xref: /netvirt/usr/src/uts/common/fs/zfs/spa.c

/*
 * CDDL HEADER START
 *
 * The contents of this file are subject to the terms of the
 * Common Development and Distribution License (the "License").
 * You may not use this file except in compliance with the License.
 *
 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
 * or http://www.opensolaris.org/os/licensing.
 * See the License for the specific language governing permissions
 * and limitations under the License.
 *
 * When distributing Covered Code, include this CDDL HEADER in each
 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
 * If applicable, add the following below this CDDL HEADER, with the
 * fields enclosed by brackets "[]" replaced with your own identifying
 * information: Portions Copyright [yyyy] [name of copyright owner]
 *
 * CDDL HEADER END
 */

/*
 * Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
 * Use is subject to license terms.
 */

#pragma ident	"@(#)spa.c	1.46	08/01/02 SMI"

/*
 * This file contains all the routines used when modifying on-disk SPA state.
 * This includes opening, importing, destroying, exporting a pool, and syncing a
 * pool.
 */

#include <sys/zfs_context.h>
#include <sys/fm/fs/zfs.h>
#include <sys/spa_impl.h>
#include <sys/zio.h>
#include <sys/zio_checksum.h>
#include <sys/zio_compress.h>
#include <sys/dmu.h>
#include <sys/dmu_tx.h>
#include <sys/zap.h>
#include <sys/zil.h>
#include <sys/vdev_impl.h>
#include <sys/metaslab.h>
#include <sys/uberblock_impl.h>
#include <sys/txg.h>
#include <sys/avl.h>
#include <sys/dmu_traverse.h>
#include <sys/dmu_objset.h>
#include <sys/unique.h>
#include <sys/dsl_pool.h>
#include <sys/dsl_dataset.h>
#include <sys/dsl_dir.h>
#include <sys/dsl_prop.h>
#include <sys/dsl_synctask.h>
#include <sys/fs/zfs.h>
#include <sys/arc.h>
#include <sys/callb.h>
#include <sys/systeminfo.h>
#include <sys/sunddi.h>

#include "zfs_prop.h"

int zio_taskq_threads = 8;

static void spa_sync_props(void *arg1, void *arg2, cred_t *cr, dmu_tx_t *tx);

/*
 * ==========================================================================
 * SPA properties routines
 * ==========================================================================
 */

/*
 * Add a (source=src, propname=propval) list to an nvlist.
 */
static int
spa_prop_add_list(nvlist_t *nvl, zpool_prop_t prop, char *strval,
    uint64_t intval, zprop_source_t src)
{
	const char *propname = zpool_prop_to_name(prop);
	nvlist_t *propval;
	int err = 0;

	if (err = nvlist_alloc(&propval, NV_UNIQUE_NAME, KM_SLEEP))
		return (err);

	if (err = nvlist_add_uint64(propval, ZPROP_SOURCE, src))
		goto out;

	if (strval != NULL) {
		if (err = nvlist_add_string(propval, ZPROP_VALUE, strval))
			goto out;
	} else {
		if (err = nvlist_add_uint64(propval, ZPROP_VALUE, intval))
			goto out;
	}

	err = nvlist_add_nvlist(nvl, propname, propval);
out:
	nvlist_free(propval);
	return (err);
}

/*
 * Get property values from the spa configuration.
 */
static int
spa_prop_get_config(spa_t *spa, nvlist_t **nvp)
{
	uint64_t size = spa_get_space(spa);
	uint64_t used = spa_get_alloc(spa);
	uint64_t cap, version;
	zprop_source_t src = ZPROP_SRC_NONE;
	int err;
	char *cachefile;
	size_t len;

	/*
	 * readonly properties
	 */
	if (err = spa_prop_add_list(*nvp, ZPOOL_PROP_NAME, spa->spa_name,
	    0, src))
		return (err);

	if (err = spa_prop_add_list(*nvp, ZPOOL_PROP_SIZE, NULL, size, src))
		return (err);

	if (err = spa_prop_add_list(*nvp, ZPOOL_PROP_USED, NULL, used, src))
		return (err);

	if (err = spa_prop_add_list(*nvp, ZPOOL_PROP_AVAILABLE, NULL,
	    size - used, src))
		return (err);

	cap = (size == 0) ? 0 : (used * 100 / size);
	if (err = spa_prop_add_list(*nvp, ZPOOL_PROP_CAPACITY, NULL, cap, src))
		return (err);

	if (err = spa_prop_add_list(*nvp, ZPOOL_PROP_GUID, NULL,
	    spa_guid(spa), src))
		return (err);

	if (err = spa_prop_add_list(*nvp, ZPOOL_PROP_HEALTH, NULL,
	    spa->spa_root_vdev->vdev_state, src))
		return (err);

