xref: /illumos-gate/usr/src/lib/libbe/common/be_create.c (revision f976337a)

/*
 * 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) 2008, 2010, Oracle and/or its affiliates. All rights reserved.
 * Copyright 2013 Nexenta Systems, Inc. All rights reserved.
 * Copyright (c) 2014, 2015 by Delphix. All rights reserved.
 * Copyright (c) 2016 Martin Matuska. All rights reserved.
 * Copyright 2021 OmniOS Community Edition (OmniOSce) Association.
 */

/*
 * System includes
 */

#include <assert.h>
#include <ctype.h>
#include <errno.h>
#include <libgen.h>
#include <libintl.h>
#include <libnvpair.h>
#include <libzfs.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/mnttab.h>
#include <sys/mount.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/wait.h>
#include <unistd.h>

#include <libbe.h>
#include <libbe_priv.h>
#include <libzfsbootenv.h>

/* Library wide variables */
libzfs_handle_t *g_zfs = NULL;

/* Private function prototypes */
static int _be_destroy(const char *, be_destroy_data_t *);
static int be_destroy_zones(char *, char *, be_destroy_data_t *);
static int be_destroy_zone_roots(char *, be_destroy_data_t *);
static int be_destroy_zone_roots_callback(zfs_handle_t *, void *);
static int be_copy_zones(char *, char *, char *);
static int be_clone_fs_callback(zfs_handle_t *, void *);
static int be_destroy_callback(zfs_handle_t *, void *);
static int be_send_fs_callback(zfs_handle_t *, void *);
static int be_demote_callback(zfs_handle_t *, void *);
static int be_demote_find_clone_callback(zfs_handle_t *, void *);
static int be_has_snapshot_callback(zfs_handle_t *, void *);
static int be_demote_get_one_clone(zfs_handle_t *, void *);
static int be_get_snap(char *, char **);
static int be_prep_clone_send_fs(zfs_handle_t *, be_transaction_data_t *,
    char *, int);
static boolean_t be_create_container_ds(char *);
static char *be_get_zone_be_name(char *root_ds, char *container_ds);
static int be_zone_root_exists_callback(zfs_handle_t *, void *);

/* ********************************************************************	*/
/*			Public Functions				*/
/* ********************************************************************	*/

/*
 * Function:	be_init
 * Description:	Creates the initial datasets for a BE and leaves them
 *		unpopulated.  The resultant BE can be mounted but can't
 *		yet be activated or booted.
 * Parameters:
 *		be_attrs - pointer to nvlist_t of attributes being passed in.
 *			The following attributes are used by this function:
 *
 *			BE_ATTR_NEW_BE_NAME		*required
 *			BE_ATTR_NEW_BE_POOL		*required
 *			BE_ATTR_ZFS_PROPERTIES		*optional
 *			BE_ATTR_FS_NAMES		*optional
 *			BE_ATTR_FS_NUM			*optional
 *			BE_ATTR_SHARED_FS_NAMES		*optional
 *			BE_ATTR_SHARED_FS_NUM		*optional
 * Return:
 *		BE_SUCCESS - Success
 *		be_errno_t - Failure
 * Scope:
 *		Public
 */
int
be_init(nvlist_t *be_attrs)
{
	be_transaction_data_t	bt = { 0 };
	zpool_handle_t	*zlp;
	nvlist_t	*zfs_props = NULL;
	char		nbe_root_ds[MAXPATHLEN];
	char		child_fs[MAXPATHLEN];
	char		**fs_names = NULL;
	char		**shared_fs_names = NULL;
	uint16_t	fs_num = 0;
	uint16_t	shared_fs_num = 0;
	int		nelem;
	int		i;
	int		zret = 0, ret = BE_SUCCESS;

	/* Initialize libzfs handle */
	if (!be_zfs_init())
		return (BE_ERR_INIT);

	/* Get new BE name */
	if (nvlist_lookup_string(be_attrs, BE_ATTR_NEW_BE_NAME, &bt.nbe_name)
	    != 0) {
		be_print_err(gettext("be_init: failed to lookup "
		    "BE_ATTR_NEW_BE_NAME attribute\n"));
		return (BE_ERR_INVAL);
	}

	/* Validate new BE name */
	if (!be_valid_be_name(bt.nbe_name)) {
		be_print_err(gettext("be_init: invalid BE name %s\n"),
		    bt.nbe_name);
		return (BE_ERR_INVAL);
	}

