fs/zfs/zfs_ioctl.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 2007 Sun Microsystems, Inc.  All rights reserved.
 * Use is subject to license terms.
 */

#pragma ident	"@(#)zfs_ioctl.c	1.56	07/12/09 SMI"

#include <sys/types.h>
#include <sys/param.h>
#include <sys/errno.h>
#include <sys/uio.h>
#include <sys/buf.h>
#include <sys/modctl.h>
#include <sys/open.h>
#include <sys/file.h>
#include <sys/kmem.h>
#include <sys/conf.h>
#include <sys/cmn_err.h>
#include <sys/stat.h>
#include <sys/zfs_ioctl.h>
#include <sys/zfs_znode.h>
#include <sys/zap.h>
#include <sys/spa.h>
#include <sys/spa_impl.h>
#include <sys/vdev.h>
#include <sys/vdev_impl.h>
#include <sys/dmu.h>
#include <sys/dsl_dir.h>
#include <sys/dsl_dataset.h>
#include <sys/dsl_prop.h>
#include <sys/dsl_deleg.h>
#include <sys/dmu_objset.h>
#include <sys/ddi.h>
#include <sys/sunddi.h>
#include <sys/sunldi.h>
#include <sys/policy.h>
#include <sys/zone.h>
#include <sys/nvpair.h>
#include <sys/pathname.h>
#include <sys/mount.h>
#include <sys/sdt.h>
#include <sys/fs/zfs.h>
#include <sys/zfs_ctldir.h>
#include <sys/zfs_dir.h>
#include <sys/zvol.h>
#include <sharefs/share.h>
#include <sys/dmu_objset.h>

#include "zfs_namecheck.h"
#include "zfs_prop.h"
#include "zfs_deleg.h"

extern struct modlfs zfs_modlfs;

extern void zfs_init(void);
extern void zfs_fini(void);

ldi_ident_t zfs_li = NULL;
dev_info_t *zfs_dip;

typedef int zfs_ioc_func_t(zfs_cmd_t *);
typedef int zfs_secpolicy_func_t(zfs_cmd_t *, cred_t *);

typedef struct zfs_ioc_vec {
	zfs_ioc_func_t		*zvec_func;
	zfs_secpolicy_func_t	*zvec_secpolicy;
	enum {
		NO_NAME,
		POOL_NAME,
		DATASET_NAME
	} zvec_namecheck;
	boolean_t		zvec_his_log;
} zfs_ioc_vec_t;

/* _NOTE(PRINTFLIKE(4)) - this is printf-like, but lint is too whiney */
void
__dprintf(const char *file, const char *func, int line, const char *fmt, ...)
{
	const char *newfile;
	char buf[256];
	va_list adx;

	/*
	 * Get rid of annoying "../common/" prefix to filename.
	 */
	newfile = strrchr(file, '/');
	if (newfile != NULL) {
		newfile = newfile + 1; /* Get rid of leading / */
	} else {
		newfile = file;
	}

	va_start(adx, fmt);
	(void) vsnprintf(buf, sizeof (buf), fmt, adx);
	va_end(adx);

	/*
	 * To get this data, use the zfs-dprintf probe as so:
	 * dtrace -q -n 'zfs-dprintf \
	 *	/stringof(arg0) == "dbuf.c"/ \
	 *	{printf("%s: %s", stringof(arg1), stringof(arg3))}'
	 * arg0 = file name
	 * arg1 = function name
	 * arg2 = line number
	 * arg3 = message
	 */
	DTRACE_PROBE4(zfs__dprintf,
	    char *, newfile, char *, func, int, line, char *, buf);
}

static void
history_str_free(char *buf)
{
	kmem_free(buf, HIS_MAX_RECORD_LEN);
}

static char *
history_str_get(zfs_cmd_t *zc)
{
	char *buf;

	if (zc->zc_history == NULL)
		return (NULL);

	buf = kmem_alloc(HIS_MAX_RECORD_LEN, KM_SLEEP);
	if (copyinstr((void *)(uintptr_t)zc->zc_history,
	    buf, HIS_MAX_RECORD_LEN, NULL) != 0) {
		history_str_free(buf);
		return (NULL);
	}

	buf[HIS_MAX_RECORD_LEN -1] = '\0';

