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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
 * Copyright (c) 2011, 2019 by Delphix. All rights reserved.
 * Copyright (c) 2015, Nexenta Systems, Inc.  All rights reserved.
 * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
 * Copyright 2013 Saso Kiselkov. All rights reserved.
 * Copyright (c) 2014 Integros [integros.com]
 * Copyright 2016 Toomas Soome <tsoome@me.com>
 * Copyright (c) 2017, 2019, Datto Inc. All rights reserved.
 * Copyright 2019 Joyent, Inc.
 * Copyright (c) 2017, Intel Corporation.
 * Copyright 2020 Joshua M. Clulow <josh@sysmgr.org>
 * Copyright 2021 OmniOS Community Edition (OmniOSce) Association.
 */

/*
 * SPA: Storage Pool Allocator
 *
 * 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/dmu.h>
#include <sys/dmu_tx.h>
#include <sys/zap.h>
#include <sys/zil.h>
#include <sys/ddt.h>
#include <sys/vdev_impl.h>
#include <sys/vdev_removal.h>
#include <sys/vdev_indirect_mapping.h>
#include <sys/vdev_indirect_births.h>
#include <sys/vdev_initialize.h>
#include <sys/vdev_trim.h>
#include <sys/metaslab.h>
#include <sys/metaslab_impl.h>
#include <sys/mmp.h>
#include <sys/uberblock_impl.h>
#include <sys/txg.h>
#include <sys/avl.h>
#include <sys/bpobj.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/spa_boot.h>
#include <sys/zfs_ioctl.h>
#include <sys/dsl_scan.h>
#include <sys/zfeature.h>
#include <sys/dsl_destroy.h>
#include <sys/abd.h>

#ifdef	_KERNEL
#include <sys/bootprops.h>
#include <sys/callb.h>
#include <sys/cpupart.h>
#include <sys/pool.h>
#include <sys/sysdc.h>
#include <sys/zone.h>
#endif	/* _KERNEL */

#include "zfs_prop.h"
#include "zfs_comutil.h"

/*
 * The interval, in seconds, at which failed configuration cache file writes
 * should be retried.
 */
int zfs_ccw_retry_interval = 300;

typedef enum zti_modes {
	ZTI_MODE_FIXED,			/* value is # of threads (min 1) */
	ZTI_MODE_BATCH,			/* cpu-intensive; value is ignored */
	ZTI_MODE_NULL,			/* don't create a taskq */
	ZTI_NMODES
} zti_modes_t;

#define	ZTI_P(n, q)	{ ZTI_MODE_FIXED, (n), (q) }
#define	ZTI_BATCH	{ ZTI_MODE_BATCH, 0, 1 }
#define	ZTI_NULL	{ ZTI_MODE_NULL, 0, 0 }

#define	ZTI_N(n)	ZTI_P(n, 1)
#define	ZTI_ONE		ZTI_N(1)

typedef struct zio_taskq_info {
	zti_modes_t zti_mode;
	uint_t zti_value;
	uint_t zti_count;
} zio_taskq_info_t;

static const char *const zio_taskq_types[ZIO_TASKQ_TYPES] = {
	"issue", "issue_high", "intr", "intr_high"
};

/*
 * This table defines the taskq settings for each ZFS I/O type. When
 * initializing a pool, we use this table to create an appropriately sized
 * taskq. Some operations are low volume and therefore have a small, static
 * number of threads assigned to their taskqs using the ZTI_N(#) or ZTI_ONE
 * macros. Other operations process a large amount of data; the ZTI_BATCH
 * macro causes us to create a taskq oriented for throughput. Some operations
 * are so high frequency and short-lived that the taskq itself can become a
 * point of lock contention. The ZTI_P(#, #) macro indicates that we need an
 * additional degree of parallelism specified by the number of threads per-
 * taskq and the number of taskqs; when dispatching an event in this case, the
 * particular taskq is chosen at random.
 *
 * The different taskq priorities are to handle the different contexts (issue
 * and interrupt) and then to reserve threads for ZIO_PRIORITY_NOW I/Os that
 * need to be handled with minimum delay.
 */
const zio_taskq_info_t zio_taskqs[ZIO_TYPES][ZIO_TASKQ_TYPES] = {
	/* ISSUE	ISSUE_HIGH	INTR		INTR_HIGH */
	{ ZTI_ONE,	ZTI_NULL,	ZTI_ONE,	ZTI_NULL }, /* NULL */
	{ ZTI_N(8),	ZTI_NULL,	ZTI_P(12, 8),	ZTI_NULL }, /* READ */
	{ ZTI_BATCH,	ZTI_N(5),	ZTI_N(8),	ZTI_N(5) }, /* WRITE */
	{ ZTI_P(12, 8),	ZTI_NULL,	ZTI_ONE,	ZTI_NULL }, /* FREE */
	{ ZTI_ONE,	ZTI_NULL,	ZTI_ONE,	ZTI_NULL }, /* CLAIM */
	{ ZTI_ONE,	ZTI_NULL,	ZTI_ONE,	ZTI_NULL }, /* IOCTL */
	{ ZTI_N(4),	ZTI_NULL,	ZTI_ONE,	ZTI_NULL }, /* TRIM */
};

static void spa_sync_version(void *arg, dmu_tx_t *tx);
static void spa_sync_props(void *arg, dmu_tx_t *tx);
static boolean_t spa_has_active_shared_spare(spa_t *spa);
static int spa_load_impl(spa_t *spa, spa_import_type_t type, char **ereport);
static void spa_vdev_resilver_done(spa_t *spa);

uint_t		zio_taskq_batch_pct = 75;	/* 1 thread per cpu in pset */
id_t		zio_taskq_psrset_bind = PS_NONE;
boolean_t	zio_taskq_sysdc = B_TRUE;	/* use SDC scheduling class */
uint_t		zio_taskq_basedc = 80;		/* base duty cycle */

boolean_t	spa_create_process = B_TRUE;	/* no process ==> no sysdc */
extern int	zfs_sync_pass_deferred_free;

/*
 * Report any spa_load_verify errors found, but do not fail spa_load.
 * This is used by zdb to analyze non-idle pools.
 */
boolean_t	spa_load_verify_dryrun = B_FALSE;

/*
 * This (illegal) pool name is used when temporarily importing a spa_t in order
 * to get the vdev stats associated with the imported devices.
 */
#define	TRYIMPORT_NAME	"$import"

/*
 * For debugging purposes: print out vdev tree during pool import.
 */
boolean_t	spa_load_print_vdev_tree = B_FALSE;

/*
 * A non-zero value for zfs_max_missing_tvds means that we allow importing
 * pools with missing top-level vdevs. This is strictly intended for advanced
 * pool recovery cases since missing data is almost inevitable. Pools with
 * missing devices can only be imported read-only for safety reasons, and their
 * fail-mode will be automatically set to "continue".
 *
 * With 1 missing vdev we should be able to import the pool and mount all
 * datasets. User data that was not modified after the missing device has been
 * added should be recoverable. This means that snapshots created prior to the
 * addition of that device should be completely intact.
 *
 * With 2 missing vdevs, some datasets may fail to mount since there are
 * dataset statistics that are stored as regular metadata. Some data might be
 * recoverable if those vdevs were added recently.
 *
 * With 3 or more missing vdevs, the pool is severely damaged and MOS entries
 * may be missing entirely. Chances of data recovery are very low. Note that
 * there are also risks of performing an inadvertent rewind as we might be
 * missing all the vdevs with the latest uberblocks.
 */
uint64_t	zfs_max_missing_tvds = 0;

/*
 * The parameters below are similar to zfs_max_missing_tvds but are only
 * intended for a preliminary open of the pool with an untrusted config which
 * might be incomplete or out-dated.
 *
 * We are more tolerant for pools opened from a cachefile since we could have
 * an out-dated cachefile where a device removal was not registered.
 * We could have set the limit arbitrarily high but in the case where devices
 * are really missing we would want to return the proper error codes; we chose
 * SPA_DVAS_PER_BP - 1 so that some copies of the MOS would still be available
 * and we get a chance to retrieve the trusted config.
 */
uint64_t	zfs_max_missing_tvds_cachefile = SPA_DVAS_PER_BP - 1;

/*
 * In the case where config was assembled by scanning device paths (/dev/dsks
 * by default) we are less tolerant since all the existing devices should have
 * been detected and we want spa_load to return the right error codes.
 */
uint64_t	zfs_max_missing_tvds_scan = 0;

/*
 * Interval in seconds at which to poll spare vdevs for health.
 * Setting this to zero disables spare polling.
 * Set to three hours by default.
 */
uint_t		spa_spare_poll_interval_seconds = 60 * 60 * 3;

/*
 * Debugging aid that pauses spa_sync() towards the end.
 */
boolean_t	zfs_pause_spa_sync = B_FALSE;

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

/*
 * Add a (source=src, propname=propval) list to an nvlist.
 */
static void
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;

	VERIFY(nvlist_alloc(&propval, NV_UNIQUE_NAME, KM_SLEEP) == 0);
	VERIFY(nvlist_add_uint64(propval, ZPROP_SOURCE, src) == 0);

	if (strval != NULL)
		VERIFY(nvlist_add_string(propval, ZPROP_VALUE, strval) == 0);
	else
		VERIFY(nvlist_add_uint64(propval, ZPROP_VALUE, intval) == 0);

	VERIFY(nvlist_add_nvlist(nvl, propname, propval) == 0);
	nvlist_free(propval);
}

/*
 * Get property values from the spa configuration.
 */
static void
spa_prop_get_config(spa_t *spa, nvlist_t **nvp)
{
	vdev_t *rvd = spa->spa_root_vdev;
	dsl_pool_t *pool = spa->spa_dsl_pool;
	uint64_t size, alloc, cap, version;
	zprop_source_t src = ZPROP_SRC_NONE;
	spa_config_dirent_t *dp;
	metaslab_class_t *mc = spa_normal_class(spa);

	ASSERT(MUTEX_HELD(&spa->spa_props_lock));

	if (rvd != NULL) {
		alloc = metaslab_class_get_alloc(mc);
		alloc += metaslab_class_get_alloc(spa_special_class(spa));
		alloc += metaslab_class_get_alloc(spa_dedup_class(spa));

		size = metaslab_class_get_space(mc);
		size += metaslab_class_get_space(spa_special_class(spa));
		size += metaslab_class_get_space(spa_dedup_class(spa));

		spa_prop_add_list(*nvp, ZPOOL_PROP_NAME, spa_name(spa), 0, src);
		spa_prop_add_list(*nvp, ZPOOL_PROP_SIZE, NULL, size, src);
		spa_prop_add_list(*nvp, ZPOOL_PROP_ALLOCATED, NULL, alloc, src);
		spa_prop_add_list(*nvp, ZPOOL_PROP_FREE, NULL,
		    size - alloc, src);
		spa_prop_add_list(*nvp, ZPOOL_PROP_CHECKPOINT, NULL,
		    spa->spa_checkpoint_info.sci_dspace, src);

		spa_prop_add_list(*nvp, ZPOOL_PROP_FRAGMENTATION, NULL,
		    metaslab_class_fragmentation(mc), src);
		spa_prop_add_list(*nvp, ZPOOL_PROP_EXPANDSZ, NULL,
		    metaslab_class_expandable_space(mc), src);
		spa_prop_add_list(*nvp, ZPOOL_PROP_READONLY, NULL,
		    (spa_mode(spa) == FREAD), src);

		cap = (size == 0) ? 0 : (alloc * 100 / size);
		spa_prop_add_list(*nvp, ZPOOL_PROP_CAPACITY, NULL, cap, src);

		spa_prop_add_list(*nvp, ZPOOL_PROP_DEDUPRATIO, NULL,
		    ddt_get_pool_dedup_ratio(spa), src);

