The Design and Implementation of the FreeBSD Operating System, Second Edition
Now available: The Design and Implementation of the FreeBSD Operating System (Second Edition)


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]

FreeBSD/Linux Kernel Cross Reference
sys/geom/geom_dev.c

Version: -  FREEBSD  -  FREEBSD-13-STABLE  -  FREEBSD-13-0  -  FREEBSD-12-STABLE  -  FREEBSD-12-0  -  FREEBSD-11-STABLE  -  FREEBSD-11-0  -  FREEBSD-10-STABLE  -  FREEBSD-10-0  -  FREEBSD-9-STABLE  -  FREEBSD-9-0  -  FREEBSD-8-STABLE  -  FREEBSD-8-0  -  FREEBSD-7-STABLE  -  FREEBSD-7-0  -  FREEBSD-6-STABLE  -  FREEBSD-6-0  -  FREEBSD-5-STABLE  -  FREEBSD-5-0  -  FREEBSD-4-STABLE  -  FREEBSD-3-STABLE  -  FREEBSD22  -  l41  -  OPENBSD  -  linux-2.6  -  MK84  -  PLAN9  -  xnu-8792 
SearchContext: -  none  -  3  -  10 

    1 /*-
    2  * Copyright (c) 2002 Poul-Henning Kamp
    3  * Copyright (c) 2002 Networks Associates Technology, Inc.
    4  * All rights reserved.
    5  *
    6  * This software was developed for the FreeBSD Project by Poul-Henning Kamp
    7  * and NAI Labs, the Security Research Division of Network Associates, Inc.
    8  * under DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the
    9  * DARPA CHATS research program.
   10  *
   11  * Redistribution and use in source and binary forms, with or without
   12  * modification, are permitted provided that the following conditions
   13  * are met:
   14  * 1. Redistributions of source code must retain the above copyright
   15  *    notice, this list of conditions and the following disclaimer.
   16  * 2. Redistributions in binary form must reproduce the above copyright
   17  *    notice, this list of conditions and the following disclaimer in the
   18  *    documentation and/or other materials provided with the distribution.
   19  * 3. The names of the authors may not be used to endorse or promote
   20  *    products derived from this software without specific prior written
   21  *    permission.
   22  *
   23  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
   24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
   25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
   26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
   27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
   28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
   29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
   30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
   31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
   32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
   33  * SUCH DAMAGE.
   34  */
   35 
   36 #include <sys/cdefs.h>
   37 __FBSDID("$FreeBSD: releng/7.4/sys/geom/geom_dev.c 212879 2010-09-19 20:02:31Z mav $");
   38 
   39 #include <sys/param.h>
   40 #include <sys/systm.h>
   41 #include <sys/malloc.h>
   42 #include <sys/kernel.h>
   43 #include <sys/conf.h>
   44 #include <sys/bio.h>
   45 #include <sys/lock.h>
   46 #include <sys/mutex.h>
   47 #include <sys/proc.h>
   48 #include <sys/errno.h>
   49 #include <sys/time.h>
   50 #include <sys/disk.h>
   51 #include <sys/fcntl.h>
   52 #include <sys/limits.h>
   53 #include <geom/geom.h>
   54 #include <geom/geom_int.h>
   55 
   56 static d_open_t         g_dev_open;
   57 static d_close_t        g_dev_close;
   58 static d_strategy_t     g_dev_strategy;
   59 static d_ioctl_t        g_dev_ioctl;
   60 
   61 static struct cdevsw g_dev_cdevsw = {
   62         .d_version =    D_VERSION,
   63         .d_open =       g_dev_open,
   64         .d_close =      g_dev_close,
   65         .d_read =       physread,
   66         .d_write =      physwrite,
   67         .d_ioctl =      g_dev_ioctl,
   68         .d_strategy =   g_dev_strategy,
   69         .d_name =       "g_dev",
   70         .d_flags =      D_DISK | D_TRACKCLOSE,
   71 };
   72 
   73 static g_taste_t g_dev_taste;
   74 static g_orphan_t g_dev_orphan;
   75 static g_init_t         g_dev_init;
   76 
   77 static struct g_class g_dev_class       = {
   78         .name = "DEV",
   79         .version = G_VERSION,
   80         .taste = g_dev_taste,
   81         .orphan = g_dev_orphan,
   82         .init = g_dev_init,
   83 };
   84 
   85 static struct unrhdr *unithdr;  /* Locked by topology */
   86 
   87 static void
   88 g_dev_init(struct g_class *mp)
   89 {
   90 
   91         unithdr = new_unrhdr(0, minor2unit(MAXMINOR), NULL);
   92 }
   93 
   94 void
   95 g_dev_print(void)
   96 {
   97         struct g_geom *gp;
   98         char const *p = "";