	return (buf);
}

/*
 * zfs_check_version
 *
 *	Return non-zero if the spa version is less than requested version.
 */
static int
zfs_check_version(const char *name, int version)
{

	spa_t *spa;

	if (spa_open(name, &spa, FTAG) == 0) {
		if (spa_version(spa) < version) {
			spa_close(spa, FTAG);
			return (1);
		}
		spa_close(spa, FTAG);
	}
	return (0);
}

static void
zfs_log_history(zfs_cmd_t *zc)
{
	spa_t *spa;
	char *buf;

	if ((buf = history_str_get(zc)) == NULL)
		return;

	if (spa_open(zc->zc_name, &spa, FTAG) == 0) {
		if (spa_version(spa) >= SPA_VERSION_ZPOOL_HISTORY)
			(void) spa_history_log(spa, buf, LOG_CMD_NORMAL);
		spa_close(spa, FTAG);
	}
	history_str_free(buf);
}

/*
 * Policy for top-level read operations (list pools).  Requires no privileges,
 * and can be used in the local zone, as there is no associated dataset.
 */
/* ARGSUSED */
static int
zfs_secpolicy_none(zfs_cmd_t *zc, cred_t *cr)
{
	return (0);
}

/*
 * Policy for dataset read operations (list children, get statistics).  Requires
 * no privileges, but must be visible in the local zone.
 */
/* ARGSUSED */
static int
zfs_secpolicy_read(zfs_cmd_t *zc, cred_t *cr)
{
	if (INGLOBALZONE(curproc) ||
	    zone_dataset_visible(zc->zc_name, NULL))
		return (0);

	return (ENOENT);
}

static int
zfs_dozonecheck(const char *dataset, cred_t *cr)
{
	uint64_t zoned;
	int writable = 1;

	/*
	 * The dataset must be visible by this zone -- check this first
	 * so they don't see EPERM on something they shouldn't know about.
	 */
	if (!INGLOBALZONE(curproc) &&
	    !zone_dataset_visible(dataset, &writable))
		return (ENOENT);

	if (dsl_prop_get_integer(dataset, "zoned", &zoned, NULL))
		return (ENOENT);

	if (INGLOBALZONE(curproc)) {
		/*
		 * If the fs is zoned, only root can access it from the
		 * global zone.
		 */
		if (secpolicy_zfs(cr) && zoned)
			return (EPERM);
	} else {
		/*
		 * If we are in a local zone, the 'zoned' property must be set.
		 */
		if (!zoned)
			return (EPERM);

		/* must be writable by this zone */
		if (!writable)
			return (EPERM);
	}
	return (0);
}

int
zfs_secpolicy_write_perms(const char *name, const char *perm, cred_t *cr)
{
	int error;

	error = zfs_dozonecheck(name, cr);
	if (error == 0) {
		error = secpolicy_zfs(cr);
		if (error)
			error = dsl_deleg_access(name, perm, cr);
	}
	return (error);
}

static int
zfs_secpolicy_setprop(const char *name, zfs_prop_t prop, cred_t *cr)
{
	/*
	 * Check permissions for special properties.
	 */
	switch (prop) {
	case ZFS_PROP_ZONED:
		/*
		 * Disallow setting of 'zoned' from within a local zone.
		 */
		if (!INGLOBALZONE(curproc))
			return (EPERM);
		break;

	case ZFS_PROP_QUOTA:
		if (!INGLOBALZONE(curproc)) {
			uint64_t zoned;
			char setpoint[MAXNAMELEN];
			/*
			 * Unprivileged users are allowed to modify the
			 * quota on things *under* (ie. contained by)
			 * the thing they own.
			 */
			if (dsl_prop_get_integer(name, "zoned", &zoned,
			    setpoint))
				return (EPERM);
			if (!zoned || strlen(name) <= strlen(setpoint))
				return (EPERM);
		}
		break;
	}

	return (zfs_secpolicy_write_perms(name, zfs_prop_to_name(prop), cr));
}

int
zfs_secpolicy_fsacl(zfs_cmd_t *zc, cred_t *cr)
{
	int error;

	error = zfs_dozonecheck(zc->zc_name, cr);
	if (error)
		return (error);

