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>
36 * Virtual device vector for files.
40 vdev_file_hold(vdev_t *vd)
42 ASSERT(vd->vdev_path != NULL);
46 vdev_file_rele(vdev_t *vd)
48 ASSERT(vd->vdev_path != NULL);
52 vdev_file_open(vdev_t *vd, uint64_t *psize, uint64_t *max_psize,
53 uint64_t *logical_ashift, uint64_t *physical_ashift)
61 * We must have a pathname, and it must be absolute.
63 if (vd->vdev_path == NULL || vd->vdev_path[0] != '/') {
64 vd->vdev_stat.vs_aux = VDEV_AUX_BAD_LABEL;
65 return (SET_ERROR(EINVAL));
69 * Reopen the device if it's not currently open. Otherwise,
70 * just update the physical size of the device.
72 if (vd->vdev_tsd != NULL) {
73 ASSERT(vd->vdev_reopening);
79 vf = vd->vdev_tsd = kmem_zalloc(sizeof (vdev_file_t), KM_SLEEP);
82 * We always open the files from the root of the global zone, even if
83 * we're in a local zone. If the user has gotten to this point, the
84 * administrator has already decided that the pool should be available
85 * to local zone users, so the underlying devices should be as well.
87 ASSERT(vd->vdev_path != NULL && vd->vdev_path[0] == '/');
88 error = vn_openat(vd->vdev_path + 1, UIO_SYSSPACE,
89 spa_mode(vd->vdev_spa) | FOFFMAX, 0, &vp, 0, 0, rootdir, -1);
92 vd->vdev_stat.vs_aux = VDEV_AUX_OPEN_FAILED;
93 kmem_free(vd->vdev_tsd, sizeof (vdev_file_t));
102 * Make sure it's a regular file.
104 if (vp->v_type != VREG) {
106 (void) VOP_CLOSE(vp, spa_mode(vd->vdev_spa), 1, 0, kcred, NULL);
108 vd->vdev_stat.vs_aux = VDEV_AUX_OPEN_FAILED;
110 kmem_free(vd->vdev_tsd, sizeof (vdev_file_t));
113 return (SET_ERROR(ENODEV));
119 * Determine the physical size of the file.
121 vattr.va_mask = AT_SIZE;
122 vn_lock(vp, LK_SHARED | LK_RETRY);
123 error = VOP_GETATTR(vp, &vattr, kcred);
126 (void) VOP_CLOSE(vp, spa_mode(vd->vdev_spa), 1, 0, kcred, NULL);
127 vd->vdev_stat.vs_aux = VDEV_AUX_OPEN_FAILED;
128 kmem_free(vd->vdev_tsd, sizeof (vdev_file_t));
133 vd->vdev_notrim = B_TRUE;
135 *max_psize = *psize = vattr.va_size;
136 *logical_ashift = SPA_MINBLOCKSHIFT;
137 *physical_ashift = SPA_MINBLOCKSHIFT;
143 vdev_file_close(vdev_t *vd)
145 vdev_file_t *vf = vd->vdev_tsd;
147 if (vd->vdev_reopening || vf == NULL)
150 if (vf->vf_vnode != NULL) {
151 (void) VOP_CLOSE(vf->vf_vnode, spa_mode(vd->vdev_spa), 1, 0,
155 vd->vdev_delayed_close = B_FALSE;
156 kmem_free(vf, sizeof (vdev_file_t));
161 vdev_file_io_start(zio_t *zio)
163 vdev_t *vd = zio->io_vd;
169 if (!vdev_readable(vd)) {
170 zio->io_error = SET_ERROR(ENXIO);
178 if (zio->io_type == ZIO_TYPE_IOCTL) {
179 switch (zio->io_cmd) {
180 case DKIOCFLUSHWRITECACHE:
181 zio->io_error = VOP_FSYNC(vp, FSYNC | FDSYNC,
185 zio->io_error = SET_ERROR(ENOTSUP);
192 ASSERT(zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE);
193 zio->io_target_timestamp = zio_handle_io_delay(zio);
195 if (zio->io_type == ZIO_TYPE_READ) {
196 addr = abd_borrow_buf(zio->io_abd, zio->io_size);
198 addr = abd_borrow_buf_copy(zio->io_abd, zio->io_size);
201 zio->io_error = vn_rdwr(zio->io_type == ZIO_TYPE_READ ?
202 UIO_READ : UIO_WRITE, vp, addr, zio->io_size,
203 zio->io_offset, UIO_SYSSPACE, 0, RLIM64_INFINITY, kcred, &resid);
205 if (zio->io_type == ZIO_TYPE_READ) {
206 abd_return_buf_copy(zio->io_abd, addr, zio->io_size);
208 abd_return_buf(zio->io_abd, addr, zio->io_size);
211 if (resid != 0 && zio->io_error == 0)
212 zio->io_error = ENOSPC;
214 zio_delay_interrupt(zio);
217 VERIFY3U(taskq_dispatch(system_taskq, vdev_file_io_strategy, bp,
224 vdev_file_io_done(zio_t *zio)
228 vdev_ops_t vdev_file_ops = {
237 VDEV_TYPE_FILE, /* name of this vdev type */
238 B_TRUE /* leaf vdev */
242 * From userland we access disks just like files.
246 vdev_ops_t vdev_disk_ops = {
255 VDEV_TYPE_DISK, /* name of this vdev type */
256 B_TRUE /* leaf vdev */