		spa_prop_add_list(*nvp, ZPOOL_PROP_HEALTH, NULL,
		    rvd->vdev_state, src);

		version = spa_version(spa);
		if (version == zpool_prop_default_numeric(ZPOOL_PROP_VERSION))
			src = ZPROP_SRC_DEFAULT;
		else
			src = ZPROP_SRC_LOCAL;
		spa_prop_add_list(*nvp, ZPOOL_PROP_VERSION, NULL, version, src);
	}

	if (pool != NULL) {
		/*
		 * The $FREE directory was introduced in SPA_VERSION_DEADLISTS,
		 * when opening pools before this version freedir will be NULL.
		 */
		if (pool->dp_free_dir != NULL) {
			spa_prop_add_list(*nvp, ZPOOL_PROP_FREEING, NULL,
			    dsl_dir_phys(pool->dp_free_dir)->dd_used_bytes,
			    src);
		} else {
			spa_prop_add_list(*nvp, ZPOOL_PROP_FREEING,
			    NULL, 0, src);
		}

		if (pool->dp_leak_dir != NULL) {
			spa_prop_add_list(*nvp, ZPOOL_PROP_LEAKED, NULL,
			    dsl_dir_phys(pool->dp_leak_dir)->dd_used_bytes,
			    src);
		} else {
			spa_prop_add_list(*nvp, ZPOOL_PROP_LEAKED,
			    NULL, 0, src);
		}
	}

	spa_prop_add_list(*nvp, ZPOOL_PROP_GUID, NULL, spa_guid(spa), src);

	if (spa->spa_comment != NULL) {
		spa_prop_add_list(*nvp, ZPOOL_PROP_COMMENT, spa->spa_comment,
		    0, ZPROP_SRC_LOCAL);
	}

	if (spa->spa_root != NULL)
		spa_prop_add_list(*nvp, ZPOOL_PROP_ALTROOT, spa->spa_root,
		    0, ZPROP_SRC_LOCAL);

	if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_BLOCKS)) {
		spa_prop_add_list(*nvp, ZPOOL_PROP_MAXBLOCKSIZE, NULL,
		    MIN(zfs_max_recordsize, SPA_MAXBLOCKSIZE), ZPROP_SRC_NONE);
	} else {
		spa_prop_add_list(*nvp, ZPOOL_PROP_MAXBLOCKSIZE, NULL,
		    SPA_OLD_MAXBLOCKSIZE, ZPROP_SRC_NONE);
	}

	if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_DNODE)) {
		spa_prop_add_list(*nvp, ZPOOL_PROP_MAXDNODESIZE, NULL,
		    DNODE_MAX_SIZE, ZPROP_SRC_NONE);
	} else {
		spa_prop_add_list(*nvp, ZPOOL_PROP_MAXDNODESIZE, NULL,
		    DNODE_MIN_SIZE, ZPROP_SRC_NONE);
	}

	if ((dp = list_head(&spa->spa_config_list)) != NULL) {
		if (dp->scd_path == NULL) {
			spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE,
			    "none", 0, ZPROP_SRC_LOCAL);
		} else if (strcmp(dp->scd_path, spa_config_path) != 0) {
			spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE,
			    dp->scd_path, 0, ZPROP_SRC_LOCAL);
		}
	}
}

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

	VERIFY(nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_SLEEP) == 0);

	mutex_enter(&spa->spa_props_lock);

	/*
	 * Get properties from the spa config.
	 */
	spa_prop_get_config(spa, nvp);

	/* If no pool property object, no more prop to get. */
	if (mos == NULL || 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)) == ZPOOL_PROP_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);
				dsl_pool_config_enter(dp, FTAG);
				err = dsl_dataset_hold_obj(dp,
				    za.za_first_integer, FTAG, &ds);
				if (err != 0) {
					dsl_pool_config_exit(dp, FTAG);
					break;
				}

				strval = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN,
				    KM_SLEEP);
				dsl_dataset_name(ds, strval);
				dsl_dataset_rele(ds, FTAG);
				dsl_pool_config_exit(dp, FTAG);
			} else {
				strval = NULL;
				intval = za.za_first_integer;
			}

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

			if (strval != NULL)
				kmem_free(strval, ZFS_MAX_DATASET_NAME_LEN);

			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;
			}
			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);
		*nvp = NULL;
		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 = 0;
	boolean_t has_feature = B_FALSE;

	elem = NULL;
	while ((elem = nvlist_next_nvpair(props, elem)) != NULL) {
		uint64_t intval;
		char *strval, *slash, *check, *fname;
		const char *propname = nvpair_name(elem);
		zpool_prop_t prop = zpool_name_to_prop(propname);

		switch (prop) {
		case ZPOOL_PROP_INVAL:
			if (!zpool_prop_feature(propname)) {
				error = SET_ERROR(EINVAL);
				break;
			}

			/*
			 * Sanitize the input.
			 */
			if (nvpair_type(elem) != DATA_TYPE_UINT64) {
				error = SET_ERROR(EINVAL);
				break;
			}

			if (nvpair_value_uint64(elem, &intval) != 0) {
				error = SET_ERROR(EINVAL);
				break;
			}

			if (intval != 0) {
				error = SET_ERROR(EINVAL);
				break;
			}

			fname = strchr(propname, '@') + 1;
			if (zfeature_lookup_name(fname, NULL) != 0) {
				error = SET_ERROR(EINVAL);
				break;
			}

			has_feature = B_TRUE;
			break;

		case ZPOOL_PROP_VERSION:
			error = nvpair_value_uint64(elem, &intval);
			if (!error &&
			    (intval < spa_version(spa) ||
			    intval > SPA_VERSION_BEFORE_FEATURES ||
			    has_feature))
				error = SET_ERROR(EINVAL);
			break;

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

		case ZPOOL_PROP_MULTIHOST:
			error = nvpair_value_uint64(elem, &intval);
			if (!error && intval > 1)
				error = SET_ERROR(EINVAL);

			if (!error && !spa_get_hostid())
				error = SET_ERROR(ENOTSUP);

			break;

		case ZPOOL_PROP_BOOTFS:
			/*
			 * If the pool version is less than SPA_VERSION_BOOTFS,
			 * or the pool is still being created (version == 0),
			 * the bootfs property cannot be set.
			 */
			if (spa_version(spa) < SPA_VERSION_BOOTFS) {
				error = SET_ERROR(ENOTSUP);
				break;
			}

			/*
			 * Make sure the vdev config is bootable
			 */
			if (!vdev_is_bootable(spa->spa_root_vdev)) {
				error = SET_ERROR(ENOTSUP);
				break;
			}

			reset_bootfs = 1;

			error = nvpair_value_string(elem, &strval);

			if (!error) {
				objset_t *os;
				uint64_t propval;

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

				error = dmu_objset_hold(strval, FTAG, &os);
				if (error != 0)
					break;

				/*
				 * Must be ZPL, and its property settings
				 * must be supported.
				 */

				if (dmu_objset_type(os) != DMU_OST_ZFS) {
					error = SET_ERROR(ENOTSUP);
				} else if ((error =
				    dsl_prop_get_int_ds(dmu_objset_ds(os),
				    zfs_prop_to_name(ZFS_PROP_COMPRESSION),
				    &propval)) == 0 &&
				    !BOOTFS_COMPRESS_VALID(propval)) {
					error = SET_ERROR(ENOTSUP);
				} else {
					objnum = dmu_objset_id(os);
				}
				dmu_objset_rele(os, FTAG);
			}
			break;

		case ZPOOL_PROP_FAILUREMODE:
			error = nvpair_value_uint64(elem, &intval);
			if (!error && (intval < ZIO_FAILURE_MODE_WAIT ||
			    intval > ZIO_FAILURE_MODE_PANIC))
				error = SET_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_suspended(spa)) {
				spa->spa_failmode = intval;
				error = SET_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 = SET_ERROR(EINVAL);
				break;
			}

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

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

		case ZPOOL_PROP_COMMENT:
			if ((error = nvpair_value_string(elem, &strval)) != 0)
				break;
			for (check = strval; *check != '\0'; check++) {
				/*
				 * The kernel doesn't have an easy isprint()
				 * check.  For this kernel check, we merely
				 * check ASCII apart from DEL.  Fix this if
				 * there is an easy-to-use kernel isprint().
				 */
				if (*check >= 0x7f) {
					error = SET_ERROR(EINVAL);
					break;
				}
			}
			if (strlen(strval) > ZPROP_MAX_COMMENT)
				error = E2BIG;
			break;

		case ZPOOL_PROP_DEDUPDITTO:
			if (spa_version(spa) < SPA_VERSION_DEDUP)
				error = SET_ERROR(ENOTSUP);
			else
				error = nvpair_value_uint64(elem, &intval);
			if (error == 0 &&
			    intval != 0 && intval < ZIO_DEDUPDITTO_MIN)
				error = SET_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);
}

void
spa_configfile_set(spa_t *spa, nvlist_t *nvp, boolean_t need_sync)
{
	char *cachefile;
	spa_config_dirent_t *dp;

	if (nvlist_lookup_string(nvp, zpool_prop_to_name(ZPOOL_PROP_CACHEFILE),
	    &cachefile) != 0)
		return;

	dp = kmem_alloc(sizeof (spa_config_dirent_t),
	    KM_SLEEP);

	if (cachefile[0] == '\0')
		dp->scd_path = spa_strdup(spa_config_path);
	else if (strcmp(cachefile, "none") == 0)
		dp->scd_path = NULL;
	else
		dp->scd_path = spa_strdup(cachefile);

	list_insert_head(&spa->spa_config_list, dp);
	if (need_sync)
		spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE);
}

int
spa_prop_set(spa_t *spa, nvlist_t *nvp)
{
	int error;
	nvpair_t *elem = NULL;
	boolean_t need_sync = B_FALSE;

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

	while ((elem = nvlist_next_nvpair(nvp, elem)) != NULL) {
		zpool_prop_t prop = zpool_name_to_prop(nvpair_name(elem));

		if (prop == ZPOOL_PROP_CACHEFILE ||
		    prop == ZPOOL_PROP_ALTROOT ||
		    prop == ZPOOL_PROP_READONLY)
			continue;

		if (prop == ZPOOL_PROP_VERSION || prop == ZPOOL_PROP_INVAL) {
			uint64_t ver;

			if (prop == ZPOOL_PROP_VERSION) {
				VERIFY(nvpair_value_uint64(elem, &ver) == 0);
			} else {
				ASSERT(zpool_prop_feature(nvpair_name(elem)));
				ver = SPA_VERSION_FEATURES;
				need_sync = B_TRUE;
			}

			/* Save time if the version is already set. */
			if (ver == spa_version(spa))
				continue;

			/*
			 * In addition to the pool directory object, we might
			 * create the pool properties object, the features for
			 * read object, the features for write object, or the
			 * feature descriptions object.
			 */
			error = dsl_sync_task(spa->spa_name, NULL,
			    spa_sync_version, &ver,
			    6, ZFS_SPACE_CHECK_RESERVED);
			if (error)
				return (error);
			continue;
		}

		need_sync = B_TRUE;
		break;
	}

	if (need_sync) {
		return (dsl_sync_task(spa->spa_name, NULL, spa_sync_props,
		    nvp, 6, ZFS_SPACE_CHECK_RESERVED));
	}

	return (0);
}

/*
 * 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;
	}
}

/*ARGSUSED*/
static int
spa_change_guid_check(void *arg, dmu_tx_t *tx)
{
	uint64_t *newguid = arg;
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;
	vdev_t *rvd = spa->spa_root_vdev;
	uint64_t vdev_state;

	if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) {
		int error = (spa_has_checkpoint(spa)) ?
		    ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT;
		return (SET_ERROR(error));
	}

	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
	vdev_state = rvd->vdev_state;
	spa_config_exit(spa, SCL_STATE, FTAG);

	if (vdev_state != VDEV_STATE_HEALTHY)
		return (SET_ERROR(ENXIO));