	/*
	 * permission to set permissions will be evaluated later in
	 * dsl_deleg_can_allow()
	 */
	return (0);
}

int
zfs_secpolicy_rollback(zfs_cmd_t *zc, cred_t *cr)
{
	int error;
	error = zfs_secpolicy_write_perms(zc->zc_name,
	    ZFS_DELEG_PERM_ROLLBACK, cr);
	if (error == 0)
		error = zfs_secpolicy_write_perms(zc->zc_name,
		    ZFS_DELEG_PERM_MOUNT, cr);
	return (error);
}

int
zfs_secpolicy_send(zfs_cmd_t *zc, cred_t *cr)
{
	return (zfs_secpolicy_write_perms(zc->zc_name,
	    ZFS_DELEG_PERM_SEND, cr));
}

int
zfs_secpolicy_share(zfs_cmd_t *zc, cred_t *cr)
{
	if (!INGLOBALZONE(curproc))
		return (EPERM);

	if (secpolicy_nfs(cr) == 0) {
		return (0);
	} else {
		vnode_t *vp;
		int error;

		if ((error = lookupname(zc->zc_value, UIO_SYSSPACE,
		    NO_FOLLOW, NULL, &vp)) != 0)
			return (error);

		/* Now make sure mntpnt and dataset are ZFS */

		if (vp->v_vfsp->vfs_fstype != zfsfstype ||
		    (strcmp((char *)refstr_value(vp->v_vfsp->vfs_resource),
		    zc->zc_name) != 0)) {
			VN_RELE(vp);
			return (EPERM);
		}

		VN_RELE(vp);
		return (dsl_deleg_access(zc->zc_name,
		    ZFS_DELEG_PERM_SHARE, cr));
	}
}

static int
zfs_get_parent(const char *datasetname, char *parent, int parentsize)
{
	char *cp;

	/*
	 * Remove the @bla or /bla from the end of the name to get the parent.
	 */
	(void) strncpy(parent, datasetname, parentsize);
	cp = strrchr(parent, '@');
	if (cp != NULL) {
		cp[0] = '\0';
	} else {
		cp = strrchr(parent, '/');
		if (cp == NULL)
			return (ENOENT);
		cp[0] = '\0';
	}

	return (0);
}

int
zfs_secpolicy_destroy_perms(const char *name, cred_t *cr)
{
	int error;

	if ((error = zfs_secpolicy_write_perms(name,
	    ZFS_DELEG_PERM_MOUNT, cr)) != 0)
		return (error);

	return (zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_DESTROY, cr));
}

static int
zfs_secpolicy_destroy(zfs_cmd_t *zc, cred_t *cr)
{
	return (zfs_secpolicy_destroy_perms(zc->zc_name, cr));
}

/*
 * Must have sys_config privilege to check the iscsi permission
 */
/* ARGSUSED */
static int
zfs_secpolicy_iscsi(zfs_cmd_t *zc, cred_t *cr)
{
	return (secpolicy_zfs(cr));
}

int
zfs_secpolicy_rename_perms(const char *from, const char *to, cred_t *cr)
{
	char 	parentname[MAXNAMELEN];
	int	error;

	if ((error = zfs_secpolicy_write_perms(from,
	    ZFS_DELEG_PERM_RENAME, cr)) != 0)
		return (error);

	if ((error = zfs_secpolicy_write_perms(from,
	    ZFS_DELEG_PERM_MOUNT, cr)) != 0)
		return (error);

	if ((error = zfs_get_parent(to, parentname,
	    sizeof (parentname))) != 0)
		return (error);

	if ((error = zfs_secpolicy_write_perms(parentname,
	    ZFS_DELEG_PERM_CREATE, cr)) != 0)
		return (error);

	if ((error = zfs_secpolicy_write_perms(parentname,
	    ZFS_DELEG_PERM_MOUNT, cr)) != 0)
		return (error);

	return (error);
}

static int
zfs_secpolicy_rename(zfs_cmd_t *zc, cred_t *cr)
{
	return (zfs_secpolicy_rename_perms(zc->zc_name, zc->zc_value, cr));
}

static int
zfs_secpolicy_promote(zfs_cmd_t *zc, cred_t *cr)
{
	char 	parentname[MAXNAMELEN];
	objset_t *clone;
	int error;

	error = zfs_secpolicy_write_perms(zc->zc_name,
	    ZFS_DELEG_PERM_PROMOTE, cr);
	if (error)
		return (error);