	/*
	 * settable properties that are not stored in the pool property object.
	 */
	version = spa_version(spa);
	if (version == zpool_prop_default_numeric(ZPOOL_PROP_VERSION))
		src = ZPROP_SRC_DEFAULT;
	else
		src = ZPROP_SRC_LOCAL;
	if (err = spa_prop_add_list(*nvp, ZPOOL_PROP_VERSION, NULL,
	    version, src))
		return (err);

	if (spa->spa_root != NULL) {
		src = ZPROP_SRC_LOCAL;
		if (err = spa_prop_add_list(*nvp, ZPOOL_PROP_ALTROOT,
		    spa->spa_root, 0, src))
			return (err);
	}

	if (spa->spa_config_dir != NULL) {
		if (strcmp(spa->spa_config_dir, "none") == 0) {
			err = spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE,
			    spa->spa_config_dir, 0, ZPROP_SRC_LOCAL);
		} else {
			len = strlen(spa->spa_config_dir) +
			    strlen(spa->spa_config_file) + 2;
			cachefile = kmem_alloc(len, KM_SLEEP);
			(void) snprintf(cachefile, len, "%s/%s",
			    spa->spa_config_dir, spa->spa_config_file);
			err = spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE,
			    cachefile, 0, ZPROP_SRC_LOCAL);
			kmem_free(cachefile, len);
		}

		if (err)
			return (err);
	}

	return (0);
}

/*
 * Get zpool property values.
 */
int
spa_prop_get(spa_t *spa, nvlist_t **nvp)
{
	zap_cursor_t zc;
	zap_attribute_t za;
	objset_t *mos = spa->spa_meta_objset;
	int err;

	if (err = nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_SLEEP))
		return (err);

	/*
	 * Get properties from the spa config.
	 */
	if (err = spa_prop_get_config(spa, nvp))
		goto out;

	mutex_enter(&spa->spa_props_lock);
	/* If no pool property object, no more prop to get. */
	if (spa->spa_pool_props_object == 0) {
		mutex_exit(&spa->spa_props_lock);
		return (0);
	}

	/*
	 * Get properties from the MOS pool property object.
	 */
	for (zap_cursor_init(&zc, mos, spa->spa_pool_props_object);
	    (err = zap_cursor_retrieve(&zc, &za)) == 0;
	    zap_cursor_advance(&zc)) {
		uint64_t intval = 0;
		char *strval = NULL;
		zprop_source_t src = ZPROP_SRC_DEFAULT;
		zpool_prop_t prop;

		if ((prop = zpool_name_to_prop(za.za_name)) == ZPROP_INVAL)
			continue;

		switch (za.za_integer_length) {
		case 8:
			/* integer property */
			if (za.za_first_integer !=
			    zpool_prop_default_numeric(prop))
				src = ZPROP_SRC_LOCAL;

			if (prop == ZPOOL_PROP_BOOTFS) {
				dsl_pool_t *dp;
				dsl_dataset_t *ds = NULL;

				dp = spa_get_dsl(spa);
				rw_enter(&dp->dp_config_rwlock, RW_READER);
				if (err = dsl_dataset_open_obj(dp,
				    za.za_first_integer, NULL, DS_MODE_NONE,
				    FTAG, &ds)) {
					rw_exit(&dp->dp_config_rwlock);
					break;
				}

				strval = kmem_alloc(
				    MAXNAMELEN + strlen(MOS_DIR_NAME) + 1,
				    KM_SLEEP);
				dsl_dataset_name(ds, strval);
				dsl_dataset_close(ds, DS_MODE_NONE, FTAG);
				rw_exit(&dp->dp_config_rwlock);
			} else {
				strval = NULL;
				intval = za.za_first_integer;
			}

			err = spa_prop_add_list(*nvp, prop, strval,
			    intval, src);

