taprefcnt --;
 344         if (taprefcnt < 0) {
 345                 taprefcnt = 0;
 346                 if_printf(ifp, "minor = %#x, refcnt = %d is out of sync. "
 347                         "set refcnt to 0\n", minor(tp->tap_dev), taprefcnt);
 348         }
 349
 350         TAPDEBUG(ifp, "closed. minor = %#x, refcnt = %d, taplastunit = %d\n",
 351                  minor(tp->tap_dev), taprefcnt, taplastunit);
 352
 353         return (0);
 354 } /* tapclose */
 355
 356
 357 /*
 358  * tapifinit
 359  *
 360  * network interface initialization function
 361  */
 362 static void
 363 tapifinit(void *xtp)
 364 {
 365         struct tap_softc        *tp = (struct tap_softc *)xtp;
 366         struct ifnet            *ifp = &tp->tap_if;
 367
 368         TAPDEBUG(ifp, "initializing, minor = %#x\n", minor(tp->tap_dev));
 369
 370         ifp->if_flags |= IFF_RUNNING;
 371         ifp->if_flags &= ~IFF_OACTIVE;
 372
 373         /* attempt to start output */
 374         tapifstart(ifp);
 375 } /* tapifinit */
 376
 377
 378 /*
 379  * tapifioctl
 380  *
 381  * Process an ioctl request on network interface
 382  *
 383  * MPSAFE
 384  */
 385 int
 386 tapifioctl(struct ifnet *ifp, u_long cmd, caddr_t data, struct ucred *cr)
 387 {
 388         struct tap_softc        *tp = (struct tap_softc *)(ifp->if_softc);
 389         struct ifstat           *ifs = NULL;
 390         int                      dummy;
 391
 392         switch (cmd) {
 393                 case SIOCSIFADDR:
 394                 case SIOCGIFADDR:
 395                 case SIOCSIFMTU:
 396                         dummy = ether_ioctl(ifp, cmd, data);
 397                         return (dummy);
 398
 399                 case SIOCSIFFLAGS: /* XXX -- just like vmnet does */
 400                 case SIOCADDMULTI:
 401                 case SIOCDELMULTI:
 402                         break;
 403
 404                 case SIOCGIFSTATUS:
 405                         ifs = (struct ifstat *)data;
 406                         dummy = strlen(ifs->ascii);
 407                         if (tp->tap_td != NULL && dummy < sizeof(ifs->ascii)) {
 408                                 if (tp->tap_td->td_proc) {
 409                                     snprintf(ifs->ascii + dummy,
 410                                         sizeof(ifs->ascii) - dummy,
 411                                         "\tOpened by pid %d\n",
 412                                         (int)tp->tap_td->td_proc->p_pid);
 413                                 } else {
 414                                     snprintf(ifs->ascii + dummy,
 415                                         sizeof(ifs->ascii) - dummy,
 416                                         "\tOpened by td %p\n", tp->tap_td);
 417                                 }
 418                         }
 419                         break;
 420
 421                 default:
 422                         return (EINVAL);
 423         }
 424
 425         return (0);
 426 } /* tapifioctl */
 427
 428
 429 /*
 430  * tapifstart
 431  *
 432  * queue packets from higher level ready to put out
 433  */
 434 static void
 435 tapifstart(struct ifnet *ifp)
 436 {
 437         struct tap_softc        *tp = ifp->if_softc;
 438
 439         TAPDEBUG(ifp, "starting, minor = %#x\n", minor(tp->tap_dev));
 440
 441         /*
 442          * do not junk pending output if we are in VMnet mode.
 443          * XXX: can this do any harm because of queue overflow?