	/* Get zpool name */
	if (nvlist_lookup_string(be_attrs, BE_ATTR_NEW_BE_POOL, &bt.nbe_zpool)
	    != 0) {
		be_print_err(gettext("be_init: failed to lookup "
		    "BE_ATTR_NEW_BE_POOL attribute\n"));
		return (BE_ERR_INVAL);
	}

	/* Get file system attributes */
	nelem = 0;
	if (nvlist_lookup_pairs(be_attrs, 0,
	    BE_ATTR_FS_NUM, DATA_TYPE_UINT16, &fs_num,
	    BE_ATTR_FS_NAMES, DATA_TYPE_STRING_ARRAY, &fs_names, &nelem,
	    NULL) != 0) {
		be_print_err(gettext("be_init: failed to lookup fs "
		    "attributes\n"));
		return (BE_ERR_INVAL);
	}
	if (nelem != fs_num) {
		be_print_err(gettext("be_init: size of FS_NAMES array (%d) "
		    "does not match FS_NUM (%d)\n"), nelem, fs_num);
		return (BE_ERR_INVAL);
	}

	/* Get shared file system attributes */
	nelem = 0;
	if (nvlist_lookup_pairs(be_attrs, NV_FLAG_NOENTOK,
	    BE_ATTR_SHARED_FS_NUM, DATA_TYPE_UINT16, &shared_fs_num,
	    BE_ATTR_SHARED_FS_NAMES, DATA_TYPE_STRING_ARRAY, &shared_fs_names,
	    &nelem, NULL) != 0) {
		be_print_err(gettext("be_init: failed to lookup "
		    "shared fs attributes\n"));
		return (BE_ERR_INVAL);
	}
	if (nelem != shared_fs_num) {
		be_print_err(gettext("be_init: size of SHARED_FS_NAMES "
		    "array does not match SHARED_FS_NUM\n"));
		return (BE_ERR_INVAL);
	}

	/* Verify that nbe_zpool exists */
	if ((zlp = zpool_open(g_zfs, bt.nbe_zpool)) == NULL) {
		be_print_err(gettext("be_init: failed to "
		    "find existing zpool (%s): %s\n"), bt.nbe_zpool,
		    libzfs_error_description(g_zfs));
		return (zfs_err_to_be_err(g_zfs));
	}
	zpool_close(zlp);

	/*
	 * Verify BE container dataset in nbe_zpool exists.
	 * If not, create it.
	 */
	if (!be_create_container_ds(bt.nbe_zpool))
		return (BE_ERR_CREATDS);

	/*
	 * Verify that nbe_name doesn't already exist in some pool.
	 */
	if ((zret = zpool_iter(g_zfs, be_exists_callback, bt.nbe_name)) > 0) {
		be_print_err(gettext("be_init: BE (%s) already exists\n"),
		    bt.nbe_name);
		return (BE_ERR_BE_EXISTS);
	} else if (zret < 0) {
		be_print_err(gettext("be_init: zpool_iter failed: %s\n"),
		    libzfs_error_description(g_zfs));
		return (zfs_err_to_be_err(g_zfs));
	}

	/* Generate string for BE's root dataset */
	be_make_root_ds(bt.nbe_zpool, bt.nbe_name, nbe_root_ds,
	    sizeof (nbe_root_ds));

	/*
	 * Create property list for new BE root dataset.  If some
	 * zfs properties were already provided by the caller, dup
	 * that list.  Otherwise initialize a new property list.
	 */
	if (nvlist_lookup_pairs(be_attrs, NV_FLAG_NOENTOK,
	    BE_ATTR_ZFS_PROPERTIES, DATA_TYPE_NVLIST, &zfs_props, NULL)
	    != 0) {
		be_print_err(gettext("be_init: failed to lookup "
		    "BE_ATTR_ZFS_PROPERTIES attribute\n"));
		return (BE_ERR_INVAL);
	}
	if (zfs_props != NULL) {
		/* Make sure its a unique nvlist */
		if (!(zfs_props->nvl_nvflag & NV_UNIQUE_NAME) &&
		    !(zfs_props->nvl_nvflag & NV_UNIQUE_NAME_TYPE)) {
			be_print_err(gettext("be_init: ZFS property list "
			    "not unique\n"));
			return (BE_ERR_INVAL);
		}

		/* Dup the list */
		if (nvlist_dup(zfs_props, &bt.nbe_zfs_props, 0) != 0) {
			be_print_err(gettext("be_init: failed to dup ZFS "
			    "property list\n"));
			return (BE_ERR_NOMEM);
		}
	} else {
		/* Initialize new nvlist */
		if (nvlist_alloc(&bt.nbe_zfs_props, NV_UNIQUE_NAME, 0) != 0) {
			be_print_err(gettext("be_init: internal "
			    "error: out of memory\n"));
			return (BE_ERR_NOMEM);
		}
	}