			if (strval != NULL)
				kmem_free(strval,
				    MAXNAMELEN + strlen(MOS_DIR_NAME) + 1);

			break;

		case 1:
			/* string property */
			strval = kmem_alloc(za.za_num_integers, KM_SLEEP);
			err = zap_lookup(mos, spa->spa_pool_props_object,
			    za.za_name, 1, za.za_num_integers, strval);
			if (err) {
				kmem_free(strval, za.za_num_integers);
				break;
			}
			err = spa_prop_add_list(*nvp, prop, strval, 0, src);
			kmem_free(strval, za.za_num_integers);
			break;

		default:
			break;
		}
	}
	zap_cursor_fini(&zc);
	mutex_exit(&spa->spa_props_lock);
out:
	if (err && err != ENOENT) {
		nvlist_free(*nvp);
		return (err);
	}

	return (0);
}

/*
 * Validate the given pool properties nvlist and modify the list
 * for the property values to be set.
 */
static int
spa_prop_validate(spa_t *spa, nvlist_t *props)
{
	nvpair_t *elem;
	int error = 0, reset_bootfs = 0;
	uint64_t objnum;

	elem = NULL;
	while ((elem = nvlist_next_nvpair(props, elem)) != NULL) {
		zpool_prop_t prop;
		char *propname, *strval;
		uint64_t intval;
		vdev_t *rvdev;
		char *vdev_type;
		objset_t *os;
		char *slash;

		propname = nvpair_name(elem);

		if ((prop = zpool_name_to_prop(propname)) == ZPROP_INVAL)
			return (EINVAL);

		switch (prop) {
		case ZPOOL_PROP_VERSION:
			error = nvpair_value_uint64(elem, &intval);
			if (!error &&
			    (intval < spa_version(spa) || intval > SPA_VERSION))
				error = EINVAL;
			break;

		case ZPOOL_PROP_DELEGATION:
		case ZPOOL_PROP_AUTOREPLACE:
			error = nvpair_value_uint64(elem, &intval);
			if (!error && intval > 1)
				error = EINVAL;
			break;

		case ZPOOL_PROP_BOOTFS:
			if (spa_version(spa) < SPA_VERSION_BOOTFS) {
				error = ENOTSUP;
				break;
			}

			/*
			 * A bootable filesystem can not be on a RAIDZ pool
			 * nor a striped pool with more than 1 device.
			 */
			rvdev = spa->spa_root_vdev;
			vdev_type =
			    rvdev->vdev_child[0]->vdev_ops->vdev_op_type;
			if (rvdev->vdev_children > 1 ||
			    strcmp(vdev_type, VDEV_TYPE_RAIDZ) == 0 ||
			    strcmp(vdev_type, VDEV_TYPE_MISSING) == 0) {
				error = ENOTSUP;
				break;
			}

			reset_bootfs = 1;

			error = nvpair_value_string(elem, &strval);

			if (!error) {
				if (strval == NULL || strval[0] == '\0') {
					objnum = zpool_prop_default_numeric(
					    ZPOOL_PROP_BOOTFS);
					break;
				}

				if (error = dmu_objset_open(strval, DMU_OST_ZFS,
				    DS_MODE_STANDARD | DS_MODE_READONLY, &os))
					break;
				objnum = dmu_objset_id(os);
				dmu_objset_close(os);
			}
			break;
		case ZPOOL_PROP_FAILUREMODE:
			error = nvpair_value_uint64(elem, &intval);
			if (!error && (intval < ZIO_FAILURE_MODE_WAIT ||
			    intval > ZIO_FAILURE_MODE_PANIC))
				error = EINVAL;

			/*
			 * This is a special case which only occurs when
			 * the pool has completely failed. This allows
			 * the user to change the in-core failmode property
			 * without syncing it out to disk (I/Os might
			 * currently be blocked). We do this by returning
			 * EIO to the caller (spa_prop_set) to trick it
			 * into thinking we encountered a property validation
			 * error.
			 */
			if (!error && spa_state(spa) == POOL_STATE_IO_FAILURE) {
				spa->spa_failmode = intval;
				error = EIO;
			}
			break;

		case ZPOOL_PROP_CACHEFILE:
			if ((error = nvpair_value_string(elem, &strval)) != 0)
				break;

			if (strval[0] == '\0')
				break;