	ASSERT3U(spa_guid(spa), !=, *newguid);

	return (0);
}

static void
spa_change_guid_sync(void *arg, dmu_tx_t *tx)
{
	uint64_t *newguid = arg;
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;
	uint64_t oldguid;
	vdev_t *rvd = spa->spa_root_vdev;

	oldguid = spa_guid(spa);

	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
	rvd->vdev_guid = *newguid;
	rvd->vdev_guid_sum += (*newguid - oldguid);
	vdev_config_dirty(rvd);
	spa_config_exit(spa, SCL_STATE, FTAG);

	spa_history_log_internal(spa, "guid change", tx, "old=%llu new=%llu",
	    oldguid, *newguid);
}

/*
 * Change the GUID for the pool.  This is done so that we can later
 * re-import a pool built from a clone of our own vdevs.  We will modify
 * the root vdev's guid, our own pool guid, and then mark all of our
 * vdevs dirty.  Note that we must make sure that all our vdevs are
 * online when we do this, or else any vdevs that weren't present
 * would be orphaned from our pool.  We are also going to issue a
 * sysevent to update any watchers.
 */
int
spa_change_guid(spa_t *spa)
{
	int error;
	uint64_t guid;

	mutex_enter(&spa->spa_vdev_top_lock);
	mutex_enter(&spa_namespace_lock);
	guid = spa_generate_guid(NULL);

	error = dsl_sync_task(spa->spa_name, spa_change_guid_check,
	    spa_change_guid_sync, &guid, 5, ZFS_SPACE_CHECK_RESERVED);

	if (error == 0) {
		spa_write_cachefile(spa, B_FALSE, B_TRUE);
		spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_REGUID);
	}

	mutex_exit(&spa_namespace_lock);
	mutex_exit(&spa->spa_vdev_top_lock);

	return (error);
}

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

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

	ret = memcmp(&sa->se_bookmark, &sb->se_bookmark,
	    sizeof (zbookmark_phys_t));

	return (TREE_ISIGN(ret));
}

/*
 * 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));
}

static void
spa_taskqs_init(spa_t *spa, zio_type_t t, zio_taskq_type_t q)
{
	const zio_taskq_info_t *ztip = &zio_taskqs[t][q];
	enum zti_modes mode = ztip->zti_mode;
	uint_t value = ztip->zti_value;
	uint_t count = ztip->zti_count;
	spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q];
	char name[32];
	uint_t flags = 0;
	boolean_t batch = B_FALSE;

	if (mode == ZTI_MODE_NULL) {
		tqs->stqs_count = 0;
		tqs->stqs_taskq = NULL;
		return;
	}

	ASSERT3U(count, >, 0);

	tqs->stqs_count = count;
	tqs->stqs_taskq = kmem_alloc(count * sizeof (taskq_t *), KM_SLEEP);

	switch (mode) {
	case ZTI_MODE_FIXED:
		ASSERT3U(value, >=, 1);
		value = MAX(value, 1);
		break;

	case ZTI_MODE_BATCH:
		batch = B_TRUE;
		flags |= TASKQ_THREADS_CPU_PCT;
		value = zio_taskq_batch_pct;
		break;

	default:
		panic("unrecognized mode for %s_%s taskq (%u:%u) in "
		    "spa_activate()",
		    zio_type_name[t], zio_taskq_types[q], mode, value);
		break;
	}

	for (uint_t i = 0; i < count; i++) {
		taskq_t *tq;

		if (count > 1) {
			(void) snprintf(name, sizeof (name), "%s_%s_%u",
			    zio_type_name[t], zio_taskq_types[q], i);
		} else {
			(void) snprintf(name, sizeof (name), "%s_%s",
			    zio_type_name[t], zio_taskq_types[q]);
		}

		if (zio_taskq_sysdc && spa->spa_proc != &p0) {
			if (batch)
				flags |= TASKQ_DC_BATCH;

			tq = taskq_create_sysdc(name, value, 50, INT_MAX,
			    spa->spa_proc, zio_taskq_basedc, flags);
		} else {
			pri_t pri = maxclsyspri;
			/*
			 * The write issue taskq can be extremely CPU
			 * intensive.  Run it at slightly lower priority
			 * than the other taskqs.
			 */
			if (t == ZIO_TYPE_WRITE && q == ZIO_TASKQ_ISSUE)
				pri--;

			tq = taskq_create_proc(name, value, pri, 50,
			    INT_MAX, spa->spa_proc, flags);
		}

		tqs->stqs_taskq[i] = tq;
	}
}

static void
spa_taskqs_fini(spa_t *spa, zio_type_t t, zio_taskq_type_t q)
{
	spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q];

	if (tqs->stqs_taskq == NULL) {
		ASSERT0(tqs->stqs_count);
		return;
	}

	for (uint_t i = 0; i < tqs->stqs_count; i++) {
		ASSERT3P(tqs->stqs_taskq[i], !=, NULL);
		taskq_destroy(tqs->stqs_taskq[i]);
	}

	kmem_free(tqs->stqs_taskq, tqs->stqs_count * sizeof (taskq_t *));
	tqs->stqs_taskq = NULL;
}

/*
 * Dispatch a task to the appropriate taskq for the ZFS I/O type and priority.
 * Note that a type may have multiple discrete taskqs to avoid lock contention
 * on the taskq itself. In that case we choose which taskq at random by using
 * the low bits of gethrtime().
 */
void
spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
    task_func_t *func, void *arg, uint_t flags, taskq_ent_t *ent)
{
	spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q];
	taskq_t *tq;

	ASSERT3P(tqs->stqs_taskq, !=, NULL);
	ASSERT3U(tqs->stqs_count, !=, 0);

	if (tqs->stqs_count == 1) {
		tq = tqs->stqs_taskq[0];
	} else {
		tq = tqs->stqs_taskq[gethrtime() % tqs->stqs_count];
	}

	taskq_dispatch_ent(tq, func, arg, flags, ent);
}

static void
spa_create_zio_taskqs(spa_t *spa)
{
	for (int t = 0; t < ZIO_TYPES; t++) {
		for (int q = 0; q < ZIO_TASKQ_TYPES; q++) {
			spa_taskqs_init(spa, t, q);
		}
	}
}

#ifdef _KERNEL
static void
spa_thread(void *arg)
{
	callb_cpr_t cprinfo;

	spa_t *spa = arg;
	user_t *pu = PTOU(curproc);

	CALLB_CPR_INIT(&cprinfo, &spa->spa_proc_lock, callb_generic_cpr,
	    spa->spa_name);

	ASSERT(curproc != &p0);
	(void) snprintf(pu->u_psargs, sizeof (pu->u_psargs),
	    "zpool-%s", spa->spa_name);
	(void) strlcpy(pu->u_comm, pu->u_psargs, sizeof (pu->u_comm));

	/* bind this thread to the requested psrset */
	if (zio_taskq_psrset_bind != PS_NONE) {
		pool_lock();
		mutex_enter(&cpu_lock);
		mutex_enter(&pidlock);
		mutex_enter(&curproc->p_lock);

		if (cpupart_bind_thread(curthread, zio_taskq_psrset_bind,
		    0, NULL, NULL) == 0)  {
			curthread->t_bind_pset = zio_taskq_psrset_bind;
		} else {
			cmn_err(CE_WARN,
			    "Couldn't bind process for zfs pool \"%s\" to "
			    "pset %d\n", spa->spa_name, zio_taskq_psrset_bind);
		}

		mutex_exit(&curproc->p_lock);
		mutex_exit(&pidlock);
		mutex_exit(&cpu_lock);
		pool_unlock();
	}

	if (zio_taskq_sysdc) {
		sysdc_thread_enter(curthread, 100, 0);
	}

	spa->spa_proc = curproc;
	spa->spa_did = curthread->t_did;

	spa_create_zio_taskqs(spa);

	mutex_enter(&spa->spa_proc_lock);
	ASSERT(spa->spa_proc_state == SPA_PROC_CREATED);

	spa->spa_proc_state = SPA_PROC_ACTIVE;
	cv_broadcast(&spa->spa_proc_cv);

	CALLB_CPR_SAFE_BEGIN(&cprinfo);
	while (spa->spa_proc_state == SPA_PROC_ACTIVE)
		cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock);
	CALLB_CPR_SAFE_END(&cprinfo, &spa->spa_proc_lock);

	ASSERT(spa->spa_proc_state == SPA_PROC_DEACTIVATE);
	spa->spa_proc_state = SPA_PROC_GONE;
	spa->spa_proc = &p0;
	cv_broadcast(&spa->spa_proc_cv);
	CALLB_CPR_EXIT(&cprinfo);	/* drops spa_proc_lock */

	mutex_enter(&curproc->p_lock);
	lwp_exit();
}
#endif

/*
 * Activate an uninitialized pool.
 */
static void
spa_activate(spa_t *spa, int mode)
{
	ASSERT(spa->spa_state == POOL_STATE_UNINITIALIZED);

	spa->spa_state = POOL_STATE_ACTIVE;
	spa->spa_mode = mode;

	spa->spa_normal_class = metaslab_class_create(spa, zfs_metaslab_ops);
	spa->spa_log_class = metaslab_class_create(spa, zfs_metaslab_ops);
	spa->spa_special_class = metaslab_class_create(spa, zfs_metaslab_ops);
	spa->spa_dedup_class = metaslab_class_create(spa, zfs_metaslab_ops);

	/* Try to create a covering process */
	mutex_enter(&spa->spa_proc_lock);
	ASSERT(spa->spa_proc_state == SPA_PROC_NONE);
	ASSERT(spa->spa_proc == &p0);
	spa->spa_did = 0;

	/* Only create a process if we're going to be around a while. */
	if (spa_create_process && strcmp(spa->spa_name, TRYIMPORT_NAME) != 0) {
		if (newproc(spa_thread, (caddr_t)spa, syscid, maxclsyspri,
		    NULL, 0) == 0) {
			spa->spa_proc_state = SPA_PROC_CREATED;
			while (spa->spa_proc_state == SPA_PROC_CREATED) {
				cv_wait(&spa->spa_proc_cv,
				    &spa->spa_proc_lock);
			}
			ASSERT(spa->spa_proc_state == SPA_PROC_ACTIVE);
			ASSERT(spa->spa_proc != &p0);
			ASSERT(spa->spa_did != 0);
		} else {
#ifdef _KERNEL
			cmn_err(CE_WARN,
			    "Couldn't create process for zfs pool \"%s\"\n",
			    spa->spa_name);
#endif
		}
	}
	mutex_exit(&spa->spa_proc_lock);

	/* If we didn't create a process, we need to create our taskqs. */
	if (spa->spa_proc == &p0) {
		spa_create_zio_taskqs(spa);
	}

	for (size_t i = 0; i < TXG_SIZE; i++) {
		spa->spa_txg_zio[i] = zio_root(spa, NULL, NULL,
		    ZIO_FLAG_CANFAIL);
	}

	list_create(&spa->spa_config_dirty_list, sizeof (vdev_t),
	    offsetof(vdev_t, vdev_config_dirty_node));
	list_create(&spa->spa_evicting_os_list, sizeof (objset_t),
	    offsetof(objset_t, os_evicting_node));
	list_create(&spa->spa_state_dirty_list, sizeof (vdev_t),
	    offsetof(vdev_t, vdev_state_dirty_node));

	txg_list_create(&spa->spa_vdev_txg_list, spa,
	    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));

	spa_keystore_init(&spa->spa_keystore);

	/*
	 * The taskq to upgrade datasets in this pool. Currently used by
	 * feature SPA_FEATURE_USEROBJ_ACCOUNTING/SPA_FEATURE_PROJECT_QUOTA.
	 */
	spa->spa_upgrade_taskq = taskq_create("z_upgrade", boot_ncpus,
	    minclsyspri, 1, INT_MAX, TASKQ_DYNAMIC);
}

