[mirror_zfs.git] / module / zfs / spa_config.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 2011 Nexenta Systems, Inc. All rights reserved.
 * Copyright (c) 2012 by Delphix. All rights reserved.
 */

#include <sys/spa.h>
#include <sys/spa_impl.h>
#include <sys/nvpair.h>
#include <sys/uio.h>
#include <sys/fs/zfs.h>
#include <sys/vdev_impl.h>
#include <sys/zfs_ioctl.h>
#include <sys/utsname.h>
#include <sys/systeminfo.h>
#include <sys/sunddi.h>
#include <sys/zfeature.h>
#ifdef _KERNEL
#include <sys/kobj.h>
#include <sys/zone.h>
#endif

/*
 * Pool configuration repository.
 *
 * Pool configuration is stored as a packed nvlist on the filesystem.  By
 * default, all pools are stored in /etc/zfs/zpool.cache and loaded on boot
 * (when the ZFS module is loaded).  Pools can also have the 'cachefile'
 * property set that allows them to be stored in an alternate location until
 * the control of external software.
 *
 * For each cache file, we have a single nvlist which holds all the
 * configuration information.  When the module loads, we read this information
 * from /etc/zfs/zpool.cache and populate the SPA namespace.  This namespace is
 * maintained independently in spa.c.  Whenever the namespace is modified, or
 * the configuration of a pool is changed, we call spa_config_sync(), which
 * walks through all the active pools and writes the configuration to disk.
 */

static uint64_t spa_config_generation = 1;

/*
 * This can be overridden in userland to preserve an alternate namespace for
 * userland pools when doing testing.
 */
char *spa_config_path = ZPOOL_CACHE;

/*
 * Called when the module is first loaded, this routine loads the configuration
 * file into the SPA namespace.  It does not actually open or load the pools; it
 * only populates the namespace.
 */
void
spa_config_load(void)
{
	void *buf = NULL;
	nvlist_t *nvlist, *child;
	nvpair_t *nvpair;
	char *pathname;
	struct _buf *file;
	uint64_t fsize;

	/*
	 * Open the configuration file.
	 */
	pathname = kmem_alloc(MAXPATHLEN, KM_PUSHPAGE);

	(void) snprintf(pathname, MAXPATHLEN, "%s%s",
	    (rootdir != NULL) ? "./" : "", spa_config_path);

	file = kobj_open_file(pathname);

	kmem_free(pathname, MAXPATHLEN);

	if (file == (struct _buf *)-1)
		return;

	if (kobj_get_filesize(file, &fsize) != 0)
		goto out;

	buf = kmem_alloc(fsize, KM_PUSHPAGE | KM_NODEBUG);

	/*
	 * Read the nvlist from the file.
	 */
	if (kobj_read_file(file, buf, fsize, 0) < 0)
		goto out;

	/*
	 * Unpack the nvlist.
	 */
	if (nvlist_unpack(buf, fsize, &nvlist, KM_PUSHPAGE) != 0)
		goto out;

	/*
	 * Iterate over all elements in the nvlist, creating a new spa_t for
	 * each one with the specified configuration.
	 */
	mutex_enter(&spa_namespace_lock);
	nvpair = NULL;
	while ((nvpair = nvlist_next_nvpair(nvlist, nvpair)) != NULL) {
		if (nvpair_type(nvpair) != DATA_TYPE_NVLIST)
			continue;

		VERIFY(nvpair_value_nvlist(nvpair, &child) == 0);

		if (spa_lookup(nvpair_name(nvpair)) != NULL)
			continue;
		(void) spa_add(nvpair_name(nvpair), child, NULL);
	}
	mutex_exit(&spa_namespace_lock);

	nvlist_free(nvlist);

out:
	if (buf != NULL)
		kmem_free(buf, fsize);

	kobj_close_file(file);
}

static void
spa_config_write(spa_config_dirent_t *dp, nvlist_t *nvl)
{
	size_t buflen;
	char *buf;
	vnode_t *vp;
	int oflags = FWRITE | FTRUNC | FCREAT | FOFFMAX;
	char *temp;

	/*
	 * If the nvlist is empty (NULL), then remove the old cachefile.
	 */
	if (nvl == NULL) {
		(void) vn_remove(dp->scd_path, UIO_SYSSPACE, RMFILE);
		return;
	}