	error = dmu_objset_open(zc->zc_name, DMU_OST_ANY,
	    DS_MODE_STANDARD | DS_MODE_READONLY, &clone);

	if (error == 0) {
		dsl_dataset_t *pclone = NULL;
		dsl_dir_t *dd;
		dd = clone->os->os_dsl_dataset->ds_dir;

		rw_enter(&dd->dd_pool->dp_config_rwlock, RW_READER);
		error = dsl_dataset_open_obj(dd->dd_pool,
		    dd->dd_phys->dd_origin_obj, NULL,
		    DS_MODE_NONE, FTAG, &pclone);
		rw_exit(&dd->dd_pool->dp_config_rwlock);
		if (error) {
			dmu_objset_close(clone);
			return (error);
		}

		error = zfs_secpolicy_write_perms(zc->zc_name,
		    ZFS_DELEG_PERM_MOUNT, cr);

		dsl_dataset_name(pclone, parentname);
		dmu_objset_close(clone);
		dsl_dataset_close(pclone, DS_MODE_NONE, FTAG);
		if (error == 0)
			error = zfs_secpolicy_write_perms(parentname,
			    ZFS_DELEG_PERM_PROMOTE, cr);
	}
	return (error);
}

static int
zfs_secpolicy_receive(zfs_cmd_t *zc, cred_t *cr)
{
	int error;

	if ((error = zfs_secpolicy_write_perms(zc->zc_name,
	    ZFS_DELEG_PERM_RECEIVE, cr)) != 0)
		return (error);

	if ((error = zfs_secpolicy_write_perms(zc->zc_name,
	    ZFS_DELEG_PERM_MOUNT, cr)) != 0)
		return (error);

	return (zfs_secpolicy_write_perms(zc->zc_name,
	    ZFS_DELEG_PERM_CREATE, cr));
}

int
zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr)
{
	int error;

	if ((error = zfs_secpolicy_write_perms(name,
	    ZFS_DELEG_PERM_SNAPSHOT, cr)) != 0)
		return (error);

	error = zfs_secpolicy_write_perms(name,
	    ZFS_DELEG_PERM_MOUNT, cr);

	return (error);
}

static int
zfs_secpolicy_snapshot(zfs_cmd_t *zc, cred_t *cr)
{

	return (zfs_secpolicy_snapshot_perms(zc->zc_name, cr));
}

static int
zfs_secpolicy_create(zfs_cmd_t *zc, cred_t *cr)
{
	char 	parentname[MAXNAMELEN];
	int 	error;

	if ((error = zfs_get_parent(zc->zc_name, parentname,
	    sizeof (parentname))) != 0)
		return (error);

	if (zc->zc_value[0] != '\0') {
		if ((error = zfs_secpolicy_write_perms(zc->zc_value,
		    ZFS_DELEG_PERM_CLONE, cr)) != 0)
			return (error);
	}

	if ((error = zfs_secpolicy_write_perms(parentname,
	    ZFS_DELEG_PERM_CREATE, cr)) != 0)
		return (error);

	error = zfs_secpolicy_write_perms(parentname,
	    ZFS_DELEG_PERM_MOUNT, cr);

	return (error);
}

static int
zfs_secpolicy_umount(zfs_cmd_t *zc, cred_t *cr)
{
	int error;

	error = secpolicy_fs_unmount(cr, NULL);
	if (error) {
		error = dsl_deleg_access(zc->zc_name, ZFS_DELEG_PERM_MOUNT, cr);
	}
	return (error);
}

/*
 * Policy for pool operations - create/destroy pools, add vdevs, etc.  Requires
 * SYS_CONFIG privilege, which is not available in a local zone.
 */
/* ARGSUSED */
static int
zfs_secpolicy_config(zfs_cmd_t *zc, cred_t *cr)
{
	if (secpolicy_sys_config(cr, B_FALSE) != 0)
		return (EPERM);

	return (0);
}

/*
 * Just like zfs_secpolicy_config, except that we will check for
 * mount permission on the dataset for permission to create/remove
 * the minor nodes.
 */
static int
zfs_secpolicy_minor(zfs_cmd_t *zc, cred_t *cr)
{
	if (secpolicy_sys_config(cr, B_FALSE) != 0) {
		return (dsl_deleg_access(zc->zc_name,
		    ZFS_DELEG_PERM_MOUNT, cr));
	}