/*
 * Opposite of spa_activate().
 */
static void
spa_deactivate(spa_t *spa)
{
	ASSERT(spa->spa_sync_on == B_FALSE);
	ASSERT(spa->spa_dsl_pool == NULL);
	ASSERT(spa->spa_root_vdev == NULL);
	ASSERT(spa->spa_async_zio_root == NULL);
	ASSERT(spa->spa_state != POOL_STATE_UNINITIALIZED);

	spa_evicting_os_wait(spa);

	if (spa->spa_upgrade_taskq) {
		taskq_destroy(spa->spa_upgrade_taskq);
		spa->spa_upgrade_taskq = NULL;
	}

	txg_list_destroy(&spa->spa_vdev_txg_list);

	list_destroy(&spa->spa_config_dirty_list);
	list_destroy(&spa->spa_evicting_os_list);
	list_destroy(&spa->spa_state_dirty_list);

	for (int t = 0; t < ZIO_TYPES; t++) {
		for (int q = 0; q < ZIO_TASKQ_TYPES; q++) {
			spa_taskqs_fini(spa, t, q);
		}
	}

	for (size_t i = 0; i < TXG_SIZE; i++) {
		ASSERT3P(spa->spa_txg_zio[i], !=, NULL);
		VERIFY0(zio_wait(spa->spa_txg_zio[i]));
		spa->spa_txg_zio[i] = 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;

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

	metaslab_class_destroy(spa->spa_dedup_class);
	spa->spa_dedup_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_keystore_fini(&spa->spa_keystore);

	spa->spa_state = POOL_STATE_UNINITIALIZED;

	mutex_enter(&spa->spa_proc_lock);
	if (spa->spa_proc_state != SPA_PROC_NONE) {
		ASSERT(spa->spa_proc_state == SPA_PROC_ACTIVE);
		spa->spa_proc_state = SPA_PROC_DEACTIVATE;
		cv_broadcast(&spa->spa_proc_cv);
		while (spa->spa_proc_state == SPA_PROC_DEACTIVATE) {
			ASSERT(spa->spa_proc != &p0);
			cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock);
		}
		ASSERT(spa->spa_proc_state == SPA_PROC_GONE);
		spa->spa_proc_state = SPA_PROC_NONE;
	}
	ASSERT(spa->spa_proc == &p0);
	mutex_exit(&spa->spa_proc_lock);

	/*
	 * We want to make sure spa_thread() has actually exited the ZFS
	 * module, so that the module can't be unloaded out from underneath
	 * it.
	 */
	if (spa->spa_did != 0) {
		thread_join(spa->spa_did);
		spa->spa_did = 0;
	}
}

/*
 * 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 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);

	error = nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	    &child, &children);

	if (error == ENOENT)
		return (0);

	if (error) {
		vdev_free(*vdp);
		*vdp = NULL;
		return (SET_ERROR(EINVAL));
	}

	for (int 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);
}

static boolean_t
spa_should_flush_logs_on_unload(spa_t *spa)
{
	if (!spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP))
		return (B_FALSE);

	if (!spa_writeable(spa))
		return (B_FALSE);

	if (!spa->spa_sync_on)
		return (B_FALSE);

	if (spa_state(spa) != POOL_STATE_EXPORTED)
		return (B_FALSE);

	if (zfs_keep_log_spacemaps_at_export)
		return (B_FALSE);

	return (B_TRUE);
}

/*
 * Opens a transaction that will set the flag that will instruct
 * spa_sync to attempt to flush all the metaslabs for that txg.
 */
static void
spa_unload_log_sm_flush_all(spa_t *spa)
{
	dmu_tx_t *tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir);

	VERIFY0(dmu_tx_assign(tx, TXG_WAIT));

	ASSERT3U(spa->spa_log_flushall_txg, ==, 0);
	spa->spa_log_flushall_txg = dmu_tx_get_txg(tx);

	dmu_tx_commit(tx);
	txg_wait_synced(spa_get_dsl(spa), spa->spa_log_flushall_txg);
}

static void
spa_unload_log_sm_metadata(spa_t *spa)
{
	void *cookie = NULL;
	spa_log_sm_t *sls;

	while ((sls = avl_destroy_nodes(&spa->spa_sm_logs_by_txg,
	    &cookie)) != NULL) {
		VERIFY0(sls->sls_mscount);
		kmem_free(sls, sizeof (spa_log_sm_t));
	}

	for (log_summary_entry_t *e = list_head(&spa->spa_log_summary);
	    e != NULL; e = list_head(&spa->spa_log_summary)) {
		VERIFY0(e->lse_mscount);
		list_remove(&spa->spa_log_summary, e);
		kmem_free(e, sizeof (log_summary_entry_t));
	}

	spa->spa_unflushed_stats.sus_nblocks = 0;
	spa->spa_unflushed_stats.sus_memused = 0;
	spa->spa_unflushed_stats.sus_blocklimit = 0;
}

/*
 * Opposite of spa_load().
 */
static void
spa_unload(spa_t *spa)
{
	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	ASSERT(spa_state(spa) != POOL_STATE_UNINITIALIZED);

	spa_import_progress_remove(spa);
	spa_load_note(spa, "UNLOADING");

	/*
	 * If the log space map feature is enabled and the pool is getting
	 * exported (but not destroyed), we want to spend some time flushing
	 * as many metaslabs as we can in an attempt to destroy log space
	 * maps and save import time.
	 */
	if (spa_should_flush_logs_on_unload(spa))
		spa_unload_log_sm_flush_all(spa);

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

	if (spa->spa_root_vdev) {
		vdev_t *root_vdev = spa->spa_root_vdev;
		vdev_initialize_stop_all(root_vdev, VDEV_INITIALIZE_ACTIVE);
		vdev_trim_stop_all(root_vdev, VDEV_TRIM_ACTIVE);
		vdev_autotrim_stop_all(spa);
	}

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

	/*
	 * This ensures that there is no async metaslab prefetching
	 * while we attempt to unload the spa.
	 */
	if (spa->spa_root_vdev != NULL) {
		for (int c = 0; c < spa->spa_root_vdev->vdev_children; c++) {
			vdev_t *vc = spa->spa_root_vdev->vdev_child[c];
			if (vc->vdev_mg != NULL)
				taskq_wait(vc->vdev_mg->mg_taskq);
		}
	}

	if (spa->spa_mmp.mmp_thread)
		mmp_thread_stop(spa);

	/*
	 * Wait for any outstanding async I/O to complete.
	 */
	if (spa->spa_async_zio_root != NULL) {
		for (int i = 0; i < max_ncpus; i++)
			(void) zio_wait(spa->spa_async_zio_root[i]);
		kmem_free(spa->spa_async_zio_root, max_ncpus * sizeof (void *));
		spa->spa_async_zio_root = NULL;
	}

	if (spa->spa_vdev_removal != NULL) {
		spa_vdev_removal_destroy(spa->spa_vdev_removal);
		spa->spa_vdev_removal = NULL;
	}

	if (spa->spa_condense_zthr != NULL) {
		zthr_destroy(spa->spa_condense_zthr);
		spa->spa_condense_zthr = NULL;
	}

	if (spa->spa_checkpoint_discard_zthr != NULL) {
		zthr_destroy(spa->spa_checkpoint_discard_zthr);
		spa->spa_checkpoint_discard_zthr = NULL;
	}

	spa_condense_fini(spa);

	bpobj_close(&spa->spa_deferred_bpobj);

	spa_config_enter(spa, SCL_ALL, spa, RW_WRITER);

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

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

	ddt_unload(spa);
	spa_unload_log_sm_metadata(spa);

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

	for (int 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;
	}
	spa->spa_spares.sav_count = 0;

	for (int i = 0; i < spa->spa_l2cache.sav_count; i++) {
		vdev_clear_stats(spa->spa_l2cache.sav_vdevs[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_l2cache.sav_count = 0;

	spa->spa_async_suspended = 0;

	spa->spa_indirect_vdevs_loaded = B_FALSE;

	if (spa->spa_comment != NULL) {
		spa_strfree(spa->spa_comment);
		spa->spa_comment = NULL;
	}

	spa_config_exit(spa, SCL_ALL, spa);
}

/*
 * 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.
 */
void
spa_load_spares(spa_t *spa)
{
	nvlist_t **spares;
	uint_t nspares;
	int i;
	vdev_t *vd, *tvd;

#ifndef _KERNEL
	/*
	 * zdb opens both the current state of the pool and the
	 * checkpointed state (if present), with a different spa_t.
	 *
	 * As spare vdevs are shared among open pools, we skip loading
	 * them when we load the checkpointed state of the pool.
	 */
	if (!spa_writeable(spa))
		return;
#endif

	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	/*
	 * 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,
		    B_FALSE)) != 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,
		    B_FALSE)) != 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);
		}

		vd->vdev_top = vd;
		vd->vdev_aux = &spa->spa_spares;

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

		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, VDEV_CONFIG_SPARE);
	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.
 */
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;

#ifndef _KERNEL
	/*
	 * zdb opens both the current state of the pool and the
	 * checkpointed state (if present), with a different spa_t.
	 *
	 * As L2 caches are part of the ARC which is shared among open
	 * pools, we skip loading them when we load the checkpointed
	 * state of the pool.
	 */
	if (!spa_writeable(spa))
		return;
#endif

	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

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

	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);

			vd->vdev_top = vd;
			vd->vdev_aux = sav;

			spa_l2cache_activate(vd);

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

			(void) vdev_validate_aux(vd);

			if (!vdev_is_dead(vd))
				l2arc_add_vdev(spa, vd);
		}
	}

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

		vd = oldvdevs[i];
		if (vd != NULL) {
			ASSERT(vd->vdev_isl2cache);

			if (spa_l2cache_exists(vd->vdev_guid, &pool) &&
			    pool != 0ULL && l2arc_vdev_present(vd))
				l2arc_remove_vdev(vd);
			vdev_clear_stats(vd);
			vdev_free(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 = NULL;
	if (sav->sav_count > 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, VDEV_CONFIG_L2CACHE);
	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;

	error = dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db);
	if (error != 0)
		return (error);

	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,
	    DMU_READ_PREFETCH);
	if (error == 0)
		error = nvlist_unpack(packed, nvsize, value, 0);
	kmem_free(packed, nvsize);

	return (error);
}

/*
 * Concrete top-level vdevs that are not missing and are not logs. At every
 * spa_sync we write new uberblocks to at least SPA_SYNC_MIN_VDEVS core tvds.
 */
static uint64_t
spa_healthy_core_tvds(spa_t *spa)
{
	vdev_t *rvd = spa->spa_root_vdev;
	uint64_t tvds = 0;

	for (uint64_t i = 0; i < rvd->vdev_children; i++) {
		vdev_t *vd = rvd->vdev_child[i];
		if (vd->vdev_islog)
			continue;
		if (vdev_is_concrete(vd) && !vdev_is_dead(vd))
			tvds++;
	}

	return (tvds);
}

/*
 * 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)
{
	for (uint64_t 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) &&
	    vdev_is_concrete(vd)) {
		zfs_post_autoreplace(vd->vdev_spa, vd);
		spa_event_notify(vd->vdev_spa, vd, NULL, ESC_ZFS_VDEV_CHECK);
	}
}

static int
spa_check_for_missing_logs(spa_t *spa)
{
	vdev_t *rvd = spa->spa_root_vdev;

	/*
	 * If we're doing a normal import, then build up any additional
	 * diagnostic information about missing log devices.
	 * We'll pass this up to the user for further processing.
	 */
	if (!(spa->spa_import_flags & ZFS_IMPORT_MISSING_LOG)) {
		nvlist_t **child, *nv;
		uint64_t idx = 0;

		child = kmem_alloc(rvd->vdev_children * sizeof (nvlist_t **),
		    KM_SLEEP);
		VERIFY(nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_SLEEP) == 0);

		for (uint64_t c = 0; c < rvd->vdev_children; c++) {
			vdev_t *tvd = rvd->vdev_child[c];