	/* Set the mountpoint property for the root dataset */
	if (nvlist_add_string(bt.nbe_zfs_props,
	    zfs_prop_to_name(ZFS_PROP_MOUNTPOINT), "/") != 0) {
		be_print_err(gettext("be_init: internal error "
		    "out of memory\n"));
		ret = BE_ERR_NOMEM;
		goto done;
	}

	/* Set the 'canmount' property */
	if (nvlist_add_string(bt.nbe_zfs_props,
	    zfs_prop_to_name(ZFS_PROP_CANMOUNT), "noauto") != 0) {
		be_print_err(gettext("be_init: internal error "
		    "out of memory\n"));
		ret = BE_ERR_NOMEM;
		goto done;
	}

	/* Create BE root dataset for the new BE */
	if (zfs_create(g_zfs, nbe_root_ds, ZFS_TYPE_FILESYSTEM,
	    bt.nbe_zfs_props) != 0) {
		be_print_err(gettext("be_init: failed to "
		    "create BE root dataset (%s): %s\n"), nbe_root_ds,
		    libzfs_error_description(g_zfs));
		ret = zfs_err_to_be_err(g_zfs);
		goto done;
	}

	/* Set UUID for new BE */
	if ((ret = be_set_uuid(nbe_root_ds)) != BE_SUCCESS) {
		be_print_err(gettext("be_init: failed to "
		    "set uuid for new BE\n"));
	}

	/*
	 * Clear the mountpoint property so that the non-shared
	 * file systems created below inherit their mountpoints.
	 */
	(void) nvlist_remove(bt.nbe_zfs_props,
	    zfs_prop_to_name(ZFS_PROP_MOUNTPOINT), DATA_TYPE_STRING);

	/* Create the new BE's non-shared file systems */
	for (i = 0; i < fs_num && fs_names[i]; i++) {
		/*
		 * If fs == "/", skip it;
		 * we already created the root dataset
		 */
		if (strcmp(fs_names[i], "/") == 0)
			continue;

		/* Generate string for file system */
		(void) snprintf(child_fs, sizeof (child_fs), "%s%s",
		    nbe_root_ds, fs_names[i]);

		/* Create file system */
		if (zfs_create(g_zfs, child_fs, ZFS_TYPE_FILESYSTEM,
		    bt.nbe_zfs_props) != 0) {
			be_print_err(gettext("be_init: failed to create "
			    "BE's child dataset (%s): %s\n"), child_fs,
			    libzfs_error_description(g_zfs));
			ret = zfs_err_to_be_err(g_zfs);
			goto done;
		}
	}

	/* Create the new BE's shared file systems */
	if (shared_fs_num > 0) {
		nvlist_t	*props = NULL;

		if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) {
			be_print_err(gettext("be_init: nvlist_alloc failed\n"));
			ret = BE_ERR_NOMEM;
			goto done;
		}

		for (i = 0; i < shared_fs_num; i++) {
			/* Generate string for shared file system */
			(void) snprintf(child_fs, sizeof (child_fs), "%s%s",
			    bt.nbe_zpool, shared_fs_names[i]);

			if (nvlist_add_string(props,
			    zfs_prop_to_name(ZFS_PROP_MOUNTPOINT),
			    shared_fs_names[i]) != 0) {
				be_print_err(gettext("be_init: "
				    "internal error: out of memory\n"));
				nvlist_free(props);
				ret = BE_ERR_NOMEM;
				goto done;
			}

			/* Create file system if it doesn't already exist */
			if (zfs_dataset_exists(g_zfs, child_fs,
			    ZFS_TYPE_FILESYSTEM)) {
				continue;
			}
			if (zfs_create(g_zfs, child_fs, ZFS_TYPE_FILESYSTEM,
			    props) != 0) {
				be_print_err(gettext("be_init: failed to "
				    "create BE's shared dataset (%s): %s\n"),
				    child_fs, libzfs_error_description(g_zfs));
				ret = zfs_err_to_be_err(g_zfs);
				nvlist_free(props);
				goto done;
			}
		}

		nvlist_free(props);
	}

done:
	nvlist_free(bt.nbe_zfs_props);

	be_zfs_fini();