	/*
	 * Pack the configuration into a buffer.
	 */
	VERIFY(nvlist_size(nvl, &buflen, NV_ENCODE_XDR) == 0);

	buf = kmem_alloc(buflen, KM_PUSHPAGE | KM_NODEBUG);
	temp = kmem_zalloc(MAXPATHLEN, KM_PUSHPAGE);

	VERIFY(nvlist_pack(nvl, &buf, &buflen, NV_ENCODE_XDR,
	    KM_PUSHPAGE) == 0);

	/*
	 * Write the configuration to disk.  We need to do the traditional
	 * 'write to temporary file, sync, move over original' to make sure we
	 * always have a consistent view of the data.
	 */
	(void) snprintf(temp, MAXPATHLEN, "%s.tmp", dp->scd_path);

	if (vn_open(temp, UIO_SYSSPACE, oflags, 0644, &vp, CRCREAT, 0) == 0) {
		if (vn_rdwr(UIO_WRITE, vp, buf, buflen, 0, UIO_SYSSPACE,
		    0, RLIM64_INFINITY, kcred, NULL) == 0 &&
		    VOP_FSYNC(vp, FSYNC, kcred, NULL) == 0) {
			(void) vn_rename(temp, dp->scd_path, UIO_SYSSPACE);
		}
		(void) VOP_CLOSE(vp, oflags, 1, 0, kcred, NULL);
	}

	(void) vn_remove(temp, UIO_SYSSPACE, RMFILE);

	kmem_free(buf, buflen);
	kmem_free(temp, MAXPATHLEN);
}

/*
 * Synchronize pool configuration to disk.  This must be called with the
 * namespace lock held.
 */
void
spa_config_sync(spa_t *target, boolean_t removing, boolean_t postsysevent)
{
	spa_config_dirent_t *dp, *tdp;
	nvlist_t *nvl;

	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	if (rootdir == NULL || !(spa_mode_global & FWRITE))
		return;

	/*
	 * Iterate over all cachefiles for the pool, past or present.  When the
	 * cachefile is changed, the new one is pushed onto this list, allowing
	 * us to update previous cachefiles that no longer contain this pool.
	 */
	for (dp = list_head(&target->spa_config_list); dp != NULL;
	    dp = list_next(&target->spa_config_list, dp)) {
		spa_t *spa = NULL;
		if (dp->scd_path == NULL)
			continue;

		/*
		 * Iterate over all pools, adding any matching pools to 'nvl'.
		 */
		nvl = NULL;
		while ((spa = spa_next(spa)) != NULL) {
			if (spa == target && removing)
				continue;

			mutex_enter(&spa->spa_props_lock);
			tdp = list_head(&spa->spa_config_list);
			if (spa->spa_config == NULL ||
			    tdp->scd_path == NULL ||
			    strcmp(tdp->scd_path, dp->scd_path) != 0) {
				mutex_exit(&spa->spa_props_lock);
				continue;
			}

			if (nvl == NULL)
				VERIFY(nvlist_alloc(&nvl, NV_UNIQUE_NAME,
				    KM_PUSHPAGE) == 0);

			VERIFY(nvlist_add_nvlist(nvl, spa->spa_name,
			    spa->spa_config) == 0);
			mutex_exit(&spa->spa_props_lock);
		}

		spa_config_write(dp, nvl);
		nvlist_free(nvl);
	}

	/*
	 * Remove any config entries older than the current one.
	 */
	dp = list_head(&target->spa_config_list);
	while ((tdp = list_next(&target->spa_config_list, dp)) != NULL) {
		list_remove(&target->spa_config_list, tdp);
		if (tdp->scd_path != NULL)
			spa_strfree(tdp->scd_path);
		kmem_free(tdp, sizeof (spa_config_dirent_t));
	}

	spa_config_generation++;

	if (postsysevent)
		spa_event_notify(target, NULL, FM_EREPORT_ZFS_CONFIG_SYNC);
}

/*
 * Sigh.  Inside a local zone, we don't have access to /etc/zfs/zpool.cache,
 * and we don't want to allow the local zone to see all the pools anyway.
 * So we have to invent the ZFS_IOC_CONFIG ioctl to grab the configuration
 * information for all pool visible within the zone.
 */
nvlist_t *
spa_all_configs(uint64_t *generation)
{
	nvlist_t *pools;
	spa_t *spa = NULL;

	if (*generation == spa_config_generation)
		return (NULL);

