xref: /illumos-gate/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 2009 Sun Microsystems, Inc.  All rights reserved.
 * Use is subject to license terms.
 */

/*
 * 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 <sys/spa_boot.h>
#include <sys/zfs_ioctl.h>

#ifdef	_KERNEL
#include <sys/zone.h>
#endif	/* _KERNEL */

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

enum zti_modes {
	zti_mode_fixed,			/* value is # of threads (min 1) */
	zti_mode_online_percent,	/* value is % of online CPUs */
	zti_mode_tune,			/* fill from zio_taskq_tune_* */
	zti_nmodes
};

#define	ZTI_THREAD_FIX(n)	{ zti_mode_fixed, (n) }
#define	ZTI_THREAD_PCT(n)	{ zti_mode_online_percent, (n) }
#define	ZTI_THREAD_TUNE		{ zti_mode_tune, 0 }

#define	ZTI_THREAD_ONE		ZTI_THREAD_FIX(1)

typedef struct zio_taskq_info {
	const char *zti_name;
	struct {
		enum zti_modes zti_mode;
		uint_t zti_value;
	} zti_nthreads[ZIO_TASKQ_TYPES];
} zio_taskq_info_t;

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

const zio_taskq_info_t zio_taskqs[ZIO_TYPES] = {
	/*			ISSUE			INTR		*/
	{ "spa_zio_null",	{ ZTI_THREAD_ONE,	ZTI_THREAD_ONE } },
	{ "spa_zio_read",	{ ZTI_THREAD_FIX(8),	ZTI_THREAD_TUNE } },
	{ "spa_zio_write",	{ ZTI_THREAD_TUNE,	ZTI_THREAD_FIX(8) } },
	{ "spa_zio_free",	{ ZTI_THREAD_ONE,	ZTI_THREAD_ONE } },
	{ "spa_zio_claim",	{ ZTI_THREAD_ONE,	ZTI_THREAD_ONE } },
	{ "spa_zio_ioctl",	{ ZTI_THREAD_ONE,	ZTI_THREAD_ONE } },
};

enum zti_modes zio_taskq_tune_mode = zti_mode_online_percent;
uint_t zio_taskq_tune_value = 80;	/* #threads = 80% of # online CPUs */

static void spa_sync_props(void *arg1, void *arg2, cred_t *cr, dmu_tx_t *tx);
static boolean_t spa_has_active_shared_spare(spa_t *spa);

/*
 * ==========================================================================
 * 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)
{
	uint64_t size;
	uint64_t used;
	uint64_t cap, version;
	zprop_source_t src = ZPROP_SRC_NONE;
	spa_config_dirent_t *dp;

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

	if (spa->spa_root_vdev != NULL) {
		size = spa_get_space(spa);
		used = spa_get_alloc(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_USED, NULL, used, src);
		spa_prop_add_list(*nvp, ZPOOL_PROP_AVAILABLE, NULL,
		    size - used, src);

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

		spa_prop_add_list(*nvp, ZPOOL_PROP_HEALTH, NULL,
		    spa->spa_root_vdev->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);
	}

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

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

	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)
{
	zap_cursor_t zc;
	zap_attribute_t za;
	objset_t *mos = spa->spa_meta_objset;
	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 (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_hold_obj(dp,
				    za.za_first_integer, 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_rele(ds, FTAG);
				rw_exit(&dp->dp_config_rwlock);
			} else {
				strval = NULL;
				intval = za.za_first_integer;
			}

			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;
			}
			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;

	elem = NULL;
	while ((elem = nvlist_next_nvpair(props, elem)) != NULL) {
		zpool_prop_t prop;
		char *propname, *strval;
		uint64_t intval;
		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:
		case ZPOOL_PROP_LISTSNAPS:
		case ZPOOL_PROP_AUTOEXPAND:
			error = nvpair_value_uint64(elem, &intval);
			if (!error && intval > 1)
				error = EINVAL;
			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 = ENOTSUP;
				break;
			}

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

			reset_bootfs = 1;

			error = nvpair_value_string(elem, &strval);

			if (!error) {
				uint64_t compress;

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

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

				/* Must be ZPL and not gzip compressed. */

				if (dmu_objset_type(os) != DMU_OST_ZFS) {
					error = ENOTSUP;
				} else if ((error = dsl_prop_get_integer(strval,
				    zfs_prop_to_name(ZFS_PROP_COMPRESSION),
				    &compress, NULL)) == 0 &&
				    !BOOTFS_COMPRESS_VALID(compress)) {
					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 = 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 = 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);
}

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;
	boolean_t need_sync = B_FALSE;
	zpool_prop_t prop;

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

	elem = NULL;
	while ((elem = nvlist_next_nvpair(nvp, elem)) != NULL) {
		if ((prop = zpool_name_to_prop(
		    nvpair_name(elem))) == ZPROP_INVAL)
			return (EINVAL);

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

		need_sync = B_TRUE;
		break;
	}

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

/*
 * ==========================================================================
 * 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 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(zfs_metaslab_ops);
	spa->spa_log_class = metaslab_class_create(zfs_metaslab_ops);

	for (int t = 0; t < ZIO_TYPES; t++) {
		const zio_taskq_info_t *ztip = &zio_taskqs[t];
		for (int q = 0; q < ZIO_TASKQ_TYPES; q++) {
			enum zti_modes mode = ztip->zti_nthreads[q].zti_mode;
			uint_t value = ztip->zti_nthreads[q].zti_value;
			char name[32];

			(void) snprintf(name, sizeof (name),
			    "%s_%s", ztip->zti_name, zio_taskq_types[q]);

			if (mode == zti_mode_tune) {
				mode = zio_taskq_tune_mode;
				value = zio_taskq_tune_value;
				if (mode == zti_mode_tune)
					mode = zti_mode_online_percent;
			}

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

				spa->spa_zio_taskq[t][q] = taskq_create(name,
				    value, maxclsyspri, 50, INT_MAX,
				    TASKQ_PREPOPULATE);
				break;

			case zti_mode_online_percent:
				spa->spa_zio_taskq[t][q] = taskq_create(name,
				    value, maxclsyspri, 50, INT_MAX,
				    TASKQ_PREPOPULATE | TASKQ_THREADS_CPU_PCT);
				break;

			case zti_mode_tune:
			default:
				panic("unrecognized mode for "
				    "zio_taskqs[%u]->zti_nthreads[%u] (%u:%u) "
				    "in spa_activate()",
				    t, q, mode, value);
				break;
			}
		}
	}

	list_create(&spa->spa_config_dirty_list, sizeof (vdev_t),
	    offsetof(vdev_t, vdev_config_dirty_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,
	    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)
{
	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);

	txg_list_destroy(&spa->spa_vdev_txg_list);

	list_destroy(&spa->spa_config_dirty_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++) {
			taskq_destroy(spa->spa_zio_taskq[t][q]);
			spa->spa_zio_taskq[t][q] = 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