 444          */
 445
 446         if (((tp->tap_flags & TAP_VMNET) == 0) &&
 447             ((tp->tap_flags & TAP_READY) != TAP_READY)) {
 448                 TAPDEBUG(ifp, "not ready. minor = %#x, tap_flags = 0x%x\n",
 449                          minor(tp->tap_dev), tp->tap_flags);
 450
 451                 ifq_purge(&ifp->if_snd);
 452                 return;
 453         }
 454
 455         ifp->if_flags |= IFF_OACTIVE;
 456
 457         if (!ifq_is_empty(&ifp->if_snd)) {
 458                 if (tp->tap_flags & TAP_RWAIT) {
 459                         tp->tap_flags &= ~TAP_RWAIT;
 460                         wakeup((caddr_t)tp);
 461                 }
 462
 463                 if ((tp->tap_flags & TAP_ASYNC) && (tp->tap_sigio != NULL))
 464                         pgsigio(tp->tap_sigio, SIGIO, 0);
 465
 466                 /*
 467                  * selwakeup is not MPSAFE.  tapifstart is.
 468                  */
 469                 get_mplock();
 470                 selwakeup(&tp->tap_rsel);
 471                 rel_mplock();
 472                 ifp->if_opackets ++; /* obytes are counted in ether_output */
 473         }
 474
 475         ifp->if_flags &= ~IFF_OACTIVE;
 476 } /* tapifstart */
 477
 478
 479 /*
 480  * tapioctl
 481  *
 482  * the ops interface is now pretty minimal
 483  */
 484 static int
 485 tapioctl(struct dev_ioctl_args *ap)
 486 {
 487         cdev_t dev = ap->a_head.a_dev;
 488         caddr_t data = ap->a_data;
 489         struct tap_softc        *tp = dev->si_drv1;
 490         struct ifnet            *ifp = &tp->tap_if;
 491         struct tapinfo          *tapp = NULL;
 492         struct mbuf *mb;
 493         short f;
 494         int error;
 495
 496         lwkt_serialize_enter(ifp->if_serializer);
 497         error = 0;
 498
 499         switch (ap->a_cmd) {
 500         case TAPSIFINFO:
 501                 tapp = (struct tapinfo *)data;
 502                 ifp->if_mtu = tapp->mtu;
 503                 ifp->if_type = tapp->type;
 504                 ifp->if_baudrate = tapp->baudrate;
 505                 break;
 506
 507         case TAPGIFINFO:
 508                 tapp = (struct tapinfo *)data;
 509                 tapp->mtu = ifp->if_mtu;
 510                 tapp->type = ifp->if_type;
 511                 tapp->baudrate = ifp->if_baudrate;
 512                 break;
 513
 514         case TAPSDEBUG:
 515                 tapdebug = *(int *)data;
 516                 break;
 517
 518         case TAPGDEBUG:
 519                 *(int *)data = tapdebug;
 520                 break;
 521
 522         case FIOASYNC:
 523                 if (*(int *)data)
 524                         tp->tap_flags |= TAP_ASYNC;
 525                 else
 526                         tp->tap_flags &= ~TAP_ASYNC;
 527                 break;
 528
 529         case FIONREAD:
 530                 *(int *)data = 0;
 531                 if ((mb = ifq_poll(&ifp->if_snd)) != NULL) {
 532                         for(; mb != NULL; mb = mb->m_next)
 533                                 *(int *)data += mb->m_len;
 534                 }
 535                 break;
 536
 537         case FIOSETOWN:
 538                 error = fsetown(*(int *)data, &tp->tap_sigio);
 539                 break;
 540
 541         case FIOGETOWN:
 542                 *(int *)data = fgetown(tp->tap_sigio);
 543                 break;
 544
 545         /* this is deprecated, FIOSETOWN should be used instead */
 546         case TIOCSPGRP:
 547                 error = fsetown(-(*(int *)data), &tp->tap_sigio);
 548                 break;
 549
 550         /* this is deprecated, FIOGETOWN should be used instead */
 551         case TIOCGPGRP:
 552                 *(int *)data = -fgetown(tp->tap_sigio);
 553                 break;
 554
 555         /* VMware/VMnet port ioctl's */
 556