			if (strcmp(strval, "none") == 0)
				break;

			if (strval[0] != '/') {
				error = EINVAL;
				break;
			}

			slash = strrchr(strval, '/');
			ASSERT(slash != NULL);

			if (slash[1] == '\0' || strcmp(slash, "/.") == 0 ||
			    strcmp(slash, "/..") == 0)
				error = EINVAL;
			break;
		}

		if (error)
			break;
	}

	if (!error && reset_bootfs) {
		error = nvlist_remove(props,
		    zpool_prop_to_name(ZPOOL_PROP_BOOTFS), DATA_TYPE_STRING);

		if (!error) {
			error = nvlist_add_uint64(props,
			    zpool_prop_to_name(ZPOOL_PROP_BOOTFS), objnum);
		}
	}

	return (error);
}

int
spa_prop_set(spa_t *spa, nvlist_t *nvp)
{
	int error;

	if ((error = spa_prop_validate(spa, nvp)) != 0)
		return (error);

	return (dsl_sync_task_do(spa_get_dsl(spa), NULL, spa_sync_props,
	    spa, nvp, 3));
}

/*
 * If the bootfs property value is dsobj, clear it.
 */
void
spa_prop_clear_bootfs(spa_t *spa, uint64_t dsobj, dmu_tx_t *tx)
{
	if (spa->spa_bootfs == dsobj && spa->spa_pool_props_object != 0) {
		VERIFY(zap_remove(spa->spa_meta_objset,
		    spa->spa_pool_props_object,
		    zpool_prop_to_name(ZPOOL_PROP_BOOTFS), tx) == 0);
		spa->spa_bootfs = 0;
	}
}

/*
 * ==========================================================================
 * SPA state manipulation (open/create/destroy/import/export)
 * ==========================================================================
 */

static int
spa_error_entry_compare(const void *a, const void *b)
{
	spa_error_entry_t *sa = (spa_error_entry_t *)a;
	spa_error_entry_t *sb = (spa_error_entry_t *)b;
	int ret;

	ret = bcmp(&sa->se_bookmark, &sb->se_bookmark,
	    sizeof (zbookmark_t));

	if (ret < 0)
		return (-1);
	else if (ret > 0)
		return (1);
	else
		return (0);
}

/*
 * Utility function which retrieves copies of the current logs and
 * re-initializes them in the process.
 */
void
spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub)
{
	ASSERT(MUTEX_HELD(&spa->spa_errlist_lock));

	bcopy(&spa->spa_errlist_last, last, sizeof (avl_tree_t));
	bcopy(&spa->spa_errlist_scrub, scrub, sizeof (avl_tree_t));

	avl_create(&spa->spa_errlist_scrub,
	    spa_error_entry_compare, sizeof (spa_error_entry_t),
	    offsetof(spa_error_entry_t, se_avl));
	avl_create(&spa->spa_errlist_last,
	    spa_error_entry_compare, sizeof (spa_error_entry_t),
	    offsetof(spa_error_entry_t, se_avl));
}

/*
 * Activate an uninitialized pool.
 */
static void
spa_activate(spa_t *spa)
{
	int t;

	ASSERT(spa->spa_state == POOL_STATE_UNINITIALIZED);

	spa->spa_state = POOL_STATE_ACTIVE;

	spa->spa_normal_class = metaslab_class_create();
	spa->spa_log_class = metaslab_class_create();

	for (t = 0; t < ZIO_TYPES; t++) {
		spa->spa_zio_issue_taskq[t] = taskq_create("spa_zio_issue",
		    zio_taskq_threads, maxclsyspri, 50, INT_MAX,
		    TASKQ_PREPOPULATE);
		spa->spa_zio_intr_taskq[t] = taskq_create("spa_zio_intr",
		    zio_taskq_threads, maxclsyspri, 50, INT_MAX,
		    TASKQ_PREPOPULATE);
	}

	list_create(&spa->spa_dirty_list, sizeof (vdev_t),
	    offsetof(vdev_t, vdev_dirty_node));
	list_create(&spa->spa_zio_list, sizeof (zio_t),
	    offsetof(zio_t, zio_link_node));

	txg_list_create(&spa->spa_vdev_txg_list,
	    offsetof(struct vdev, vdev_txg_node));

	avl_create(&spa->spa_errlist_scrub,
	    spa_error_entry_compare, sizeof (spa_error_entry_t),
	    offsetof(spa_error_entry_t, se_avl));
	avl_create(&spa->spa_errlist_last,
	    spa_error_entry_compare, sizeof (spa_error_entry_t),
	    offsetof(spa_error_entry_t, se_avl));
}