   99 
  100         LIST_FOREACH(gp, &g_dev_class.geom, geom) {
  101                 printf("%s%s", p, gp->name);
  102                 p = " ";
  103         }
  104         printf("\n");
  105 }
  106 
  107 struct g_provider *
  108 g_dev_getprovider(struct cdev *dev)
  109 {
  110         struct g_consumer *cp;
  111 
  112         g_topology_assert();
  113         if (dev == NULL)
  114                 return (NULL);
  115         if (dev->si_devsw != &g_dev_cdevsw)
  116                 return (NULL);
  117         cp = dev->si_drv2;
  118         return (cp->provider);
  119 }
  120 
  121 
  122 static struct g_geom *
  123 g_dev_taste(struct g_class *mp, struct g_provider *pp, int insist __unused)
  124 {
  125         struct g_geom *gp;
  126         struct g_consumer *cp;
  127         char *alias;
  128         int error;
  129         struct cdev *dev;
  130         u_int unit;
  131 
  132         g_trace(G_T_TOPOLOGY, "dev_taste(%s,%s)", mp->name, pp->name);
  133         g_topology_assert();
  134         LIST_FOREACH(cp, &pp->consumers, consumers)
  135                 if (cp->geom->class == mp)
  136                         return (NULL);
  137         gp = g_new_geomf(mp, pp->name);
  138         cp = g_new_consumer(gp);
  139         error = g_attach(cp, pp);
  140         KASSERT(error == 0,
  141             ("g_dev_taste(%s) failed to g_attach, err=%d", pp->name, error));
  142         unit = alloc_unr(unithdr);
  143         dev = make_dev(&g_dev_cdevsw, unit2minor(unit),
  144             UID_ROOT, GID_OPERATOR, 0640, gp->name);
  145         if (pp->flags & G_PF_CANDELETE)
  146                 dev->si_flags |= SI_CANDELETE;
  147         dev->si_iosize_max = MAXPHYS;
  148         gp->softc = dev;
  149         dev->si_drv1 = gp;
  150         dev->si_drv2 = cp;
  151 
  152         g_topology_unlock();
  153 
  154         alias = g_malloc(MAXPATHLEN, M_WAITOK | M_ZERO);
  155         error = (pp->geom->ioctl == NULL) ? ENODEV :
  156             pp->geom->ioctl(pp, DIOCGPROVIDERALIAS, alias, 0, curthread);
  157         if (!error && alias[0] != '\0')
  158                 make_dev_alias(dev, "%s", alias);
  159         g_free(alias);
  160 
  161         g_topology_lock();
  162         return (gp);
  163 }
  164 
  165 static int
  166 g_dev_open(struct cdev *dev, int flags, int fmt, struct thread *td)
  167 {
  168         struct g_geom *gp;
  169         struct g_consumer *cp;
  170         int error, r, w, e;
  171 
  172         gp = dev->si_drv1;
  173         cp = dev->si_drv2;
  174         if (gp == NULL || cp == NULL || gp->softc != dev)
  175                 return(ENXIO);          /* g_dev_taste() not done yet */
  176 
  177         g_trace(G_T_ACCESS, "g_dev_open(%s, %d, %d, %p)",
  178             gp->name, flags, fmt, td);
  179 
  180         r = flags & FREAD ? 1 : 0;
  181         w = flags & FWRITE ? 1 : 0;
  182 #ifdef notyet
  183         e = flags & O_EXCL ? 1 : 0;
  184 #else
  185         e = 0;
  186 #endif
  187         if (w) {
  188                 /*
  189                  * When running in very secure mode, do not allow
  190                  * opens for writing of any disks.
  191                  */
  192                 error = securelevel_ge(td->td_ucred, 2);
  193                 if (error)
  194                         return (error);
  195         }
  196         g_topology_lock();
  197         if (dev->si_devsw == NULL)
  198                 error = ENXIO;          /* We were orphaned */
  199         else
  200                 error = g_access(cp, r, w, e);
  201         g_topology_unlock();
  202         return(error);
  203 }
  204 
  205 static int
  206 g_dev_close(struct cdev *dev, int flags, int fmt, struct thread *td)
  207 {
  208         struct g_geom *gp;
  209         struct g_consumer *cp;
  210         int error, r, w, e, i;
  211 
  212         gp = dev->si_drv1;
  213         cp = dev->si_drv2;
  214         if (gp == NULL || cp == NULL)
  215                 return(ENXIO);
  216         g_trace(G_T_ACCESS, "g_dev_close(%s, %d, %d, %p)",
  217             gp->name, flags, fmt, td);
  218         r = flags & FREAD ? -1 : 0;
  219         w = flags & FWRITE ? -1 : 0;
  220 #ifdef notyet
  221         e = flags & O_EXCL ? -1 : 0;
  222 #else
  223         e = 0;
  224 #endif
  225         g_topology_lock();
  226         if (dev->si_devsw == NULL)
  227                 error = ENXIO;          /* We were orphaned */
  228         else
  229                 error = g_access(cp, r, w, e);