	return (0);
}

/*
 * Policy for fault injection.  Requires all privileges.
 */
/* ARGSUSED */
static int
zfs_secpolicy_inject(zfs_cmd_t *zc, cred_t *cr)
{
	return (secpolicy_zinject(cr));
}

static int
zfs_secpolicy_inherit(zfs_cmd_t *zc, cred_t *cr)
{
	zfs_prop_t prop = zfs_name_to_prop(zc->zc_value);

	if (prop == ZPROP_INVAL) {
		if (!zfs_prop_user(zc->zc_value))
			return (EINVAL);
		return (zfs_secpolicy_write_perms(zc->zc_name,
		    ZFS_DELEG_PERM_USERPROP, cr));
	} else {
		if (!zfs_prop_inheritable(prop))
			return (EINVAL);
		return (zfs_secpolicy_setprop(zc->zc_name, prop, cr));
	}
}

/*
 * Returns the nvlist as specified by the user in the zfs_cmd_t.
 */
static int
get_nvlist(uint64_t nvl, uint64_t size, nvlist_t **nvp)
{
	char *packed;
	int error;
	nvlist_t *list = NULL;

	/*
	 * Read in and unpack the user-supplied nvlist.
	 */
	if (size == 0)
		return (EINVAL);

	packed = kmem_alloc(size, KM_SLEEP);

	if ((error = xcopyin((void *)(uintptr_t)nvl, packed, size)) != 0) {
		kmem_free(packed, size);
		return (error);
	}

	if ((error = nvlist_unpack(packed, size, &list, 0)) != 0) {
		kmem_free(packed, size);
		return (error);
	}

	kmem_free(packed, size);

	*nvp = list;
	return (0);
}

static int
put_nvlist(zfs_cmd_t *zc, nvlist_t *nvl)
{
	char *packed = NULL;
	size_t size;
	int error;

	VERIFY(nvlist_size(nvl, &size, NV_ENCODE_NATIVE) == 0);

	if (size > zc->zc_nvlist_dst_size) {
		error = ENOMEM;
	} else {
		packed = kmem_alloc(size, KM_SLEEP);
		VERIFY(nvlist_pack(nvl, &packed, &size, NV_ENCODE_NATIVE,
		    KM_SLEEP) == 0);
		error = xcopyout(packed, (void *)(uintptr_t)zc->zc_nvlist_dst,
		    size);
		kmem_free(packed, size);
	}

	zc->zc_nvlist_dst_size = size;
	return (error);
}

static int
zfs_ioc_pool_create(zfs_cmd_t *zc)
{
	int error;
	nvlist_t *config, *props = NULL;
	char *buf;

	if (error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
	    &config))
		return (error);

	if (zc->zc_nvlist_src_size != 0 && (error =
	    get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, &props))) {
		nvlist_free(config);
		return (error);
	}

	buf = history_str_get(zc);

	error = spa_create(zc->zc_name, config, props, buf);

	if (buf != NULL)
		history_str_free(buf);

	nvlist_free(config);

	if (props)
		nvlist_free(props);

	return (error);
}

static int
zfs_ioc_pool_destroy(zfs_cmd_t *zc)
{
	int error;
	zfs_log_history(zc);
	error = spa_destroy(zc->zc_name);
	return (error);
}

static int
zfs_ioc_pool_import(zfs_cmd_t *zc)
{
	int error;
	nvlist_t *config, *props = NULL;
	uint64_t guid;

	if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
	    &config)) != 0)
		return (error);

	if (zc->zc_nvlist_src_size != 0 && (error =
	    get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, &props))) {
		nvlist_free(config);
		return (error);
	}

	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &guid) != 0 ||
	    guid != zc->zc_guid)
		error = EINVAL;
	else
		error = spa_import(zc->zc_name, config, props);

	nvlist_free(config);

	if (props)
		nvlist_free(props);

	return (error);
}

static int
zfs_ioc_pool_export(zfs_cmd_t *zc)
{
	int error;
	zfs_log_history(zc);
	error = spa_export(zc->zc_name, NULL);
	return (error);
}

static int
zfs_ioc_pool_configs(zfs_cmd_t *zc)
{
	nvlist_t *configs;
	int error;

	if ((configs = spa_all_configs(&zc->zc_cookie)) == NULL)
		return (EEXIST);

	error = put_nvlist(zc, configs);

	nvlist_free(configs);

	return (error);
}

static int
zfs_ioc_pool_stats(zfs_cmd_t *zc)
{
	nvlist_t *config;
	int error;
	int ret = 0;

	error = spa_get_stats(zc->zc_name, &config, zc->zc_value,
	    sizeof (zc->zc_value));

	if (config != NULL) {
		ret = put_nvlist(zc, config);
		nvlist_free(config);