/*
 * Opposite of spa_activate().
 */
static void
spa_deactivate(spa_t *spa)
{
	int t;

	ASSERT(spa->spa_sync_on == B_FALSE);
	ASSERT(spa->spa_dsl_pool == NULL);
	ASSERT(spa->spa_root_vdev == NULL);

	ASSERT(spa->spa_state != POOL_STATE_UNINITIALIZED);

	txg_list_destroy(&spa->spa_vdev_txg_list);

	list_destroy(&spa->spa_dirty_list);
	list_destroy(&spa->spa_zio_list);

	for (t = 0; t < ZIO_TYPES; t++) {
		taskq_destroy(spa->spa_zio_issue_taskq[t]);
		taskq_destroy(spa->spa_zio_intr_taskq[t]);
		spa->spa_zio_issue_taskq[t] = NULL;
		spa->spa_zio_intr_taskq[t] = NULL;
	}

	metaslab_class_destroy(spa->spa_normal_class);
	spa->spa_normal_class = NULL;

	metaslab_class_destroy(spa->spa_log_class);
	spa->spa_log_class = NULL;

	/*
	 * If this was part of an import or the open otherwise failed, we may
	 * still have errors left in the queues.  Empty them just in case.
	 */
	spa_errlog_drain(spa);

	avl_destroy(&spa->spa_errlist_scrub);
	avl_destroy(&spa->spa_errlist_last);

	spa->spa_state = POOL_STATE_UNINITIALIZED;
}

/*
 * Verify a pool configuration, and construct the vdev tree appropriately.  This
 * will create all the necessary vdevs in the appropriate layout, with each vdev
 * in the CLOSED state.  This will prep the pool before open/creation/import.
 * All vdev validation is done by the vdev_alloc() routine.
 */
static int
spa_config_parse(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent,
    uint_t id, int atype)
{
	nvlist_t **child;
	uint_t c, children;
	int error;

	if ((error = vdev_alloc(spa, vdp, nv, parent, id, atype)) != 0)
		return (error);

	if ((*vdp)->vdev_ops->vdev_op_leaf)
		return (0);

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	    &child, &children) != 0) {
		vdev_free(*vdp);
		*vdp = NULL;
		return (EINVAL);
	}

	for (c = 0; c < children; c++) {
		vdev_t *vd;
		if ((error = spa_config_parse(spa, &vd, child[c], *vdp, c,
		    atype)) != 0) {
			vdev_free(*vdp);
			*vdp = NULL;
			return (error);
		}
	}

	ASSERT(*vdp != NULL);

	return (0);
}

/*
 * Opposite of spa_load().
 */
static void
spa_unload(spa_t *spa)
{
	int i;

	/*
	 * Stop async tasks.
	 */
	spa_async_suspend(spa);

	/*
	 * Stop syncing.
	 */
	if (spa->spa_sync_on) {
		txg_sync_stop(spa->spa_dsl_pool);
		spa->spa_sync_on = B_FALSE;
	}

	/*
	 * Wait for any outstanding prefetch I/O to complete.
	 */
	spa_config_enter(spa, RW_WRITER, FTAG);
	spa_config_exit(spa, FTAG);

	/*
	 * Drop and purge level 2 cache
	 */
	spa_l2cache_drop(spa);

	/*
	 * Close the dsl pool.
	 */
	if (spa->spa_dsl_pool) {
		dsl_pool_close(spa->spa_dsl_pool);
		spa->spa_dsl_pool = NULL;
	}