 557         case SIOCGIFFLAGS:      /* get ifnet flags */
 558                 bcopy(&ifp->if_flags, data, sizeof(ifp->if_flags));
 559                 break;
 560
 561         case VMIO_SIOCSIFFLAGS: /* VMware/VMnet SIOCSIFFLAGS */
 562                 f = *(short *)data;
 563                 f &= 0x0fff;
 564                 f &= ~IFF_CANTCHANGE;
 565                 f |= IFF_UP;
 566                 ifp->if_flags = f | (ifp->if_flags & IFF_CANTCHANGE);
 567                 break;
 568
 569         case OSIOCGIFADDR:      /* get MAC address of the remote side */
 570         case SIOCGIFADDR:
 571                 bcopy(tp->ether_addr, data, sizeof(tp->ether_addr));
 572                 break;
 573
 574         case SIOCSIFADDR:       /* set MAC address of the remote side */
 575                 bcopy(data, tp->ether_addr, sizeof(tp->ether_addr));
 576                 break;
 577
 578         default:
 579                 error = ENOTTY;
 580                 break;
 581         }
 582         lwkt_serialize_exit(ifp->if_serializer);
 583         return (error);
 584 } /* tapioctl */
 585
 586
 587 /*
 588  * tapread
 589  *
 590  * the ops read interface - reads a packet at a time, or at
 591  * least as much of a packet as can be read
 592  */
 593 static int
 594 tapread(struct dev_read_args *ap)
 595 {
 596         cdev_t dev = ap->a_head.a_dev;
 597         struct uio *uio = ap->a_uio;
 598         struct tap_softc        *tp = dev->si_drv1;
 599         struct ifnet            *ifp = &tp->tap_if;
 600         struct mbuf             *m0 = NULL;
 601         int                      error = 0, len;
 602
 603         TAPDEBUG(ifp, "reading, minor = %#x\n", minor(tp->tap_dev));
 604
 605         if ((tp->tap_flags & TAP_READY) != TAP_READY) {
 606                 TAPDEBUG(ifp, "not ready. minor = %#x, tap_flags = 0x%x\n",
 607                          minor(tp->tap_dev), tp->tap_flags);
 608
 609                 return (EHOSTDOWN);
 610         }
 611
 612         tp->tap_flags &= ~TAP_RWAIT;
 613
 614         /* sleep until we get a packet */
 615         do {
 616                 lwkt_serialize_enter(ifp->if_serializer);
 617                 m0 = ifq_dequeue(&ifp->if_snd, NULL);
 618                 if (m0 == NULL) {
 619                         if (ap->a_ioflag & IO_NDELAY) {
 620                                 lwkt_serialize_exit(ifp->if_serializer);
 621                                 return (EWOULDBLOCK);
 622                         }
 623                         tp->tap_flags |= TAP_RWAIT;
 624                         crit_enter();
 625                         tsleep_interlock(tp);
 626                         lwkt_serialize_exit(ifp->if_serializer);
 627                         error = tsleep(tp, PCATCH, "taprd", 0);
 628                         crit_exit();
 629                         if (error)
 630                                 return (error);
 631                 } else {
 632                         lwkt_serialize_exit(ifp->if_serializer);
 633                 }
 634         } while (m0 == NULL);
 635
 636         BPF_MTAP(ifp, m0);
 637
 638         /* xfer packet to user space */
 639         while ((m0 != NULL) && (uio->uio_resid > 0) && (error == 0)) {
 640                 len = min(uio->uio_resid, m0->m_len);
 641                 if (len == 0)
 642                         break;
 643
 644                 error = uiomove(mtod(m0, caddr_t), len, uio);
 645                 m0 = m_free(m0);
 646         }
 647
 648         if (m0 != NULL) {
 649                 TAPDEBUG(ifp, "dropping mbuf, minor = %#x\n",
 650                          minor(tp->tap_dev));
 651                 m_freem(m0);
 652         }
 653
 654         return (error);
 655 } /* tapread */
 656
 657
 658 /*
 659  * tapwrite
 660  *
 661  * the ops write interface - an atomic write is a packet - or else!