		/*
		 * The config may be present even if 'error' is non-zero.
		 * In this case we return success, and preserve the real errno
		 * in 'zc_cookie'.
		 */
		zc->zc_cookie = error;
	} else {
		ret = error;
	}

	return (ret);
}

/*
 * Try to import the given pool, returning pool stats as appropriate so that
 * user land knows which devices are available and overall pool health.
 */
static int
zfs_ioc_pool_tryimport(zfs_cmd_t *zc)
{
	nvlist_t *tryconfig, *config;
	int error;

	if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
	    &tryconfig)) != 0)
		return (error);

	config = spa_tryimport(tryconfig);

	nvlist_free(tryconfig);

	if (config == NULL)
		return (EINVAL);

	error = put_nvlist(zc, config);
	nvlist_free(config);

	return (error);
}

static int
zfs_ioc_pool_scrub(zfs_cmd_t *zc)
{
	spa_t *spa;
	int error;

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
		return (error);

	mutex_enter(&spa_namespace_lock);
	error = spa_scrub(spa, zc->zc_cookie, B_FALSE);
	mutex_exit(&spa_namespace_lock);

	spa_close(spa, FTAG);

	return (error);
}

static int
zfs_ioc_pool_freeze(zfs_cmd_t *zc)
{
	spa_t *spa;
	int error;

	error = spa_open(zc->zc_name, &spa, FTAG);
	if (error == 0) {
		spa_freeze(spa);
		spa_close(spa, FTAG);
	}
	return (error);
}

static int
zfs_ioc_pool_upgrade(zfs_cmd_t *zc)
{
	spa_t *spa;
	int error;

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
		return (error);

	if (zc->zc_cookie < spa_version(spa) || zc->zc_cookie > SPA_VERSION) {
		spa_close(spa, FTAG);
		return (EINVAL);
	}

	spa_upgrade(spa, zc->zc_cookie);
	spa_close(spa, FTAG);

	return (error);
}

static int
zfs_ioc_pool_get_history(zfs_cmd_t *zc)
{
	spa_t *spa;
	char *hist_buf;
	uint64_t size;
	int error;

	if ((size = zc->zc_history_len) == 0)
		return (EINVAL);

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
		return (error);

	if (spa_version(spa) < SPA_VERSION_ZPOOL_HISTORY) {
		spa_close(spa, FTAG);
		return (ENOTSUP);
	}

	hist_buf = kmem_alloc(size, KM_SLEEP);
	if ((error = spa_history_get(spa, &zc->zc_history_offset,
	    &zc->zc_history_len, hist_buf)) == 0) {
		error = xcopyout(hist_buf,
		    (char *)(uintptr_t)zc->zc_history,
		    zc->zc_history_len);
	}

	spa_close(spa, FTAG);
	kmem_free(hist_buf, size);
	return (error);
}

static int
zfs_ioc_dsobj_to_dsname(zfs_cmd_t *zc)
{
	int error;

	if (error = dsl_dsobj_to_dsname(zc->zc_name, zc->zc_obj, zc->zc_value))
		return (error);

	return (0);
}

static int
zfs_ioc_obj_to_path(zfs_cmd_t *zc)
{
	objset_t *osp;
	int error;

	if ((error = dmu_objset_open(zc->zc_name, DMU_OST_ZFS,
	    DS_MODE_NONE | DS_MODE_READONLY, &osp)) != 0)
		return (error);

	error = zfs_obj_to_path(osp, zc->zc_obj, zc->zc_value,
	    sizeof (zc->zc_value));
	dmu_objset_close(osp);

	return (error);
}

static int
zfs_ioc_vdev_add(zfs_cmd_t *zc)
{
	spa_t *spa;
	int error;
	nvlist_t *config, **l2cache;
	uint_t nl2cache;

	error = spa_open(zc->zc_name, &spa, FTAG);
	if (error != 0)
		return (error);

	error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
	    &config);
	(void) nvlist_lookup_nvlist_array(config, ZPOOL_CONFIG_L2CACHE,
	    &l2cache, &nl2cache);