	return (ret);
}

/*
 * Function:	be_destroy
 * Description:	Destroy a BE and all of its children datasets, snapshots and
 *		zones that belong to the parent BE.
 * Parameters:
 *		be_attrs - pointer to nvlist_t of attributes being passed in.
 *			The following attributes are used by this function:
 *
 *			BE_ATTR_ORIG_BE_NAME		*required
 *			BE_ATTR_DESTROY_FLAGS		*optional
 * Return:
 *		BE_SUCCESS - Success
 *		be_errno_t - Failure
 * Scope:
 *		Public
 */
int
be_destroy(nvlist_t *be_attrs)
{
	zfs_handle_t		*zhp = NULL;
	be_transaction_data_t	bt = { 0 };
	be_transaction_data_t	cur_bt = { 0 };
	be_destroy_data_t	dd = { 0 };
	int			ret = BE_SUCCESS;
	uint16_t		flags = 0;
	boolean_t		bs_found = B_FALSE;
	int			zret;
	char			obe_root_ds[MAXPATHLEN];
	char			*mp = NULL;

	/* Initialize libzfs handle */
	if (!be_zfs_init())
		return (BE_ERR_INIT);

	/* Get name of BE to delete */
	if (nvlist_lookup_string(be_attrs, BE_ATTR_ORIG_BE_NAME, &bt.obe_name)
	    != 0) {
		be_print_err(gettext("be_destroy: failed to lookup "
		    "BE_ATTR_ORIG_BE_NAME attribute\n"));
		return (BE_ERR_INVAL);
	}

	/*
	 * Validate BE name. If valid, then check that the original BE is not
	 * the active BE. If it is the 'active' BE then return an error code
	 * since we can't destroy the active BE.
	 */
	if (!be_valid_be_name(bt.obe_name)) {
		be_print_err(gettext("be_destroy: invalid BE name %s\n"),
		    bt.obe_name);
		return (BE_ERR_INVAL);
	} else if (bt.obe_name != NULL) {
		if ((ret = be_find_current_be(&cur_bt)) != BE_SUCCESS) {
			return (ret);
		}
		if (strcmp(cur_bt.obe_name, bt.obe_name) == 0) {
			return (BE_ERR_DESTROY_CURR_BE);
		}
	}

	/* Get destroy flags if provided */
	if (nvlist_lookup_pairs(be_attrs, NV_FLAG_NOENTOK,
	    BE_ATTR_DESTROY_FLAGS, DATA_TYPE_UINT16, &flags, NULL)
	    != 0) {
		be_print_err(gettext("be_destroy: failed to lookup "
		    "BE_ATTR_DESTROY_FLAGS attribute\n"));
		return (BE_ERR_INVAL);
	}

	dd.destroy_snaps = flags & BE_DESTROY_FLAG_SNAPSHOTS;
	dd.force_unmount = flags & BE_DESTROY_FLAG_FORCE_UNMOUNT;

	/* Find which zpool obe_name lives in */
	if ((zret = zpool_iter(g_zfs, be_find_zpool_callback, &bt)) == 0) {
		be_print_err(gettext("be_destroy: failed to find zpool "
		    "for BE (%s)\n"), bt.obe_name);
		return (BE_ERR_BE_NOENT);
	} else if (zret < 0) {
		be_print_err(gettext("be_destroy: zpool_iter failed: %s\n"),
		    libzfs_error_description(g_zfs));
		return (zfs_err_to_be_err(g_zfs));
	}

	/* Generate string for obe_name's root dataset */
	be_make_root_ds(bt.obe_zpool, bt.obe_name, obe_root_ds,
	    sizeof (obe_root_ds));
	bt.obe_root_ds = obe_root_ds;

	if (getzoneid() != GLOBAL_ZONEID) {
		if (!be_zone_compare_uuids(bt.obe_root_ds)) {
			if (be_is_active_on_boot(bt.obe_name)) {
				be_print_err(gettext("be_destroy: destroying "
				    "active zone root dataset from non-active "
				    "global BE is not supported\n"));
				return (BE_ERR_NOTSUP);
			}
		}
	}

	/*
	 * Detect if the BE to destroy has the 'active on boot' property set.
	 * If so, set the 'active on boot' property on the the 'active' BE.
	 */
	if (be_is_active_on_boot(bt.obe_name)) {
		if ((ret = be_activate_current_be()) != BE_SUCCESS) {
			be_print_err(gettext("be_destroy: failed to "
			    "make the current BE 'active on boot'\n"));
			return (ret);
		}
	}