	VERIFY(nvlist_alloc(&pools, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);

	mutex_enter(&spa_namespace_lock);
	while ((spa = spa_next(spa)) != NULL) {
		if (INGLOBALZONE(curproc) ||
		    zone_dataset_visible(spa_name(spa), NULL)) {
			mutex_enter(&spa->spa_props_lock);
			VERIFY(nvlist_add_nvlist(pools, spa_name(spa),
			    spa->spa_config) == 0);
			mutex_exit(&spa->spa_props_lock);
		}
	}
	*generation = spa_config_generation;
	mutex_exit(&spa_namespace_lock);

	return (pools);
}

void
spa_config_set(spa_t *spa, nvlist_t *config)
{
	mutex_enter(&spa->spa_props_lock);
	if (spa->spa_config != NULL)
		nvlist_free(spa->spa_config);
	spa->spa_config = config;
	mutex_exit(&spa->spa_props_lock);
}

/*
 * Generate the pool's configuration based on the current in-core state.
 * We infer whether to generate a complete config or just one top-level config
 * based on whether vd is the root vdev.
 */
nvlist_t *
spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, int getstats)
{
	nvlist_t *config, *nvroot;
	vdev_t *rvd = spa->spa_root_vdev;
	unsigned long hostid = 0;
	boolean_t locked = B_FALSE;
	uint64_t split_guid;

	if (vd == NULL) {
		vd = rvd;
		locked = B_TRUE;
		spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER);
	}

	ASSERT(spa_config_held(spa, SCL_CONFIG | SCL_STATE, RW_READER) ==
	    (SCL_CONFIG | SCL_STATE));

	/*
	 * If txg is -1, report the current value of spa->spa_config_txg.
	 */
	if (txg == -1ULL)
		txg = spa->spa_config_txg;

	VERIFY(nvlist_alloc(&config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);

	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_VERSION,
	    spa_version(spa)) == 0);
	VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME,
	    spa_name(spa)) == 0);
	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE,
	    spa_state(spa)) == 0);
	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG,
	    txg) == 0);
	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID,
	    spa_guid(spa)) == 0);
	VERIFY(spa->spa_comment == NULL || nvlist_add_string(config,
	    ZPOOL_CONFIG_COMMENT, spa->spa_comment) == 0);


#ifdef	_KERNEL
	hostid = zone_get_hostid(NULL);
#else	/* _KERNEL */
	/*
	 * We're emulating the system's hostid in userland, so we can't use
	 * zone_get_hostid().
	 */
	(void) ddi_strtoul(hw_serial, NULL, 10, &hostid);
#endif	/* _KERNEL */
	if (hostid != 0) {
		VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_HOSTID,
		    hostid) == 0);
	}
	VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_HOSTNAME,
	    utsname.nodename) == 0);

	if (vd != rvd) {
		VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_TOP_GUID,
		    vd->vdev_top->vdev_guid) == 0);
		VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_GUID,
		    vd->vdev_guid) == 0);
		if (vd->vdev_isspare)
			VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_IS_SPARE,
			    1ULL) == 0);
		if (vd->vdev_islog)
			VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_IS_LOG,
			    1ULL) == 0);
		vd = vd->vdev_top;		/* label contains top config */
	} else {
		/*
		 * Only add the (potentially large) split information
		 * in the mos config, and not in the vdev labels
		 */
		if (spa->spa_config_splitting != NULL)
			VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_SPLIT,
			    spa->spa_config_splitting) == 0);
	}

	/*
	 * Add the top-level config.  We even add this on pools which
	 * don't support holes in the namespace.
	 */
	vdev_top_config_generate(spa, config);

	/*
	 * If we're splitting, record the original pool's guid.
	 */
	if (spa->spa_config_splitting != NULL &&
	    nvlist_lookup_uint64(spa->spa_config_splitting,
	    ZPOOL_CONFIG_SPLIT_GUID, &split_guid) == 0) {
		VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_SPLIT_GUID,
		    split_guid) == 0);
	}

	nvroot = vdev_config_generate(spa, vd, getstats, 0);
	VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, nvroot) == 0);
	nvlist_free(nvroot);