  230         for (i = 0; i < 10 * hz;) {
  231                 if (cp->acr != 0 || cp->acw != 0)
  232                         break;
  233                 if (cp->nstart == cp->nend)
  234                         break;
  235                 pause("gdevwclose", hz / 10);
  236                 i += hz / 10;
  237         }
  238         if (cp->acr == 0 && cp->acw == 0 && cp->nstart != cp->nend) {
  239                 printf("WARNING: Final close of geom_dev(%s) %s %s\n",
  240                     gp->name,
  241                     "still has outstanding I/O after 10 seconds.",
  242                     "Completing close anyway, panic may happen later.");
  243         }
  244         g_topology_unlock();
  245         return (error);
  246 }
  247 
  248 /*
  249  * XXX: Until we have unmessed the ioctl situation, there is a race against
  250  * XXX: a concurrent orphanization.  We cannot close it by holding topology
  251  * XXX: since that would prevent us from doing our job, and stalling events
  252  * XXX: will break (actually: stall) the BSD disklabel hacks.
  253  */
  254 static int
  255 g_dev_ioctl(struct cdev *dev, u_long cmd, caddr_t data, int fflag, struct thread *td)
  256 {
  257         struct g_geom *gp;
  258         struct g_consumer *cp;
  259         struct g_kerneldump kd;
  260         off_t offset, length;
  261         int i, error;
  262         u_int u;
  263 
  264         gp = dev->si_drv1;
  265         cp = dev->si_drv2;
  266 
  267         error = 0;
  268         KASSERT(cp->acr || cp->acw,
  269             ("Consumer with zero access count in g_dev_ioctl"));
  270 
  271         i = IOCPARM_LEN(cmd);
  272         switch (cmd) {
  273         case DIOCGSECTORSIZE:
  274                 *(u_int *)data = cp->provider->sectorsize;
  275                 if (*(u_int *)data == 0)
  276                         error = ENOENT;
  277                 break;
  278         case DIOCGMEDIASIZE:
  279                 *(off_t *)data = cp->provider->mediasize;
  280                 if (*(off_t *)data == 0)
  281                         error = ENOENT;
  282                 break;
  283         case DIOCGFWSECTORS:
  284                 error = g_io_getattr("GEOM::fwsectors", cp, &i, data);
  285                 if (error == 0 && *(u_int *)data == 0)
  286                         error = ENOENT;
  287                 break;
  288         case DIOCGFWHEADS:
  289                 error = g_io_getattr("GEOM::fwheads", cp, &i, data);
  290                 if (error == 0 && *(u_int *)data == 0)
  291                         error = ENOENT;
  292                 break;
  293         case DIOCGFRONTSTUFF:
  294                 error = g_io_getattr("GEOM::frontstuff", cp, &i, data);
  295                 break;
  296         case DIOCSKERNELDUMP:
  297                 u = *((u_int *)data);
  298                 if (!u) {
  299                         set_dumper(NULL);
  300                         error = 0;
  301                         break;
  302                 }
  303                 kd.offset = 0;
  304                 kd.length = OFF_MAX;
  305                 i = sizeof kd;
  306                 error = g_io_getattr("GEOM::kerneldump", cp, &i, &kd);
  307                 if (!error)
  308                         dev->si_flags |= SI_DUMPDEV;
  309                 break;
  310         case DIOCGFLUSH:
  311                 error = g_io_flush(cp);
  312                 break;
  313         case DIOCGDELETE:
  314                 offset = ((off_t *)data)[0];
  315                 length = ((off_t *)data)[1];
  316                 if ((offset % cp->provider->sectorsize) != 0 ||
  317                     (length % cp->provider->sectorsize) != 0 || length <= 0) {
  318                         printf("%s: offset=%jd length=%jd\n", __func__, offset,
  319                             length);
  320                         error = EINVAL;
  321                         break;
  322                 }
  323                 error = g_delete_data(cp, offset, length);
  324                 break;
  325         case DIOCGIDENT:
  326                 error = g_io_getattr("GEOM::ident", cp, &i, data);
  327                 break;
  328 
  329         default:
  330                 if (cp->provider->geom->ioctl != NULL) {
  331                         error = cp->provider->geom->ioctl(cp->provider, cmd, data, fflag, td);
  332                 } else {
  333                         error = ENOIOCTL;
  334                 }
  335         }
  336 
  337         return (error);
