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.
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.
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]
22 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
23 * Copyright (c) 2011, 2015 by Delphix. All rights reserved.
26 #include <sys/zfs_context.h>
28 #include <sys/vdev_file.h>
29 #include <sys/vdev_impl.h>
31 #include <sys/fs/zfs.h>
32 #include <sys/fm/fs/zfs.h>
35 * Virtual device vector for files.
39 vdev_file_hold(vdev_t *vd)
41 ASSERT(vd->vdev_path != NULL);
45 vdev_file_rele(vdev_t *vd)
47 ASSERT(vd->vdev_path != NULL);
51 vdev_file_open(vdev_t *vd, uint64_t *psize, uint64_t *max_psize,
52 uint64_t *logical_ashift, uint64_t *physical_ashift)
60 * We must have a pathname, and it must be absolute.
62 if (vd->vdev_path == NULL || vd->vdev_path[0] != '/') {
63 vd->vdev_stat.vs_aux = VDEV_AUX_BAD_LABEL;
64 return (SET_ERROR(EINVAL));
68 * Reopen the device if it's not currently open. Otherwise,
69 * just update the physical size of the device.
71 if (vd->vdev_tsd != NULL) {
72 ASSERT(vd->vdev_reopening);
78 vf = vd->vdev_tsd = kmem_zalloc(sizeof (vdev_file_t), KM_SLEEP);
81 * We always open the files from the root of the global zone, even if
82 * we're in a local zone. If the user has gotten to this point, the
83 * administrator has already decided that the pool should be available
84 * to local zone users, so the underlying devices should be as well.
86 ASSERT(vd->vdev_path != NULL && vd->vdev_path[0] == '/');
87 error = vn_openat(vd->vdev_path + 1, UIO_SYSSPACE,
88 spa_mode(vd->vdev_spa) | FOFFMAX, 0, &vp, 0, 0, rootdir, -1);
91 vd->vdev_stat.vs_aux = VDEV_AUX_OPEN_FAILED;
92 kmem_free(vd->vdev_tsd, sizeof (vdev_file_t));
101 * Make sure it's a regular file.
103 if (vp->v_type != VREG) {
105 (void) VOP_CLOSE(vp, spa_mode(vd->vdev_spa), 1, 0, kcred, NULL);
107 vd->vdev_stat.vs_aux = VDEV_AUX_OPEN_FAILED;
109 kmem_free(vd->vdev_tsd, sizeof (vdev_file_t));
112 return (SET_ERROR(ENODEV));
118 * Determine the physical size of the file.
120 vattr.va_mask = AT_SIZE;
121 vn_lock(vp, LK_SHARED | LK_RETRY);
122 error = VOP_GETATTR(vp, &vattr, kcred);
125 (void) VOP_CLOSE(vp, spa_mode(vd->vdev_spa), 1, 0, kcred, NULL);
126 vd->vdev_stat.vs_aux = VDEV_AUX_OPEN_FAILED;
127 kmem_free(vd->vdev_tsd, sizeof (vdev_file_t));
132 vd->vdev_notrim = B_TRUE;
134 *max_psize = *psize = vattr.va_size;
135 *logical_ashift = SPA_MINBLOCKSHIFT;
136 *physical_ashift = SPA_MINBLOCKSHIFT;
142 vdev_file_close(vdev_t *vd)
144 vdev_file_t *vf = vd->vdev_tsd;
146 if (vd->vdev_reopening || vf == NULL)
149 if (vf->vf_vnode != NULL) {
150 (void) VOP_CLOSE(vf->vf_vnode, spa_mode(vd->vdev_spa), 1, 0,
154 vd->vdev_delayed_close = B_FALSE;
155 kmem_free(vf, sizeof (vdev_file_t));
160 vdev_file_io_start(zio_t *zio)
162 vdev_t *vd = zio->io_vd;
167 if (!vdev_readable(vd)) {
168 zio->io_error = SET_ERROR(ENXIO);
176 if (zio->io_type == ZIO_TYPE_IOCTL) {
177 switch (zio->io_cmd) {
178 case DKIOCFLUSHWRITECACHE:
179 zio->io_error = VOP_FSYNC(vp, FSYNC | FDSYNC,
183 zio->io_error = SET_ERROR(ENOTSUP);
190 ASSERT(zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE);
191 zio->io_target_timestamp = zio_handle_io_delay(zio);
193 zio->io_error = vn_rdwr(zio->io_type == ZIO_TYPE_READ ?
194 UIO_READ : UIO_WRITE, vp, zio->io_data, zio->io_size,
195 zio->io_offset, UIO_SYSSPACE, 0, RLIM64_INFINITY, kcred, &resid);
197 if (resid != 0 && zio->io_error == 0)
198 zio->io_error = ENOSPC;
200 zio_delay_interrupt(zio);
203 VERIFY3U(taskq_dispatch(system_taskq, vdev_file_io_strategy, bp,
210 vdev_file_io_done(zio_t *zio)
214 vdev_ops_t vdev_file_ops = {
223 VDEV_TYPE_FILE, /* name of this vdev type */
224 B_TRUE /* leaf vdev */
228 * From userland we access disks just like files.
232 vdev_ops_t vdev_disk_ops = {
241 VDEV_TYPE_DISK, /* name of this vdev type */
242 B_TRUE /* leaf vdev */