 662  */
 663 static int
 664 tapwrite(struct dev_write_args *ap)
 665 {
 666         cdev_t dev = ap->a_head.a_dev;
 667         struct uio *uio = ap->a_uio;
 668         struct tap_softc        *tp = dev->si_drv1;
 669         struct ifnet            *ifp = &tp->tap_if;
 670         struct mbuf             *top = NULL, **mp = NULL, *m = NULL;
 671         int                      error = 0, tlen, mlen;
 672
 673         TAPDEBUG(ifp, "writing, minor = %#x\n", minor(tp->tap_dev));
 674
 675         if (uio->uio_resid == 0)
 676                 return (0);
 677
 678         if ((uio->uio_resid < 0) || (uio->uio_resid > TAPMRU)) {
 679                 TAPDEBUG(ifp, "invalid packet len = %d, minor = %#x\n",
 680                          uio->uio_resid, minor(tp->tap_dev));
 681
 682                 return (EIO);
 683         }
 684         tlen = uio->uio_resid;
 685
 686         /* get a header mbuf */
 687         MGETHDR(m, MB_DONTWAIT, MT_DATA);
 688         if (m == NULL)
 689                 return (ENOBUFS);
 690         mlen = MHLEN;
 691
 692         top = 0;
 693         mp = &top;
 694         while ((error == 0) && (uio->uio_resid > 0)) {
 695                 m->m_len = min(mlen, uio->uio_resid);
 696                 error = uiomove(mtod(m, caddr_t), m->m_len, uio);
 697                 *mp = m;
 698                 mp = &m->m_next;
 699                 if (uio->uio_resid > 0) {
 700                         MGET(m, MB_DONTWAIT, MT_DATA);
 701                         if (m == NULL) {
 702                                 error = ENOBUFS;
 703                                 break;
 704                         }
 705                         mlen = MLEN;
 706                 }
 707         }
 708         if (error) {
 709                 ifp->if_ierrors ++;
 710                 if (top)
 711                         m_freem(top);
 712                 return (error);
 713         }
 714
 715         top->m_pkthdr.len = tlen;
 716         top->m_pkthdr.rcvif = ifp;
 717
 718         /*
 719          * Ethernet bridge and bpf are handled in ether_input
 720          *
 721          * adjust mbuf and give packet to the ether_input
 722          */
 723         lwkt_serialize_enter(ifp->if_serializer);
 724         ifp->if_input(ifp, top);
 725         ifp->if_ipackets ++; /* ibytes are counted in ether_input */
 726         lwkt_serialize_exit(ifp->if_serializer);
 727
 728         return (0);
 729 } /* tapwrite */
 730
 731
 732 /*
 733  * tappoll
 734  *
 735  * the poll interface, this is only useful on reads
 736  * really. the write detect always returns true, write never blocks
 737  * anyway, it either accepts the packet or drops it
 738  */
 739 static int
 740 tappoll(struct dev_poll_args *ap)
 741 {
 742         cdev_t dev = ap->a_head.a_dev;
 743         struct tap_softc        *tp = dev->si_drv1;
 744         struct ifnet            *ifp = &tp->tap_if;
 745         int                      revents = 0;
 746
 747         TAPDEBUG(ifp, "polling, minor = %#x\n", minor(tp->tap_dev));
 748
 749         lwkt_serialize_enter(ifp->if_serializer);
 750         if (ap->a_events & (POLLIN | POLLRDNORM)) {
 751                 if (!ifq_is_empty(&ifp->if_snd)) {
 752                         TAPDEBUG(ifp,
 753                                  "has data in queue. minor = %#x\n",
 754                                  minor(tp->tap_dev));
 755
 756                         revents |= (ap->a_events & (POLLIN | POLLRDNORM));
 757                 }
 758                 else {
 759                         TAPDEBUG(ifp, "waiting for data, minor = %#x\n",
 760                                  minor(tp->tap_dev));
 761
 762                         selrecord(curthread, &tp->tap_rsel);
 763                 }
 764         }
 765         lwkt_serialize_exit(ifp->if_serializer);
 766
 767         if (ap->a_events & (POLLOUT | POLLWRNORM))
 768                 revents |= (ap->a_events & (POLLOUT | POLLWRNORM));
 769         ap->a_events = revents;
 770         return(0);
 771 } /* tappoll */