	/*
	 * Close all vdevs.
	 */
	if (spa->spa_root_vdev)
		vdev_free(spa->spa_root_vdev);
	ASSERT(spa->spa_root_vdev == NULL);

	for (i = 0; i < spa->spa_spares.sav_count; i++)
		vdev_free(spa->spa_spares.sav_vdevs[i]);
	if (spa->spa_spares.sav_vdevs) {
		kmem_free(spa->spa_spares.sav_vdevs,
		    spa->spa_spares.sav_count * sizeof (void *));
		spa->spa_spares.sav_vdevs = NULL;
	}
	if (spa->spa_spares.sav_config) {
		nvlist_free(spa->spa_spares.sav_config);
		spa->spa_spares.sav_config = NULL;
	}

	for (i = 0; i < spa->spa_l2cache.sav_count; i++)
		vdev_free(spa->spa_l2cache.sav_vdevs[i]);
	if (spa->spa_l2cache.sav_vdevs) {
		kmem_free(spa->spa_l2cache.sav_vdevs,
		    spa->spa_l2cache.sav_count * sizeof (void *));
		spa->spa_l2cache.sav_vdevs = NULL;
	}
	if (spa->spa_l2cache.sav_config) {
		nvlist_free(spa->spa_l2cache.sav_config);
		spa->spa_l2cache.sav_config = NULL;
	}

	spa->spa_async_suspended = 0;
}

/*
 * Load (or re-load) the current list of vdevs describing the active spares for
 * this pool.  When this is called, we have some form of basic information in
 * 'spa_spares.sav_config'.  We parse this into vdevs, try to open them, and
 * then re-generate a more complete list including status information.
 */
static void
spa_load_spares(spa_t *spa)
{
	nvlist_t **spares;
	uint_t nspares;
	int i;
	vdev_t *vd, *tvd;

	/*
	 * First, close and free any existing spare vdevs.
	 */
	for (i = 0; i < spa->spa_spares.sav_count; i++) {
		vd = spa->spa_spares.sav_vdevs[i];

		/* Undo the call to spa_activate() below */
		if ((tvd = spa_lookup_by_guid(spa, vd->vdev_guid)) != NULL &&
		    tvd->vdev_isspare)
			spa_spare_remove(tvd);
		vdev_close(vd);
		vdev_free(vd);
	}

	if (spa->spa_spares.sav_vdevs)
		kmem_free(spa->spa_spares.sav_vdevs,
		    spa->spa_spares.sav_count * sizeof (void *));

	if (spa->spa_spares.sav_config == NULL)
		nspares = 0;
	else
		VERIFY(nvlist_lookup_nvlist_array(spa->spa_spares.sav_config,
		    ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0);

	spa->spa_spares.sav_count = (int)nspares;
	spa->spa_spares.sav_vdevs = NULL;

	if (nspares == 0)
		return;

	/*
	 * Construct the array of vdevs, opening them to get status in the
	 * process.   For each spare, there is potentially two different vdev_t
	 * structures associated with it: one in the list of spares (used only
	 * for basic validation purposes) and one in the active vdev
	 * configuration (if it's spared in).  During this phase we open and
	 * validate each vdev on the spare list.  If the vdev also exists in the
	 * active configuration, then we also mark this vdev as an active spare.
	 */
	spa->spa_spares.sav_vdevs = kmem_alloc(nspares * sizeof (void *),
	    KM_SLEEP);
	for (i = 0; i < spa->spa_spares.sav_count; i++) {
		VERIFY(spa_config_parse(spa, &vd, spares[i], NULL, 0,
		    VDEV_ALLOC_SPARE) == 0);
		ASSERT(vd != NULL);

		spa->spa_spares.sav_vdevs[i] = vd;

		if ((tvd = spa_lookup_by_guid(spa, vd->vdev_guid)) != NULL) {
			if (!tvd->vdev_isspare)
				spa_spare_add(tvd);

			/*
			 * We only mark the spare active if we were successfully
			 * able to load the vdev.  Otherwise, importing a pool
			 * with a bad active spare would result in strange
			 * behavior, because multiple pool would think the spare
			 * is actively in use.
			 *
			 * There is a vulnerability here to an equally bizarre
			 * circumstance, where a dead active spare is later
			 * brought back to life (onlined or otherwise).  Given
			 * the rarity of this scenario, and the extra complexity
			 * it adds, we ignore the possibility.
			 */
			if (!vdev_is_dead(tvd))
				spa_spare_activate(tvd);
		}

		if (vdev_open(vd) != 0)
			continue;

		vd->vdev_top = vd;
		if (vdev_validate_aux(vd) == 0)
			spa_spare_add(vd);
	}

	/*
	 * Recompute the stashed list of spares, with status information
	 * this time.
	 */
	VERIFY(nvlist_remove(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES,
	    DATA_TYPE_NVLIST_ARRAY) == 0);