  338 }
  339 
  340 static void
  341 g_dev_done(struct bio *bp2)
  342 {
  343         struct bio *bp;
  344 
  345         bp = bp2->bio_parent;
  346         bp->bio_error = bp2->bio_error;
  347         if (bp->bio_error != 0) {
  348                 g_trace(G_T_BIO, "g_dev_done(%p) had error %d",
  349                     bp2, bp->bio_error);
  350                 bp->bio_flags |= BIO_ERROR;
  351         } else {
  352                 g_trace(G_T_BIO, "g_dev_done(%p/%p) resid %ld completed %jd",
  353                     bp2, bp, bp->bio_resid, (intmax_t)bp2->bio_completed);
  354         }
  355         bp->bio_resid = bp->bio_length - bp2->bio_completed;
  356         bp->bio_completed = bp2->bio_completed;
  357         g_destroy_bio(bp2);
  358         biodone(bp);
  359 }
  360 
  361 static void
  362 g_dev_strategy(struct bio *bp)
  363 {
  364         struct g_consumer *cp;
  365         struct bio *bp2;
  366         struct cdev *dev;
  367 
  368         KASSERT(bp->bio_cmd == BIO_READ ||
  369                 bp->bio_cmd == BIO_WRITE ||
  370                 bp->bio_cmd == BIO_DELETE,
  371                 ("Wrong bio_cmd bio=%p cmd=%d", bp, bp->bio_cmd));
  372         dev = bp->bio_dev;
  373         cp = dev->si_drv2;
  374         KASSERT(cp->acr || cp->acw,
  375             ("Consumer with zero access count in g_dev_strategy"));
  376 #ifdef INVARIANTS
  377         if ((bp->bio_offset % cp->provider->sectorsize) != 0 ||
  378             (bp->bio_bcount % cp->provider->sectorsize) != 0) {
  379                 bp->bio_resid = bp->bio_bcount;
  380                 biofinish(bp, NULL, EINVAL);
  381                 return;
  382         }
  383 #endif
  384         for (;;) {
  385                 /*
  386                  * XXX: This is not an ideal solution, but I belive it to
  387                  * XXX: deadlock safe, all things considered.
  388                  */
  389                 bp2 = g_clone_bio(bp);
  390                 if (bp2 != NULL)
  391                         break;
  392                 pause("gdstrat", hz / 10);
  393         }
  394         KASSERT(bp2 != NULL, ("XXX: ENOMEM in a bad place"));
  395         bp2->bio_done = g_dev_done;
  396         g_trace(G_T_BIO,
  397             "g_dev_strategy(%p/%p) offset %jd length %jd data %p cmd %d",
  398             bp, bp2, (intmax_t)bp->bio_offset, (intmax_t)bp2->bio_length,
  399             bp2->bio_data, bp2->bio_cmd);
  400         g_io_request(bp2, cp);
  401         KASSERT(cp->acr || cp->acw,
  402             ("g_dev_strategy raced with g_dev_close and lost"));
  403 
  404 }
  405 
  406 /*
  407  * g_dev_orphan()
  408  *
  409  * Called from below when the provider orphaned us.
  410  * - Clear any dump settings.
  411  * - Destroy the struct cdev *to prevent any more request from coming in.  The
  412  *   provider is already marked with an error, so anything which comes in
  413  *   in the interrim will be returned immediately.
  414  * - Wait for any outstanding I/O to finish.
  415  * - Set our access counts to zero, whatever they were.
  416  * - Detach and self-destruct.
  417  */
  418 
  419 static void
  420 g_dev_orphan(struct g_consumer *cp)
  421 {
  422         struct g_geom *gp;
  423         struct cdev *dev;
  424         u_int unit;
  425 
  426         g_topology_assert();
  427         gp = cp->geom;
  428         dev = gp->softc;
  429         g_trace(G_T_TOPOLOGY, "g_dev_orphan(%p(%s))", cp, gp->name);
  430 
  431         /* Reset any dump-area set on this device */
  432         if (dev->si_flags & SI_DUMPDEV)
  433                 set_dumper(NULL);
  434 
  435         /* Destroy the struct cdev *so we get no more requests */
  436         unit = dev2unit(dev);
  437         destroy_dev(dev);
  438         free_unr(unithdr, unit);
  439 
  440         /* Wait for the cows to come home */
  441         while (cp->nstart != cp->nend)
  442                 pause("gdevorphan", hz / 10);
  443 
  444         if (cp->acr > 0 || cp->acw > 0 || cp->ace > 0)
  445                 g_access(cp, -cp->acr, -cp->acw, -cp->ace);
  446 
  447         g_detach(cp);
  448         g_destroy_consumer(cp);
  449         g_destroy_geom(gp);
  450 }
  451 
  452 DECLARE_GEOM_CLASS(g_dev_class, g_dev);

Cache object: d0e934f419709b3e2dde3bd747c4bffc


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]


This page is part of the FreeBSD/Linux Linux Kernel Cross-Reference, and was automatically generated using a modified version of the LXR engine.