	/*
	 * Detect if the BE to destroy is referenced in the pool's nextboot
	 * field, and unset it if so.
	 */
	if (getzoneid() == GLOBAL_ZONEID) {
		char *nextboot = NULL;

		if (lzbe_get_boot_device(bt.obe_zpool, &nextboot) == 0 &&
		    nextboot != NULL && strcmp(nextboot, bt.obe_root_ds) == 0) {
			if (lzbe_set_boot_device(bt.obe_zpool,
			    lzbe_add, "") != 0) {
				be_print_err(gettext("be_destroy: failed to "
				    "remove temporary activation for "
				    "dataset %s on pool %s\n"),
				    bt.obe_root_ds, bt.obe_zpool);
				free(nextboot);
				return (BE_ERR_UNKNOWN);
			}
		}
		free(nextboot);
	}

	/* Get handle to BE's root dataset */
	if ((zhp = zfs_open(g_zfs, bt.obe_root_ds, ZFS_TYPE_FILESYSTEM)) ==
	    NULL) {
		be_print_err(gettext("be_destroy: failed to "
		    "open BE root dataset (%s): %s\n"), bt.obe_root_ds,
		    libzfs_error_description(g_zfs));
		return (zfs_err_to_be_err(g_zfs));
	}

	/*
	 * Check if BE has snapshots and BE_DESTROY_FLAG_SNAPSHOTS
	 * is not set.
	 */
	(void) zfs_iter_snapshots(zhp, B_FALSE, be_has_snapshot_callback,
	    &bs_found);
	if (!dd.destroy_snaps && bs_found) {
		ZFS_CLOSE(zhp);
		return (BE_ERR_SS_EXISTS);
	}

	/* Get the UUID of the global BE */
	if (getzoneid() == GLOBAL_ZONEID) {
		if (be_get_uuid(zfs_get_name(zhp),
		    &dd.gz_be_uuid) != BE_SUCCESS) {
			be_print_err(gettext("be_destroy: BE has no "
			"UUID (%s)\n"), zfs_get_name(zhp));
		}
	}

	/*
	 * If the global BE is mounted, make sure we've been given the
	 * flag to forcibly unmount it.
	 */
	if (zfs_is_mounted(zhp, &mp)) {
		if (!(dd.force_unmount)) {
			be_print_err(gettext("be_destroy: "
			    "%s is currently mounted at %s, cannot destroy\n"),
			    bt.obe_name, mp != NULL ? mp : "<unknown>");

			free(mp);
			ZFS_CLOSE(zhp);
			return (BE_ERR_MOUNTED);
		}
		free(mp);
	}

	/*
	 * Destroy the non-global zone BE's if we are in the global zone
	 * and there is a UUID associated with the global zone BE
	 */
	if (getzoneid() == GLOBAL_ZONEID && !uuid_is_null(dd.gz_be_uuid)) {
		if ((ret = be_destroy_zones(bt.obe_name, bt.obe_root_ds, &dd))
		    != BE_SUCCESS) {
			be_print_err(gettext("be_destroy: failed to "
			    "destroy one or more zones for BE %s\n"),
			    bt.obe_name);
			goto done;
		}
	}

	/* Unmount the BE if it was mounted */
	if (zfs_is_mounted(zhp, NULL)) {
		if ((ret = _be_unmount(bt.obe_name, BE_UNMOUNT_FLAG_FORCE))
		    != BE_SUCCESS) {
			be_print_err(gettext("be_destroy: "
			    "failed to unmount %s\n"), bt.obe_name);
			ZFS_CLOSE(zhp);
			return (ret);
		}
	}
	ZFS_CLOSE(zhp);

	/* Destroy this BE */
	if ((ret = _be_destroy((const char *)bt.obe_root_ds, &dd))
	    != BE_SUCCESS) {
		goto done;
	}

	/* Remove BE's entry from the boot menu */
	if (getzoneid() == GLOBAL_ZONEID) {
		if ((ret = be_remove_menu(bt.obe_name, bt.obe_zpool, NULL))
		    != BE_SUCCESS) {
			be_print_err(gettext("be_destroy: failed to "
			    "remove BE %s from the boot menu\n"),
			    bt.obe_root_ds);
			goto done;
		}
	}

done:
	be_zfs_fini();

	return (ret);
}

/*
 * Function:	be_copy
 * Description:	This function makes a copy of an existing BE.  If the original
 *		BE and the new BE are in the same pool, it uses zfs cloning to
 *		create the new BE, otherwise it does a physical copy.
 *		If the original BE name isn't provided, it uses the currently
 *		booted BE.  If the new BE name isn't provided, it creates an
 *		auto named BE and returns that name to the caller.
 * Parameters:
 *		be_attrs - pointer to nvlist_t of attributes being passed in.
 *			The following attributes are used by this function:
 *