	spares = kmem_alloc(spa->spa_spares.sav_count * sizeof (void *),
	    KM_SLEEP);
	for (i = 0; i < spa->spa_spares.sav_count; i++)
		spares[i] = vdev_config_generate(spa,
		    spa->spa_spares.sav_vdevs[i], B_TRUE, B_TRUE, B_FALSE);
	VERIFY(nvlist_add_nvlist_array(spa->spa_spares.sav_config,
	    ZPOOL_CONFIG_SPARES, spares, spa->spa_spares.sav_count) == 0);
	for (i = 0; i < spa->spa_spares.sav_count; i++)
		nvlist_free(spares[i]);
	kmem_free(spares, spa->spa_spares.sav_count * sizeof (void *));
}

/*
 * Load (or re-load) the current list of vdevs describing the active l2cache for
 * this pool.  When this is called, we have some form of basic information in
 * 'spa_l2cache.sav_config'.  We parse this into vdevs, try to open them, and
 * then re-generate a more complete list including status information.
 * Devices which are already active have their details maintained, and are
 * not re-opened.
 */
static void
spa_load_l2cache(spa_t *spa)
{
	nvlist_t **l2cache;
	uint_t nl2cache;
	int i, j, oldnvdevs;
	uint64_t guid;
	vdev_t *vd, **oldvdevs, **newvdevs;
	spa_aux_vdev_t *sav = &spa->spa_l2cache;

	if (sav->sav_config != NULL) {
		VERIFY(nvlist_lookup_nvlist_array(sav->sav_config,
		    ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0);
		newvdevs = kmem_alloc(nl2cache * sizeof (void *), KM_SLEEP);
	} else {
		nl2cache = 0;
	}

	oldvdevs = sav->sav_vdevs;
	oldnvdevs = sav->sav_count;
	sav->sav_vdevs = NULL;
	sav->sav_count = 0;

	/*
	 * Process new nvlist of vdevs.
	 */
	for (i = 0; i < nl2cache; i++) {
		VERIFY(nvlist_lookup_uint64(l2cache[i], ZPOOL_CONFIG_GUID,
		    &guid) == 0);

		newvdevs[i] = NULL;
		for (j = 0; j < oldnvdevs; j++) {
			vd = oldvdevs[j];
			if (vd != NULL && guid == vd->vdev_guid) {
				/*
				 * Retain previous vdev for add/remove ops.
				 */
				newvdevs[i] = vd;
				oldvdevs[j] = NULL;
				break;
			}
		}

		if (newvdevs[i] == NULL) {
			/*
			 * Create new vdev
			 */
			VERIFY(spa_config_parse(spa, &vd, l2cache[i], NULL, 0,
			    VDEV_ALLOC_L2CACHE) == 0);
			ASSERT(vd != NULL);
			newvdevs[i] = vd;

			/*
			 * Commit this vdev as an l2cache device,
			 * even if it fails to open.
			 */
			spa_l2cache_add(vd);

			if (vdev_open(vd) != 0)
				continue;

			vd->vdev_top = vd;
			(void) vdev_validate_aux(vd);

			if (!vdev_is_dead(vd)) {
				uint64_t size;
				size = vdev_get_rsize(vd);
				ASSERT3U(size, >, 0);
				if (spa_mode & FWRITE) {
					l2arc_add_vdev(spa, vd,
					    VDEV_LABEL_START_SIZE,
					    size - VDEV_LABEL_START_SIZE);
				}
				spa_l2cache_activate(vd);
			}
		}
	}

	/*
	 * Purge vdevs that were dropped
	 */
	for (i = 0; i < oldnvdevs; i++) {
		uint64_t pool;

		vd = oldvdevs[i];
		if (vd != NULL) {
			if (spa_mode & FWRITE &&
			    spa_l2cache_exists(vd->vdev_guid, &pool) &&
			    pool != 0ULL) {
				l2arc_remove_vdev(vd);
			}
			(void) vdev_close(vd);
			spa_l2cache_remove(vd);
		}
	}

	if (oldvdevs)
		kmem_free(oldvdevs, oldnvdevs * sizeof (void *));

	if (sav->sav_config == NULL)
		goto out;

	sav->sav_vdevs = newvdevs;
	sav->sav_count = (int)nl2cache;