			/*
			 * We consider a device as missing only if it failed
			 * to open (i.e. offline or faulted is not considered
			 * as missing).
			 */
			if (tvd->vdev_islog &&
			    tvd->vdev_state == VDEV_STATE_CANT_OPEN) {
				child[idx++] = vdev_config_generate(spa, tvd,
				    B_FALSE, VDEV_CONFIG_MISSING);
			}
		}

		if (idx > 0) {
			fnvlist_add_nvlist_array(nv,
			    ZPOOL_CONFIG_CHILDREN, child, idx);
			fnvlist_add_nvlist(spa->spa_load_info,
			    ZPOOL_CONFIG_MISSING_DEVICES, nv);

			for (uint64_t i = 0; i < idx; i++)
				nvlist_free(child[i]);
		}
		nvlist_free(nv);
		kmem_free(child, rvd->vdev_children * sizeof (char **));

		if (idx > 0) {
			spa_load_failed(spa, "some log devices are missing");
			vdev_dbgmsg_print_tree(rvd, 2);
			return (SET_ERROR(ENXIO));
		}
	} else {
		for (uint64_t c = 0; c < rvd->vdev_children; c++) {
			vdev_t *tvd = rvd->vdev_child[c];

			if (tvd->vdev_islog &&
			    tvd->vdev_state == VDEV_STATE_CANT_OPEN) {
				spa_set_log_state(spa, SPA_LOG_CLEAR);
				spa_load_note(spa, "some log devices are "
				    "missing, ZIL is dropped.");
				vdev_dbgmsg_print_tree(rvd, 2);
				break;
			}
		}
	}

	return (0);
}

/*
 * Check for missing log devices
 */
static boolean_t
spa_check_logs(spa_t *spa)
{
	boolean_t rv = B_FALSE;
	dsl_pool_t *dp = spa_get_dsl(spa);

	switch (spa->spa_log_state) {
	case SPA_LOG_MISSING:
		/* need to recheck in case slog has been restored */
	case SPA_LOG_UNKNOWN:
		rv = (dmu_objset_find_dp(dp, dp->dp_root_dir_obj,
		    zil_check_log_chain, NULL, DS_FIND_CHILDREN) != 0);
		if (rv)
			spa_set_log_state(spa, SPA_LOG_MISSING);
		break;
	}
	return (rv);
}

static boolean_t
spa_passivate_log(spa_t *spa)
{
	vdev_t *rvd = spa->spa_root_vdev;
	boolean_t slog_found = B_FALSE;

	ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER));

	if (!spa_has_slogs(spa))
		return (B_FALSE);

	for (int c = 0; c < rvd->vdev_children; c++) {
		vdev_t *tvd = rvd->vdev_child[c];
		metaslab_group_t *mg = tvd->vdev_mg;

		if (tvd->vdev_islog) {
			metaslab_group_passivate(mg);
			slog_found = B_TRUE;
		}
	}

	return (slog_found);
}

static void
spa_activate_log(spa_t *spa)
{
	vdev_t *rvd = spa->spa_root_vdev;

	ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER));

	for (int c = 0; c < rvd->vdev_children; c++) {
		vdev_t *tvd = rvd->vdev_child[c];
		metaslab_group_t *mg = tvd->vdev_mg;

		if (tvd->vdev_islog)
			metaslab_group_activate(mg);
	}
}

int
spa_reset_logs(spa_t *spa)
{
	int error;

	error = dmu_objset_find(spa_name(spa), zil_reset,
	    NULL, DS_FIND_CHILDREN);
	if (error == 0) {
		/*
		 * We successfully offlined the log device, sync out the
		 * current txg so that the "stubby" block can be removed
		 * by zil_sync().
		 */
		txg_wait_synced(spa->spa_dsl_pool, 0);
	}
	return (error);
}

static void
spa_aux_check_removed(spa_aux_vdev_t *sav)
{
	for (int i = 0; i < sav->sav_count; i++)
		spa_check_removed(sav->sav_vdevs[i]);
}

void
spa_claim_notify(zio_t *zio)
{
	spa_t *spa = zio->io_spa;

	if (zio->io_error)
		return;

	mutex_enter(&spa->spa_props_lock);	/* any mutex will do */
	if (spa->spa_claim_max_txg < zio->io_bp->blk_birth)
		spa->spa_claim_max_txg = zio->io_bp->blk_birth;
	mutex_exit(&spa->spa_props_lock);
}

typedef struct spa_load_error {
	uint64_t	sle_meta_count;
	uint64_t	sle_data_count;
} spa_load_error_t;

static void
spa_load_verify_done(zio_t *zio)
{
	blkptr_t *bp = zio->io_bp;
	spa_load_error_t *sle = zio->io_private;
	dmu_object_type_t type = BP_GET_TYPE(bp);
	int error = zio->io_error;
	spa_t *spa = zio->io_spa;

	abd_free(zio->io_abd);
	if (error) {
		if ((BP_GET_LEVEL(bp) != 0 || DMU_OT_IS_METADATA(type)) &&
		    type != DMU_OT_INTENT_LOG)
			atomic_inc_64(&sle->sle_meta_count);
		else
			atomic_inc_64(&sle->sle_data_count);
	}

	mutex_enter(&spa->spa_scrub_lock);
	spa->spa_load_verify_ios--;
	cv_broadcast(&spa->spa_scrub_io_cv);
	mutex_exit(&spa->spa_scrub_lock);
}

/*
 * Maximum number of concurrent scrub i/os to create while verifying
 * a pool while importing it.
 */
int spa_load_verify_maxinflight = 10000;
boolean_t spa_load_verify_metadata = B_TRUE;
boolean_t spa_load_verify_data = B_TRUE;

/*ARGSUSED*/
static int
spa_load_verify_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
    const zbookmark_phys_t *zb, const dnode_phys_t *dnp, void *arg)
{
	if (bp == NULL || BP_IS_HOLE(bp) || BP_IS_EMBEDDED(bp))
		return (0);
	/*
	 * Note: normally this routine will not be called if
	 * spa_load_verify_metadata is not set.  However, it may be useful
	 * to manually set the flag after the traversal has begun.
	 */
	if (!spa_load_verify_metadata)
		return (0);
	if (!BP_IS_METADATA(bp) && !spa_load_verify_data)
		return (0);

	zio_t *rio = arg;
	size_t size = BP_GET_PSIZE(bp);

	mutex_enter(&spa->spa_scrub_lock);
	while (spa->spa_load_verify_ios >= spa_load_verify_maxinflight)
		cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock);
	spa->spa_load_verify_ios++;
	mutex_exit(&spa->spa_scrub_lock);

	zio_nowait(zio_read(rio, spa, bp, abd_alloc_for_io(size, B_FALSE), size,
	    spa_load_verify_done, rio->io_private, ZIO_PRIORITY_SCRUB,
	    ZIO_FLAG_SPECULATIVE | ZIO_FLAG_CANFAIL |
	    ZIO_FLAG_SCRUB | ZIO_FLAG_RAW, zb));
	return (0);
}

/* ARGSUSED */
int
verify_dataset_name_len(dsl_pool_t *dp, dsl_dataset_t *ds, void *arg)
{
	if (dsl_dataset_namelen(ds) >= ZFS_MAX_DATASET_NAME_LEN)
		return (SET_ERROR(ENAMETOOLONG));

	return (0);
}

static int
spa_load_verify(spa_t *spa)
{
	zio_t *rio;
	spa_load_error_t sle = { 0 };
	zpool_load_policy_t policy;
	boolean_t verify_ok = B_FALSE;
	int error = 0;

	zpool_get_load_policy(spa->spa_config, &policy);

	if (policy.zlp_rewind & ZPOOL_NEVER_REWIND)
		return (0);

	dsl_pool_config_enter(spa->spa_dsl_pool, FTAG);
	error = dmu_objset_find_dp(spa->spa_dsl_pool,
	    spa->spa_dsl_pool->dp_root_dir_obj, verify_dataset_name_len, NULL,
	    DS_FIND_CHILDREN);
	dsl_pool_config_exit(spa->spa_dsl_pool, FTAG);
	if (error != 0)
		return (error);

	rio = zio_root(spa, NULL, &sle,
	    ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE);

	if (spa_load_verify_metadata) {
		if (spa->spa_extreme_rewind) {
			spa_load_note(spa, "performing a complete scan of the "
			    "pool since extreme rewind is on. This may take "
			    "a very long time.\n  (spa_load_verify_data=%u, "
			    "spa_load_verify_metadata=%u)",
			    spa_load_verify_data, spa_load_verify_metadata);
		}
		error = traverse_pool(spa, spa->spa_verify_min_txg,
		    TRAVERSE_PRE | TRAVERSE_PREFETCH_METADATA |
		    TRAVERSE_NO_DECRYPT, spa_load_verify_cb, rio);
	}

	(void) zio_wait(rio);

	spa->spa_load_meta_errors = sle.sle_meta_count;
	spa->spa_load_data_errors = sle.sle_data_count;

	if (sle.sle_meta_count != 0 || sle.sle_data_count != 0) {
		spa_load_note(spa, "spa_load_verify found %llu metadata errors "
		    "and %llu data errors", (u_longlong_t)sle.sle_meta_count,
		    (u_longlong_t)sle.sle_data_count);
	}

	if (spa_load_verify_dryrun ||
	    (!error && sle.sle_meta_count <= policy.zlp_maxmeta &&
	    sle.sle_data_count <= policy.zlp_maxdata)) {
		int64_t loss = 0;

		verify_ok = B_TRUE;
		spa->spa_load_txg = spa->spa_uberblock.ub_txg;
		spa->spa_load_txg_ts = spa->spa_uberblock.ub_timestamp;

		loss = spa->spa_last_ubsync_txg_ts - spa->spa_load_txg_ts;
		VERIFY(nvlist_add_uint64(spa->spa_load_info,
		    ZPOOL_CONFIG_LOAD_TIME, spa->spa_load_txg_ts) == 0);
		VERIFY(nvlist_add_int64(spa->spa_load_info,
		    ZPOOL_CONFIG_REWIND_TIME, loss) == 0);
		VERIFY(nvlist_add_uint64(spa->spa_load_info,
		    ZPOOL_CONFIG_LOAD_DATA_ERRORS, sle.sle_data_count) == 0);
	} else {
		spa->spa_load_max_txg = spa->spa_uberblock.ub_txg;
	}

	if (spa_load_verify_dryrun)
		return (0);

	if (error) {
		if (error != ENXIO && error != EIO)
			error = SET_ERROR(EIO);
		return (error);
	}

	return (verify_ok ? 0 : EIO);
}

/*
 * Find a value in the pool props object.
 */
static void
spa_prop_find(spa_t *spa, zpool_prop_t prop, uint64_t *val)
{
	(void) zap_lookup(spa->spa_meta_objset, spa->spa_pool_props_object,
	    zpool_prop_to_name(prop), sizeof (uint64_t), 1, val);
}

/*
 * Find a value in the pool directory object.
 */
static int
spa_dir_prop(spa_t *spa, const char *name, uint64_t *val, boolean_t log_enoent)
{
	int error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
	    name, sizeof (uint64_t), 1, val);

	if (error != 0 && (error != ENOENT || log_enoent)) {
		spa_load_failed(spa, "couldn't get '%s' value in MOS directory "
		    "[error=%d]", name, error);
	}

	return (error);
}

static int
spa_vdev_err(vdev_t *vdev, vdev_aux_t aux, int err)
{
	vdev_set_state(vdev, B_TRUE, VDEV_STATE_CANT_OPEN, aux);
	return (SET_ERROR(err));
}

static void
spa_spawn_aux_threads(spa_t *spa)
{
	ASSERT(spa_writeable(spa));

