module/zfs/vdev_file.c

   1 /*
   2  * CDDL HEADER START
   3  *
   4  * The contents of this file are subject to the terms of the
   5  * Common Development and Distribution License (the "License").
   6  * You may not use this file except in compliance with the License.
   7  *
   8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
   9  * or http://www.opensolaris.org/os/licensing.
  10  * See the License for the specific language governing permissions
  11  * and limitations under the License.
  12  *
  13  * When distributing Covered Code, include this CDDL HEADER in each
  14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  15  * If applicable, add the following below this CDDL HEADER, with the
  16  * fields enclosed by brackets "[]" replaced with your own identifying
  17  * information: Portions Copyright [yyyy] [name of copyright owner]
  18  *
  19  * CDDL HEADER END
  20  */
  21 /*
  22  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
  23  * Copyright (c) 2012 by Delphix. All rights reserved.
  24  */
  25
  26 #include <sys/zfs_context.h>
  27 #include <sys/spa.h>
  28 #include <sys/vdev_file.h>
  29 #include <sys/vdev_impl.h>
  30 #include <sys/zio.h>
  31 #include <sys/fs/zfs.h>
  32 #include <sys/fm/fs/zfs.h>
  33
  34 /*
  35  * Virtual device vector for files.
  36  */
  37
  38 static void
  39 vdev_file_hold(vdev_t *vd)
  40 {
  41         ASSERT(vd->vdev_path != NULL);
  42 }
  43
  44 static void
  45 vdev_file_rele(vdev_t *vd)
  46 {
  47         ASSERT(vd->vdev_path != NULL);
  48 }
  49
  50 static int
  51 vdev_file_open(vdev_t *vd, uint64_t *psize, uint64_t *max_psize,
  52     uint64_t *ashift)
  53 {
  54         vdev_file_t *vf;
  55         vnode_t *vp;
  56         vattr_t vattr;
  57         int error;
  58
  59         /*
  60          * We must have a pathname, and it must be absolute.
  61          */
  62         if (vd->vdev_path == NULL || vd->vdev_path[0] != '/') {
  63                 vd->vdev_stat.vs_aux = VDEV_AUX_BAD_LABEL;
  64                 return (EINVAL);
  65         }
  66
  67         /*
  68          * Reopen the device if it's not currently open.  Otherwise,
  69          * just update the physical size of the device.
  70          */
  71         if (vd->vdev_tsd != NULL) {
  72                 ASSERT(vd->vdev_reopening);
  73                 vf = vd->vdev_tsd;
  74                 goto skip_open;
  75         }
  76
  77         vf = vd->vdev_tsd = kmem_zalloc(sizeof (vdev_file_t), KM_PUSHPAGE);
  78
  79         /*
  80          * We always open the files from the root of the global zone, even if
  81          * we're in a local zone.  If the user has gotten to this point, the
  82          * administrator has already decided that the pool should be available
  83          * to local zone users, so the underlying devices should be as well.
  84          */
  85         ASSERT(vd->vdev_path != NULL && vd->vdev_path[0] == '/');
  86         error = vn_openat(vd->vdev_path + 1, UIO_SYSSPACE,
  87             spa_mode(vd->vdev_spa) | FOFFMAX, 0, &vp, 0, 0, rootdir, -1);
  88
  89         if (error) {
  90                 vd->vdev_stat.vs_aux = VDEV_AUX_OPEN_FAILED;
  91                 return (error);
  92         }
  93
  94         vf->vf_vnode = vp;
  95
  96 #ifdef _KERNEL
  97         /*
  98          * Make sure it's a regular file.
  99          */
 100         if (vp->v_type != VREG) {
 101                 vd->vdev_stat.vs_aux = VDEV_AUX_OPEN_FAILED;
 102                 return (ENODEV);
 103         }
 104 #endif
 105
 106 skip_open:
 107         /*
 108          * Determine the physical size of the file.