	/*
	 * Recompute the stashed list of l2cache devices, with status
	 * information this time.
	 */
	VERIFY(nvlist_remove(sav->sav_config, ZPOOL_CONFIG_L2CACHE,
	    DATA_TYPE_NVLIST_ARRAY) == 0);

	l2cache = kmem_alloc(sav->sav_count * sizeof (void *), KM_SLEEP);
	for (i = 0; i < sav->sav_count; i++)
		l2cache[i] = vdev_config_generate(spa,
		    sav->sav_vdevs[i], B_TRUE, B_FALSE, B_TRUE);
	VERIFY(nvlist_add_nvlist_array(sav->sav_config,
	    ZPOOL_CONFIG_L2CACHE, l2cache, sav->sav_count) == 0);
out:
	for (i = 0; i < sav->sav_count; i++)
		nvlist_free(l2cache[i]);
	if (sav->sav_count)
		kmem_free(l2cache, sav->sav_count * sizeof (void *));
}

static int
load_nvlist(spa_t *spa, uint64_t obj, nvlist_t **value)
{
	dmu_buf_t *db;
	char *packed = NULL;
	size_t nvsize = 0;
	int error;
	*value = NULL;

	VERIFY(0 == dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db));
	nvsize = *(uint64_t *)db->db_data;
	dmu_buf_rele(db, FTAG);

	packed = kmem_alloc(nvsize, KM_SLEEP);
	error = dmu_read(spa->spa_meta_objset, obj, 0, nvsize, packed);
	if (error == 0)
		error = nvlist_unpack(packed, nvsize, value, 0);
	kmem_free(packed, nvsize);

	return (error);
}

/*
 * Checks to see if the given vdev could not be opened, in which case we post a
 * sysevent to notify the autoreplace code that the device has been removed.
 */
static void
spa_check_removed(vdev_t *vd)
{
	int c;

	for (c = 0; c < vd->vdev_children; c++)
		spa_check_removed(vd->vdev_child[c]);

	if (vd->vdev_ops->vdev_op_leaf && vdev_is_dead(vd)) {
		zfs_post_autoreplace(vd->vdev_spa, vd);
		spa_event_notify(vd->vdev_spa, vd, ESC_ZFS_VDEV_CHECK);
	}
}

/*
 * Load an existing storage pool, using the pool's builtin spa_config as a
 * source of configuration information.
 */
static int
spa_load(spa_t *spa, nvlist_t *config, spa_load_state_t state, int mosconfig)
{
	int error = 0;
	nvlist_t *nvroot = NULL;
	vdev_t *rvd;
	uberblock_t *ub = &spa->spa_uberblock;
	uint64_t config_cache_txg = spa->spa_config_txg;
	uint64_t pool_guid;
	uint64_t version;
	zio_t *zio;
	uint64_t autoreplace = 0;

	spa->spa_load_state = state;

	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) ||
	    nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &pool_guid)) {
		error = EINVAL;
		goto out;
	}

	/*
	 * Versioning wasn't explicitly added to the label until later, so if
	 * it's not present treat it as the initial version.
	 */
	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION, &version) != 0)
		version = SPA_VERSION_INITIAL;

	(void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG,
	    &spa->spa_config_txg);

	if ((state == SPA_LOAD_IMPORT || state == SPA_LOAD_TRYIMPORT) &&
	    spa_guid_exists(pool_guid, 0)) {
		error = EEXIST;
		goto out;
	}

	spa->spa_load_guid = pool_guid;

	/*
	 * Parse the configuration into a vdev tree.  We explicitly set the
	 * value that will be returned by spa_version() since parsing the
	 * configuration requires knowing the version number.
	 */
	spa_config_enter(spa, RW_WRITER, FTAG);
	spa->spa_ubsync.ub_version = version;
	error = spa_config_parse(spa, &rvd, nvroot, NULL, 0, VDEV_ALLOC_LOAD);
	spa_config_exit(spa, FTAG);

	if (error != 0)
		goto out;

	ASSERT(spa->spa_root_vdev == rvd);
	ASSERT(spa_guid(spa) == pool_guid);

	/*
	 * Try to open all vdevs, loading each label in the process.
	 */
	error = vdev_open(rvd);
	if (error != 0)
		goto