	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	spa_start_indirect_condensing_thread(spa);

	ASSERT3P(spa->spa_checkpoint_discard_zthr, ==, NULL);
	spa->spa_checkpoint_discard_zthr =
	    zthr_create(spa_checkpoint_discard_thread_check,
	    spa_checkpoint_discard_thread, spa);
}

/*
 * Fix up config after a partly-completed split.  This is done with the
 * ZPOOL_CONFIG_SPLIT nvlist.  Both the splitting pool and the split-off
 * pool have that entry in their config, but only the splitting one contains
 * a list of all the guids of the vdevs that are being split off.
 *
 * This function determines what to do with that list: either rejoin
 * all the disks to the pool, or complete the splitting process.  To attempt
 * the rejoin, each disk that is offlined is marked online again, and
 * we do a reopen() call.  If the vdev label for every disk that was
 * marked online indicates it was successfully split off (VDEV_AUX_SPLIT_POOL)
 * then we call vdev_split() on each disk, and complete the split.
 *
 * Otherwise we leave the config alone, with all the vdevs in place in
 * the original pool.
 */
static void
spa_try_repair(spa_t *spa, nvlist_t *config)
{
	uint_t extracted;
	uint64_t *glist;
	uint_t i, gcount;
	nvlist_t *nvl;
	vdev_t **vd;
	boolean_t attempt_reopen;

	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT, &nvl) != 0)
		return;

	/* check that the config is complete */
	if (nvlist_lookup_uint64_array(nvl, ZPOOL_CONFIG_SPLIT_LIST,
	    &glist, &gcount) != 0)
		return;

	vd = kmem_zalloc(gcount * sizeof (vdev_t *), KM_SLEEP);

	/* attempt to online all the vdevs & validate */
	attempt_reopen = B_TRUE;
	for (i = 0; i < gcount; i++) {
		if (glist[i] == 0)	/* vdev is hole */
			continue;

		vd[i] = spa_lookup_by_guid(spa, glist[i], B_FALSE);
		if (vd[i] == NULL) {
			/*
			 * Don't bother attempting to reopen the disks;
			 * just do the split.
			 */
			attempt_reopen = B_FALSE;
		} else {
			/* attempt to re-online it */
			vd[i]->vdev_offline = B_FALSE;
		}
	}

	if (attempt_reopen) {
		vdev_reopen(spa->spa_root_vdev);

		/* check each device to see what state it's in */
		for (extracted = 0, i = 0; i < gcount; i++) {
			if (vd[i] != NULL &&
			    vd[i]->vdev_stat.vs_aux != VDEV_AUX_SPLIT_POOL)
				break;
			++extracted;
		}
	}

	/*
	 * If every disk has been moved to the new pool, or if we never
	 * even attempted to look at them, then we split them off for
	 * good.
	 */
	if (!attempt_reopen || gcount == extracted) {
		for (i = 0; i < gcount; i++)
			if (vd[i] != NULL)
				vdev_split(vd[i]);
		vdev_reopen(spa->spa_root_vdev);
	}

	kmem_free(vd, gcount * sizeof (vdev_t *));
}

static int
spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type)
{
	char *ereport = FM_EREPORT_ZFS_POOL;
	int error;

	spa->spa_load_state = state;
	(void) spa_import_progress_set_state(spa, spa_load_state(spa));

	gethrestime(&spa->spa_loaded_ts);
	error = spa_load_impl(spa, type, &ereport);

	/*
	 * Don't count references from objsets that are already closed
	 * and are making their way through the eviction process.
	 */
	spa_evicting_os_wait(spa);
	spa->spa_minref = zfs_refcount_count(&spa->spa_refcount);
	if (error) {
		if (error != EEXIST) {
			spa->spa_loaded_ts.tv_sec = 0;
			spa->spa_loaded_ts.tv_nsec = 0;
		}
		if (error != EBADF) {
			(void) zfs_ereport_post(ereport, spa,
			    NULL, NULL, NULL, 0, 0);
		}
	}
	spa->spa_load_state = error ? SPA_LOAD_ERROR : SPA_LOAD_NONE;
	spa->spa_ena = 0;

	(void) spa_import_progress_set_state(spa, spa_load_state(spa));

	return (error);
}

/*
 * Count the number of per-vdev ZAPs associated with all of the vdevs in the
 * vdev tree rooted in the given vd, and ensure that each ZAP is present in the
 * spa's per-vdev ZAP list.
 */
static uint64_t
vdev_count_verify_zaps(vdev_t *vd)
{
	spa_t *spa = vd->vdev_spa;
	uint64_t total = 0;
	if (vd->vdev_top_zap != 0) {
		total++;
		ASSERT0(zap_lookup_int(spa->spa_meta_objset,
		    spa->spa_all_vdev_zaps, vd->vdev_top_zap));
	}
	if (vd->vdev_leaf_zap != 0) {
		total++;
		ASSERT0(zap_lookup_int(spa->spa_meta_objset,
		    spa->spa_all_vdev_zaps, vd->vdev_leaf_zap));
	}

	for (uint64_t i = 0; i < vd->vdev_children; i++) {
		total += vdev_count_verify_zaps(vd->vdev_child[i]);
	}

	return (total);
}

/*
 * Determine whether the activity check is required.
 */
static boolean_t
spa_activity_check_required(spa_t *spa, uberblock_t *ub, nvlist_t *label,
    nvlist_t *config)
{
	uint64_t state = 0;
	uint64_t hostid = 0;
	uint64_t tryconfig_txg = 0;
	uint64_t tryconfig_timestamp = 0;
	uint16_t tryconfig_mmp_seq = 0;
	nvlist_t *nvinfo;

	if (nvlist_exists(config, ZPOOL_CONFIG_LOAD_INFO)) {
		nvinfo = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO);
		(void) nvlist_lookup_uint64(nvinfo, ZPOOL_CONFIG_MMP_TXG,
		    &tryconfig_txg);
		(void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_TIMESTAMP,
		    &tryconfig_timestamp);
		(void) nvlist_lookup_uint16(nvinfo, ZPOOL_CONFIG_MMP_SEQ,
		    &tryconfig_mmp_seq);
	}

	(void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE, &state);

	/*
	 * Disable the MMP activity check - This is used by zdb which
	 * is intended to be used on potentially active pools.
	 */
	if (spa->spa_import_flags & ZFS_IMPORT_SKIP_MMP)
		return (B_FALSE);

	/*
	 * Skip the activity check when the MMP feature is disabled.
	 */
	if (ub->ub_mmp_magic == MMP_MAGIC && ub->ub_mmp_delay == 0)
		return (B_FALSE);

	/*
	 * If the tryconfig_ values are nonzero, they are the results of an
	 * earlier tryimport.  If they all match the uberblock we just found,
	 * then the pool has not changed and we return false so we do not test
	 * a second time.
	 */
	if (tryconfig_txg && tryconfig_txg == ub->ub_txg &&
	    tryconfig_timestamp && tryconfig_timestamp == ub->ub_timestamp &&
	    tryconfig_mmp_seq && tryconfig_mmp_seq ==
	    (MMP_SEQ_VALID(ub) ? MMP_SEQ(ub) : 0))
		return (B_FALSE);

	/*
	 * Allow the activity check to be skipped when importing the pool
	 * on the same host which last imported it.  Since the hostid from
	 * configuration may be stale use the one read from the label.
	 */
	if (nvlist_exists(label, ZPOOL_CONFIG_HOSTID))
		hostid = fnvlist_lookup_uint64(label, ZPOOL_CONFIG_HOSTID);

	if (hostid == spa_get_hostid())
		return (B_FALSE);

	/*
	 * Skip the activity test when the pool was cleanly exported.
	 */
	if (state != POOL_STATE_ACTIVE)
		return (B_FALSE);

	return (B_TRUE);
}

/*
 * Nanoseconds the activity check must watch for changes on-disk.
 */
static uint64_t
spa_activity_check_duration(spa_t *spa, uberblock_t *ub)
{
	uint64_t import_intervals = MAX(zfs_multihost_import_intervals, 1);
	uint64_t multihost_interval = MSEC2NSEC(
	    MMP_INTERVAL_OK(zfs_multihost_interval));
	uint64_t import_delay = MAX(NANOSEC, import_intervals *
	    multihost_interval);

	/*
	 * Local tunables determine a minimum duration except for the case
	 * where we know when the remote host will suspend the pool if MMP
	 * writes do not land.
	 *
	 * See Big Theory comment at the top of mmp.c for the reasoning behind
	 * these cases and times.
	 */

	ASSERT(MMP_IMPORT_SAFETY_FACTOR >= 100);

	if (MMP_INTERVAL_VALID(ub) && MMP_FAIL_INT_VALID(ub) &&
	    MMP_FAIL_INT(ub) > 0) {

		/* MMP on remote host will suspend pool after failed writes */
		import_delay = MMP_FAIL_INT(ub) * MSEC2NSEC(MMP_INTERVAL(ub)) *
		    MMP_IMPORT_SAFETY_FACTOR / 100;

		zfs_dbgmsg("fail_intvals>0 import_delay=%llu ub_mmp "
		    "mmp_fails=%llu ub_mmp mmp_interval=%llu "
		    "import_intervals=%u", import_delay, MMP_FAIL_INT(ub),
		    MMP_INTERVAL(ub), import_intervals);

	} else if (MMP_INTERVAL_VALID(ub) && MMP_FAIL_INT_VALID(ub) &&
	    MMP_FAIL_INT(ub) == 0) {

		/* MMP on remote host will never suspend pool */
		import_delay = MAX(import_delay, (MSEC2NSEC(MMP_INTERVAL(ub)) +
		    ub->ub_mmp_delay) * import_intervals);

		zfs_dbgmsg("fail_intvals=0 import_delay=%llu ub_mmp "
		    "mmp_interval=%llu ub_mmp_delay=%llu "
		    "import_intervals=%u", import_delay, MMP_INTERVAL(ub),
		    ub->ub_mmp_delay, import_intervals);

	} else if (MMP_VALID(ub)) {
		/*
		 * zfs-0.7 compatability case
		 */

		import_delay = MAX(import_delay, (multihost_interval +
		    ub->ub_mmp_delay) * import_intervals);

		zfs_dbgmsg("import_delay=%llu ub_mmp_delay=%llu "
		    "import_intervals=%u leaves=%u", import_delay,
		    ub->ub_mmp_delay, import_intervals,
		    vdev_count_leaves(spa));
	} else {
		/* Using local tunings is the only reasonable option */
		zfs_dbgmsg("pool last imported on non-MMP aware "
		    "host using import_delay=%llu multihost_interval=%llu "
		    "import_intervals=%u", import_delay, multihost_interval,
		    import_intervals);
	}

	return (import_delay);
}

/*
 * Perform the import activity check.  If the user canceled the import or
 * we detected activity then fail.
 */
static int
spa_activity_check(spa_t *spa, uberblock_t *ub, nvlist_t *config)
{
	uint64_t txg = ub->ub_txg;
	uint64_t timestamp = ub->ub_timestamp;
	uint64_t mmp_config = ub->ub_mmp_config;
	uint16_t mmp_seq = MMP_SEQ_VALID(ub) ? MMP_SEQ(ub) : 0;
	uint64_t import_delay;
	hrtime_t import_expire;
	nvlist_t *mmp_label = NULL;
	vdev_t *rvd = spa->spa_root_vdev;
	kcondvar_t cv;
	kmutex_t mtx;
	int error = 0;

	cv_init(&cv, NULL, CV_DEFAULT, NULL);
	mutex_init(&mtx, NULL, MUTEX_DEFAULT, NULL);
	mutex_enter(&mtx);