	/*
	 * A root pool with concatenated devices is not supported.
	 * Thus, can not add a device to a root pool with one device.
	 * Allow for l2cache devices to be added.
	 */
	if (spa->spa_root_vdev->vdev_children == 1 && spa->spa_bootfs != 0 &&
	    nl2cache == 0) {
		spa_close(spa, FTAG);
		return (EDOM);
	}

	if (error == 0) {
		error = spa_vdev_add(spa, config);
		nvlist_free(config);
	}
	spa_close(spa, FTAG);
	return (error);
}

static int
zfs_ioc_vdev_remove(zfs_cmd_t *zc)
{
	spa_t *spa;
	int error;

	error = spa_open(zc->zc_name, &spa, FTAG);
	if (error != 0)
		return (error);
	error = spa_vdev_remove(spa, zc->zc_guid, B_FALSE);
	spa_close(spa, FTAG);
	return (error);
}

static int
zfs_ioc_vdev_set_state(zfs_cmd_t *zc)
{
	spa_t *spa;
	int error;
	vdev_state_t newstate = VDEV_STATE_UNKNOWN;

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
		return (error);
	switch (zc->zc_cookie) {
	case VDEV_STATE_ONLINE:
		error = vdev_online(spa, zc->zc_guid, zc->zc_obj, &newstate);
		break;

	case VDEV_STATE_OFFLINE:
		error = vdev_offline(spa, zc->zc_guid, zc->zc_obj);
		break;

	case VDEV_STATE_FAULTED:
		error = vdev_fault(spa, zc->zc_guid);
		break;

	case VDEV_STATE_DEGRADED:
		error = vdev_degrade(spa, zc->zc_guid);
		break;

	default:
		error = EINVAL;
	}
	zc->zc_cookie = newstate;
	spa_close(spa, FTAG);
	return (error);
}

static int
zfs_ioc_vdev_attach(zfs_cmd_t *zc)
{
	spa_t *spa;
	int replacing = zc->zc_cookie;
	nvlist_t *config;
	int error;

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
		return (error);

	if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
	    &config)) == 0) {
		error = spa_vdev_attach(spa, zc->zc_guid, config, replacing);
		nvlist_free(config);
	}

	spa_close(spa, FTAG);
	return (error);
}

static int
zfs_ioc_vdev_detach(zfs_cmd_t *zc)
{
	spa_t *spa;
	int error;

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
		return (error);

	error = spa_vdev_detach(spa, zc->zc_guid, B_FALSE);

	spa_close(spa, FTAG);
	return (error);
}

static int
zfs_ioc_vdev_setpath(zfs_cmd_t *zc)
{
	spa_t *spa;
	char *path = zc->zc_value;
	uint64_t guid = zc->zc_guid;
	int error;

	error = spa_open(zc->zc_name, &spa, FTAG);
	if (error != 0)
		return (error);

	error = spa_vdev_setpath(spa, guid, path);
	spa_close(spa, FTAG);
	return (error);
}

static int
zfs_os_open_retry(char *name, objset_t **os)
{
	int error;

retry:
	error = dmu_objset_open(name, DMU_OST_ANY,
	    DS_MODE_STANDARD | DS_MODE_READONLY, os);
	if (error != 0) {
		/*
		 * This is ugly: dmu_objset_open() can return EBUSY if
		 * the objset is held exclusively. Fortunately this hold is
		 * only for a short while, so we retry here.
		 * This avoids user code having to handle EBUSY,
		 * for example for a "zfs list".
		 */
		if (error == EBUSY) {
			delay(1);
			goto retry;
		}
	}
	return (error);
}

/*
 * inputs:
 * zc_name		name of filesystem
 * zc_nvlist_dst_size	size of buffer for property nvlist
 *
 * outputs:
 * zc_objset_stats	stats
 * zc_nvlist_dst	property nvlist
 * zc_nvlist_dst_size	size of property nvlist
 * zc_value		alternate root
 */
static int
zfs_ioc_objset_stats(zfs_cmd_t *zc)
{
	objset_t *os = NULL;
	int error