 109          */
 110         vattr.va_mask = AT_SIZE;
 111         error = VOP_GETATTR(vf->vf_vnode, &vattr, 0, kcred, NULL);
 112         if (error) {
 113                 vd->vdev_stat.vs_aux = VDEV_AUX_OPEN_FAILED;
 114                 return (error);
 115         }
 116
 117         *max_psize = *psize = vattr.va_size;
 118         *ashift = SPA_MINBLOCKSHIFT;
 119
 120         return (0);
 121 }
 122
 123 static void
 124 vdev_file_close(vdev_t *vd)
 125 {
 126         vdev_file_t *vf = vd->vdev_tsd;
 127
 128         if (vd->vdev_reopening || vf == NULL)
 129                 return;
 130
 131         if (vf->vf_vnode != NULL) {
 132                 (void) VOP_PUTPAGE(vf->vf_vnode, 0, 0, B_INVAL, kcred, NULL);
 133                 (void) VOP_CLOSE(vf->vf_vnode, spa_mode(vd->vdev_spa), 1, 0,
 134                     kcred, NULL);
 135         }
 136
 137         vd->vdev_delayed_close = B_FALSE;
 138         kmem_free(vf, sizeof (vdev_file_t));
 139         vd->vdev_tsd = NULL;
 140 }
 141
 142 static int
 143 vdev_file_io_start(zio_t *zio)
 144 {
 145         vdev_t *vd = zio->io_vd;
 146         vdev_file_t *vf;
 147         ssize_t resid = 0;
 148
 149         if (!vdev_readable(vd)) {
 150                 zio->io_error = ENXIO;
 151                 return (ZIO_PIPELINE_CONTINUE);
 152         }
 153
 154         vf = vd->vdev_tsd;
 155
 156         if (zio->io_type == ZIO_TYPE_IOCTL) {
 157                 switch (zio->io_cmd) {
 158                 case DKIOCFLUSHWRITECACHE:
 159                         zio->io_error = VOP_FSYNC(vf->vf_vnode, FSYNC | FDSYNC,
 160                             kcred, NULL);
 161                         break;
 162                 default:
 163                         zio->io_error = ENOTSUP;
 164                 }
 165
 166                 return (ZIO_PIPELINE_CONTINUE);
 167         }
 168
 169         zio->io_error = vn_rdwr(zio->io_type == ZIO_TYPE_READ ?
 170             UIO_READ : UIO_WRITE, vf->vf_vnode, zio->io_data,
 171             zio->io_size, zio->io_offset, UIO_SYSSPACE,
 172             0, RLIM64_INFINITY, kcred, &resid);
 173
 174         if (resid != 0 && zio->io_error == 0)
 175                 zio->io_error = ENOSPC;
 176
 177         zio_interrupt(zio);
 178
 179         return (ZIO_PIPELINE_STOP);
 180 }
 181
 182 /* ARGSUSED */
 183 static void
 184 vdev_file_io_done(zio_t *zio)
 185 {
 186 }
 187
 188 vdev_ops_t vdev_file_ops = {
 189         vdev_file_open,
 190         vdev_file_close,
 191         vdev_default_asize,
 192         vdev_file_io_start,
 193         vdev_file_io_done,
 194         NULL,
 195         vdev_file_hold,
 196         vdev_file_rele,
 197         VDEV_TYPE_FILE,         /* name of this vdev type */
 198         B_TRUE                  /* leaf vdev */
 199 };
 200
 201 /*
 202  * From userland we access disks just like files.
 203  */
 204 #ifndef _KERNEL
 205
 206 vdev_ops_t vdev_disk_ops = {
 207         vdev_file_open,
 208         vdev_file_close,
 209         vdev_default_asize,
 210         vdev_file_io_start,
 211         vdev_file_io_done,
 212         NULL,
 213         vdev_file_hold,
 214         vdev_file_rele,
 215         VDEV_TYPE_DISK,         /* name of this vdev type */
 216         B_TRUE                  /* leaf vdev */
 217 };
 218
 219 #endif