	/*
	 * If ZPOOL_CONFIG_MMP_TXG is present an activity check was performed
	 * during the earlier tryimport.  If the txg recorded there is 0 then
	 * the pool is known to be active on another host.
	 *
	 * Otherwise, the pool might be in use on another host.  Check for
	 * changes in the uberblocks on disk if necessary.
	 */
	if (nvlist_exists(config, ZPOOL_CONFIG_LOAD_INFO)) {
		nvlist_t *nvinfo = fnvlist_lookup_nvlist(config,
		    ZPOOL_CONFIG_LOAD_INFO);

		if (nvlist_exists(nvinfo, ZPOOL_CONFIG_MMP_TXG) &&
		    fnvlist_lookup_uint64(nvinfo, ZPOOL_CONFIG_MMP_TXG) == 0) {
			vdev_uberblock_load(rvd, ub, &mmp_label);
			error = SET_ERROR(EREMOTEIO);
			goto out;
		}
	}

	import_delay = spa_activity_check_duration(spa, ub);

	/* Add a small random factor in case of simultaneous imports (0-25%) */
	import_delay += import_delay * spa_get_random(250) / 1000;

	import_expire = gethrtime() + import_delay;

	while (gethrtime() < import_expire) {
		(void) spa_import_progress_set_mmp_check(spa,
		    NSEC2SEC(import_expire - gethrtime()));

		vdev_uberblock_load(rvd, ub, &mmp_label);

		if (txg != ub->ub_txg || timestamp != ub->ub_timestamp ||
		    mmp_seq != (MMP_SEQ_VALID(ub) ? MMP_SEQ(ub) : 0)) {
			zfs_dbgmsg("multihost activity detected "
			    "txg %llu ub_txg  %llu "
			    "timestamp %llu ub_timestamp  %llu "
			    "mmp_config %#llx ub_mmp_config %#llx",
			    txg, ub->ub_txg, timestamp, ub->ub_timestamp,
			    mmp_config, ub->ub_mmp_config);

			error = SET_ERROR(EREMOTEIO);
			break;
		}

		if (mmp_label) {
			nvlist_free(mmp_label);
			mmp_label = NULL;
		}

		error = cv_timedwait_sig(&cv, &mtx, ddi_get_lbolt() + hz);
		if (error != -1) {
			error = SET_ERROR(EINTR);
			break;
		}
		error = 0;
	}

out:
	mutex_exit(&mtx);
	mutex_destroy(&mtx);
	cv_destroy(&cv);

	/*
	 * If the pool is determined to be active store the status in the
	 * spa->spa_load_info nvlist.  If the remote hostname or hostid are
	 * available from configuration read from disk store them as well.
	 * This allows 'zpool import' to generate a more useful message.
	 *
	 * ZPOOL_CONFIG_MMP_STATE    - observed pool status (mandatory)
	 * ZPOOL_CONFIG_MMP_HOSTNAME - hostname from the active pool
	 * ZPOOL_CONFIG_MMP_HOSTID   - hostid from the active pool
	 */
	if (error == EREMOTEIO) {
		char *hostname = "<unknown>";
		uint64_t hostid = 0;

		if (mmp_label) {
			if (nvlist_exists(mmp_label, ZPOOL_CONFIG_HOSTNAME)) {
				hostname = fnvlist_lookup_string(mmp_label,
				    ZPOOL_CONFIG_HOSTNAME);
				fnvlist_add_string(spa->spa_load_info,
				    ZPOOL_CONFIG_MMP_HOSTNAME, hostname);
			}

			if (nvlist_exists(mmp_label, ZPOOL_CONFIG_HOSTID)) {
				hostid = fnvlist_lookup_uint64(mmp_label,
				    ZPOOL_CONFIG_HOSTID);
				fnvlist_add_uint64(spa->spa_load_info,
				    ZPOOL_CONFIG_MMP_HOSTID, hostid);
			}
		}

		fnvlist_add_uint64(spa->spa_load_info,
		    ZPOOL_CONFIG_MMP_STATE, MMP_STATE_ACTIVE);
		fnvlist_add_uint64(spa->spa_load_info,
		    ZPOOL_CONFIG_MMP_TXG, 0);

		error = spa_vdev_err(rvd, VDEV_AUX_ACTIVE, EREMOTEIO);
	}

	if (mmp_label)
		nvlist_free(mmp_label);

	return (error);
}

static int
spa_verify_host(spa_t *spa, nvlist_t *mos_config)
{
	uint64_t hostid;
	char *hostname;
	uint64_t myhostid = 0;

	if (!spa_is_root(spa) && nvlist_lookup_uint64(mos_config,
	    ZPOOL_CONFIG_HOSTID, &hostid) == 0) {
		hostname = fnvlist_lookup_string(mos_config,
		    ZPOOL_CONFIG_HOSTNAME);

		myhostid = zone_get_hostid(NULL);

		if (hostid != 0 && myhostid != 0 && hostid != myhostid) {
			cmn_err(CE_WARN, "pool '%s' could not be "
			    "loaded as it was last accessed by "
			    "another system (host: %s hostid: 0x%llx). "
			    "See: http://illumos.org/msg/ZFS-8000-EY",
			    spa_name(spa), hostname, (u_longlong_t)hostid);
			spa_load_failed(spa, "hostid verification failed: pool "
			    "last accessed by host: %s (hostid: 0x%llx)",
			    hostname, (u_longlong_t)hostid);
			return (SET_ERROR(EBADF));
		}
	}

	return (0);
}

static int
spa_ld_parse_config(spa_t *spa, spa_import_type_t type)
{
	int error = 0;
	nvlist_t *nvtree, *nvl, *config = spa->spa_config;
	int parse;
	vdev_t *rvd;
	uint64_t pool_guid;
	char *comment;

	/*
	 * 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,
	    &spa->spa_ubsync.ub_version) != 0)
		spa->spa_ubsync.ub_version = SPA_VERSION_INITIAL;

	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &pool_guid)) {
		spa_load_failed(spa, "invalid config provided: '%s' missing",
		    ZPOOL_CONFIG_POOL_GUID);
		return (SET_ERROR(EINVAL));
	}

	/*
	 * If we are doing an import, ensure that the pool is not already
	 * imported by checking if its pool guid already exists in the
	 * spa namespace.
	 *
	 * The only case that we allow an already imported pool to be
	 * imported again, is when the pool is checkpointed and we want to
	 * look at its checkpointed state from userland tools like zdb.
	 */
#ifdef _KERNEL
	if ((spa->spa_load_state == SPA_LOAD_IMPORT ||
	    spa->spa_load_state == SPA_LOAD_TRYIMPORT) &&
	    spa_guid_exists(pool_guid, 0)) {
#else
	if ((spa->spa_load_state == SPA_LOAD_IMPORT ||
	    spa->spa_load_state == SPA_LOAD_TRYIMPORT) &&
	    spa_guid_exists(pool_guid, 0) &&
	    !spa_importing_readonly_checkpoint(spa)) {
#endif
		spa_load_failed(spa, "a pool with guid %llu is already open",
		    (u_longlong_t)pool_guid);
		return (SET_ERROR(EEXIST));
	}

	spa->spa_config_guid = pool_guid;

	nvlist_free(spa->spa_load_info);
	spa->spa_load_info = fnvlist_alloc();

	ASSERT(spa->spa_comment == NULL);
	if (nvlist_lookup_string(config, ZPOOL_CONFIG_COMMENT, &comment) == 0)
		spa->spa_comment = spa_strdup(comment);

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

	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT, &nvl) == 0)
		spa->spa_config_splitting = fnvlist_dup(nvl);

	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvtree)) {
		spa_load_failed(spa, "invalid config provided: '%s' missing",
		    ZPOOL_CONFIG_VDEV_TREE);
		return (SET_ERROR(EINVAL));
	}

	/*
	 * Create "The Godfather" zio to hold all async IOs
	 */
	spa->spa_async_zio_root = kmem_alloc(max_ncpus * sizeof (void *),
	    KM_SLEEP);
	for (int i = 0; i < max_ncpus; i++) {
		spa->spa_async_zio_root[i] = zio_root(spa, NULL, NULL,
		    ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE |
		    ZIO_FLAG_GODFATHER);
	}

	/*
	 * 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, SCL_ALL, FTAG, RW_WRITER);
	parse = (type == SPA_IMPORT_EXISTING ?
	    VDEV_ALLOC_LOAD : VDEV_ALLOC_SPLIT);
	error = spa_config_parse(spa, &rvd, nvtree, NULL, 0, parse);
	spa_config_exit(spa, SCL_ALL, FTAG);

	if (error != 0) {
		spa_load_failed(spa, "unable to parse config [error=%d]",
		    error);
		return (error);
	}

	ASSERT(spa->spa_root_vdev == rvd);
	ASSERT3U(spa->spa_min_ashift, >=, SPA_MINBLOCKSHIFT);
	ASSERT3U(spa->spa_max_ashift, <=, SPA_MAXBLOCKSHIFT);

	if (type != SPA_IMPORT_ASSEMBLE) {
		ASSERT(spa_guid(spa) == pool_guid);
	}

	return (0);
}

/*
 * Recursively open all vdevs in the vdev tree. This function is called twice:
 * first with the untrusted config, then with the trusted config.
 */
static int
spa_ld_open_vdevs(spa_t *spa)
{
	int error = 0;

	/*
	 * spa_missing_tvds_allowed defines how many top-level vdevs can be
	 * missing/unopenable for the root vdev to be still considered openable.
	 */
	if (spa->spa_trust_config) {
		spa->spa_missing_tvds_allowed = zfs_max_missing_tvds;
	} else if (spa->spa_config_source == SPA_CONFIG_SRC_CACHEFILE) {
		spa->spa_missing_tvds_allowed = zfs_max_missing_tvds_cachefile;
	} else if (spa->spa_config_source == SPA_CONFIG_SRC_SCAN) {
		spa->spa_missing_tvds_allowed = zfs_max_missing_tvds_scan;
	} else {
		spa->spa_missing_tvds_allowed = 0;
	}

	spa->spa_missing_tvds_allowed =
	    MAX(zfs_max_missing_tvds, spa->spa_missing_tvds_allowed);

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	error = vdev_open(spa->spa_root_vdev);
	spa_config_exit(spa, SCL_ALL, FTAG);

	if (spa->spa_missing_tvds != 0) {
		spa_load_note(spa, "vdev tree has %lld missing top-level "
		    "vdevs.", (u_longlong_t)spa->spa_missing_tvds);
		if (spa->spa_trust_config && (spa->spa_mode & FWRITE)) {
			/*
			 * Although theoretically we could allow users to open
			 * incomplete pools in RW mode, we'd need to add a lot
			 * of extra logic (e.g. adjust pool space to account
			 * for missing vdevs).
			 * This limitation also prevents users from accidentally
			 * opening the pool in RW mode during data recovery and
			 * damaging it further.
			 */
			spa_load_note(spa, "pools with missing top-level "
			    "vdevs can only be opened in read-only mode.");
			error = SET_ERROR(ENXIO);
		} else {
			spa_load_note(spa, "current settings allow for maximum "
			    "%lld missing top-level vdevs at this stage.",
			    (u_longlong_t)spa->spa_missing_tvds_allowed);
		}
	}
	if (error != 0) {
		spa_load_failed(spa, "unable to open vdev tree [error=%d]",
		    error);
	}
	if (spa->spa_missing_tvds != 0 || error != 0)
		vdev_dbgmsg_print_tree(spa->spa_root_vdev, 2);

	return (error);
}

/*
 * We need to validate the vdev labels against the configuration that
 * we have in hand. This function is called twice: first with an untrusted
 * config, then with a trusted config. The validation is more strict when the
 * config is trusted.
 */
static int
spa_ld_validate_vdevs(spa_t *spa)
{
	int error = 0;
	vdev_t *rvd = spa->spa_root_vdev;

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	error = vdev_validate(rvd);
	spa_config_exit(spa, SCL_ALL, FTAG);

	if (error != 0) {
		spa_load_failed(spa, "vdev_validate failed [error=%d]", error);
		return (error);
	}

	if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN) {
		spa_load_failed(spa, "cannot open vdev tree after invalidating "
		    "some vdevs");
		vdev_dbgmsg_print_tree(rvd, 2);
		return (SET_ERROR(ENXIO));
	}

	return (0);
}

static void
spa_ld_select_uberblock_done(spa_t *spa, uberblock_t *ub)
{
	spa->spa_state = POOL_STATE_ACTIVE;
	spa->spa_ubsync = spa->spa_uberblock;
	spa->spa_verify_min_txg = spa->spa_extreme_rewind ?
	    TXG_INITIAL - 1 : spa_last_synced_txg(spa) - TXG_DEFER_SIZE - 1;
	spa->spa_first_txg = spa->spa_last_ubsync_txg ?
	    spa->spa_last_ubsync_txg : spa_last_synced_txg(spa) + 1;
	spa->spa_claim_max_txg = spa->spa_first_txg;
	spa->spa_prev_software_version = ub->ub_software_version;
}

static int
spa_ld_select_uberblock(spa_t *spa, spa_import_type_t type)
{
	vdev_t *rvd = spa->spa_root_vdev;
	nvlist_t *label;
	uberblock_t *ub = &spa->spa_uberblock;
	boolean_t activity_check = B_FALSE;

	/*
	 * If we are opening the checkpointed state of the pool by
	 * rewinding to it, at this point we will have written the
	 * checkpointed uberblock to the vdev labels, so searching
	 * the labels will find the right uberblock.  However, if
	 * we are opening the checkpointed state read-only, we have
	 * not modified the labels. Therefore, we must ignore the
	 * labels and continue using the spa_uberblock that was set
	 * by spa_ld_checkpoint_rewind.
	 *
	 * Note that it would be fine to ignore the labels when
	 * rewinding (opening writeable) as well. However, if we
	 * crash just after writing the labels, we will end up
	 * searching the labels. Doing so in the common case means
	 * that this code path gets exercised normally, rather than
	 * just in the edge case.
	 */
	if (ub->ub_checkpoint_txg != 0 &&
	    spa_importing_readonly_checkpoint(spa)) {
		spa_ld_select_uberblock_done(spa, ub);
		return (0);
	}

	/*
	 * Find the best uberblock.
	 */
	vdev_uberblock_load(rvd, ub, &label);

	/*
	 * If we weren't able to find a single valid uberblock, return failure.
	 */
	if (ub->ub_txg == 0) {
		nvlist_free(label);
		spa_load_failed(spa, "no valid uberblock found");
		return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, ENXIO));
	}

	if (spa->spa_load_max_txg != UINT64_MAX) {
		(void) spa_import_progress_set_max_txg(spa,
		    (u_longlong_t)spa->spa_load_max_txg);
	}
	spa_load_note(spa, "using uberblock with txg=%llu",
	    (u_longlong_t)ub->ub_txg);

	/*
	 * For pools which have the multihost property on determine if the
	 * pool is truly inactive and can be safely imported.  Prevent
	 * hosts which don't have a hostid set from importing the pool.
	 */
	activity_check = spa_activity_check_required(spa, ub, label,
	    spa->spa_config);
	if (activity_check) {
		if (ub->ub_mmp_magic == MMP_MAGIC && ub->ub_mmp_delay &&
		    spa_get_hostid() == 0) {
			nvlist_free(label);
			fnvlist_add_uint64(spa->spa_load_info,
			    ZPOOL_CONFIG_MMP_STATE, MMP_STATE_NO_HOSTID);
			return (spa_vdev_err(rvd, VDEV_AUX_ACTIVE, EREMOTEIO));
		}

		int error = spa_activity_check(spa, ub, spa->spa_config);
		if (error) {
			nvlist_free(label);
			return (error);
		}

		fnvlist_add_uint64(spa->spa_load_info,
		    ZPOOL_CONFIG_MMP_STATE, MMP_STATE_INACTIVE);
		fnvlist_add_uint64(spa->spa_load_info,
		    ZPOOL_CONFIG_MMP_TXG, ub->ub_txg);
		fnvlist_add_uint16(spa->spa_load_info,
		    ZPOOL_CONFIG_MMP_SEQ,
		    (MMP_SEQ_VALID(ub) ? MMP_SEQ(ub) : 0));
	}

	/*
	 * If the pool has an unsupported version we can't open it.
	 */
	if (!SPA_VERSION_IS_SUPPORTED(ub->ub_version)) {
		nvlist_free(label);
		spa_load_failed(spa, "version %llu is not supported",
		    (u_longlong_t)ub->ub_version);
		return (spa_vdev_err(rvd, VDEV_AUX_VERSION_NEWER, ENOTSUP));
	}

	if (ub->ub_version >= SPA_VERSION_FEATURES) {
		nvlist_t *features;

		/*
		 * If we weren't able to find what's necessary for reading the
		 * MOS in the label, return failure.
		 */
		if (label == NULL) {
			spa_load_failed(spa, "label config unavailable");
			return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA,
			    ENXIO));
		}

		if (nvlist_lookup_nvlist(label, ZPOOL_CONFIG_FEATURES_FOR_READ,
		    &features) != 0) {
			nvlist_free(label);
			spa_load_failed(spa, "invalid label: '%s' missing",
			    ZPOOL_CONFIG_FEATURES_FOR_READ);
			return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA,
			    ENXIO));
		}

		/*
		 * Update our in-core representation with the definitive values
		 * from the label.
		 */
		nvlist_free(spa->spa_label_features);
		VERIFY(nvlist_dup(features, &spa->spa_label_features, 0) == 0);
	}

	nvlist_free(label);

	/*
	 * Look through entries in the label nvlist's features_for_read. If
	 * there is a feature listed there which we don't understand then we
	 * cannot open a pool.
	 */
	if (ub->ub_version >= SPA_VERSION_FEATURES) {
		nvlist_t *unsup_feat;

		VERIFY(nvlist_alloc(&unsup_feat, NV_UNIQUE_NAME, KM_SLEEP) ==
		    0);

		for (nvpair_t *nvp = nvlist_next_nvpair(spa->spa_label_features,
		    NULL); nvp != NULL;
		    nvp = nvlist_next_nvpair(spa->spa_label_features, nvp)) {
			if (!zfeature_is_supported(nvpair_name(nvp))) {
				VERIFY(nvlist_add_string(unsup_feat,
				    nvpair_name(nvp), "") == 0);
			}
		}

		if (!nvlist_empty(unsup_feat)) {
			VERIFY(nvlist_add_nvlist(spa->spa_load_info,
			    ZPOOL_CONFIG_UNSUP_FEAT, unsup_feat) == 0);
			nvlist_free(unsup_feat);
			spa_load_failed(spa, "some features are unsupported");
			return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT,
			    ENOTSUP));
		}

		nvlist_free(unsup_feat);
	}

	if (type != SPA_IMPORT_ASSEMBLE && spa->spa_config_splitting) {
		spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
		spa_try_repair(spa, spa->spa_config);
		spa_config_exit(spa, SCL_ALL, FTAG);
		nvlist_free(spa->spa_config_splitting);
		spa->spa_config_splitting = NULL;
	}

	/*
	 * Initialize internal SPA structures.
	 */
	spa_ld_select_uberblock_done(spa, ub);

	return (0);
}

static int
spa_ld_open_rootbp(spa_t *spa)
{
	int error = 0;
	vdev_t *rvd = spa->spa_root_vdev;

	error = dsl_pool_init(spa, spa->spa_first_txg, &spa->spa_dsl_pool);
	if (error != 0) {
		spa_load_failed(spa, "unable to open rootbp in dsl_pool_init "
		    "[error=%d]", error);
		return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}
	spa->spa_meta_objset = spa->spa_dsl_pool->dp_meta_objset;

	return (0);
}

static int
spa_ld_trusted_config(spa_t *spa, spa_import_type_t type,
    boolean_t reloading)
{
	vdev_t *mrvd, *rvd = spa->spa_root_vdev;
	nvlist_t *nv, *mos_config, *policy;
	int error = 0, copy_error;
	uint64_t healthy_tvds, healthy_tvds_mos;
	uint64_t mos_config_txg;

	if (spa_dir_prop(spa, DMU_POOL_CONFIG, &spa->spa_config_object, B_TRUE)
	    != 0)
		return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));

	/*
	 * If we're assembling a pool from a split, the config provided is
	 * already trusted so there is nothing to do.
	 */
	if (type == SPA_IMPORT_ASSEMBLE)
		return (0);

	healthy_tvds = spa_healthy_core_tvds(spa);

	if (load_nvlist(spa, spa->spa_config_object, &mos_config)
	    != 0) {
		spa_load_failed(spa, "unable to retrieve MOS config");
		return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	/*
	 * If we are doing an open, pool owner wasn't verified yet, thus do
	 * the verification here.
	 */
	if (spa->spa_load_state == SPA_LOAD_OPEN) {
		error = spa_verify_host(spa, mos_config);
		if (error != 0) {
			nvlist_free(mos_config);
			return (error);
		}
	}

	nv = fnvlist_lookup_nvlist(mos_config, ZPOOL_CONFIG_VDEV_TREE);

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);

	/*
	 * Build a new vdev tree from the trusted config
	 */
	VERIFY(spa_config_parse(spa, &mrvd, nv, NULL, 0, VDEV_ALLOC_LOAD) == 0);

	/*
	 * Vdev paths in the MOS may be obsolete. If the untrusted config was
	 * obtained by scanning /dev/dsk, then it will have the right vdev
	 * paths. We update the trusted MOS config with this information.
	 * We first try to copy the paths with vdev_copy_path_strict, which
	 * succeeds only when both configs have exactly the same vdev tree.
	 * If that fails, we fall back to a more flexible method that has a
	 * best effort policy.
	 */
	copy_error = vdev_copy_path_strict(rvd, mrvd);
	if (copy_error != 0 || spa_load_print_vdev_tree) {
		spa_load_note(spa, "provided vdev tree:");
		vdev_dbgmsg_print_tree(rvd, 2);
		spa_load_note(spa, "MOS vdev tree:");
		vdev_dbgmsg_print_tree(mrvd, 2);
	}
	if (copy_error != 0) {
		spa_load_note(spa, "vdev_copy_path_strict failed, falling "
		    "back to vdev_copy_path_relaxed");
		vdev_copy_path_relaxed(rvd, mrvd);
	}

	vdev_close(rvd);
	vdev_free(rvd);
	spa->spa_root_vdev = mrvd;
	rvd = mrvd;
	spa_config_exit(spa, SCL_ALL, FTAG);

	/*
	 * We will use spa_config if we decide to reload the spa or if spa_load
	 * fails and we rewind. We must thus regenerate the config using the
	 * MOS information with the updated paths. ZPOOL_LOAD_POLICY is used to
	 * pass settings on how to load the pool and is not stored in the MOS.
	 * We copy it over to our new, trusted config.
	 */
	mos_config_txg = fnvlist_lookup_uint64(mos_config,
	    ZPOOL_CONFIG_POOL_TXG);
	nvlist_free(mos_config);
	mos_config = spa_config_generate(spa, NULL, mos_config_txg, B_FALSE);
	if (nvlist_lookup_nvlist(spa->spa_config, ZPOOL_LOAD_POLICY,
	    &policy) == 0)
		fnvlist_add_nvlist(mos_config, ZPOOL_LOAD_POLICY, policy);
	spa_config_set(spa, mos_config);
	spa->spa_config_source = SPA_CONFIG_SRC_MOS;

	/*
	 * Now that we got the config from the MOS, we should be more strict