	/*
	 * Store what's necessary for reading the MOS in the label.
	 */
	VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_FEATURES_FOR_READ,
	    spa->spa_label_features) == 0);

	if (getstats && spa_load_state(spa) == SPA_LOAD_NONE) {
		ddt_histogram_t *ddh;
		ddt_stat_t *dds;
		ddt_object_t *ddo;

		ddh = kmem_zalloc(sizeof (ddt_histogram_t), KM_PUSHPAGE);
		ddt_get_dedup_histogram(spa, ddh);
		VERIFY(nvlist_add_uint64_array(config,
		    ZPOOL_CONFIG_DDT_HISTOGRAM,
		    (uint64_t *)ddh, sizeof (*ddh) / sizeof (uint64_t)) == 0);
		kmem_free(ddh, sizeof (ddt_histogram_t));

		ddo = kmem_zalloc(sizeof (ddt_object_t), KM_PUSHPAGE);
		ddt_get_dedup_object_stats(spa, ddo);
		VERIFY(nvlist_add_uint64_array(config,
		    ZPOOL_CONFIG_DDT_OBJ_STATS,
		    (uint64_t *)ddo, sizeof (*ddo) / sizeof (uint64_t)) == 0);
		kmem_free(ddo, sizeof (ddt_object_t));

		dds = kmem_zalloc(sizeof (ddt_stat_t), KM_PUSHPAGE);
		ddt_get_dedup_stats(spa, dds);
		VERIFY(nvlist_add_uint64_array(config,
		    ZPOOL_CONFIG_DDT_STATS,
		    (uint64_t *)dds, sizeof (*dds) / sizeof (uint64_t)) == 0);
		kmem_free(dds, sizeof (ddt_stat_t));
	}

	if (locked)
		spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG);

	return (config);
}

/*
 * Update all disk labels, generate a fresh config based on the current
 * in-core state, and sync the global config cache (do not sync the config
 * cache if this is a booting rootpool).
 */
void
spa_config_update(spa_t *spa, int what)
{
	vdev_t *rvd = spa->spa_root_vdev;
	uint64_t txg;
	int c;

	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	txg = spa_last_synced_txg(spa) + 1;
	if (what == SPA_CONFIG_UPDATE_POOL) {
		vdev_config_dirty(rvd);
	} else {
		/*
		 * If we have top-level vdevs that were added but have
		 * not yet been prepared for allocation, do that now.
		 * (It's safe now because the config cache is up to date,
		 * so it will be able to translate the new DVAs.)
		 * See comments in spa_vdev_add() for full details.
		 */
		for (c = 0; c < rvd->vdev_children; c++) {
			vdev_t *tvd = rvd->vdev_child[c];
			if (tvd->vdev_ms_array == 0)
				vdev_metaslab_set_size(tvd);
			vdev_expand(tvd, txg);
		}
	}
	spa_config_exit(spa, SCL_ALL, FTAG);

	/*
	 * Wait for the mosconfig to be regenerated and synced.
	 */
	txg_wait_synced(spa->spa_dsl_pool, txg);

	/*
	 * Update the global config cache to reflect the new mosconfig.
	 */
	if (!spa->spa_is_root)
		spa_config_sync(spa, B_FALSE, what != SPA_CONFIG_UPDATE_POOL);

	if (what == SPA_CONFIG_UPDATE_POOL)
		spa_config_update(spa, SPA_CONFIG_UPDATE_VDEVS);
}

#if defined(_KERNEL) && defined(HAVE_SPL)
EXPORT_SYMBOL(spa_config_sync);
EXPORT_SYMBOL(spa_config_load);
EXPORT_SYMBOL(spa_all_configs);
EXPORT_SYMBOL(spa_config_set);
EXPORT_SYMBOL(spa_config_generate);
EXPORT_SYMBOL(spa_config_update);

module_param(spa_config_path, charp, 0444);
MODULE_PARM_DESC(spa_config_path, "SPA config file (/etc/zfs/zpool.cache)");
#endif
Commit	Line	Data
34dc7c2f BB	1	/*
	2	* CDDL HEADER START
	3	*
	4	* The contents of this file are subject to the terms of the
	5	* Common Development and Distribution License (the "License").
	6	* You may not use this file except in compliance with the License.
	7	*
	8	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	9	* or http://www.opensolaris.org/os/licensing.
	10	* See the License for the specific language governing permissions
	11	* and limitations under the License.
	12	*
	13	* When distributing Covered Code, include this CDDL HEADER in each
	14	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	15	* If applicable, add the following below this CDDL HEADER, with the
	16	* fields enclosed by brackets "[]" replaced with your own identifying
	17	* information: Portions Copyright [yyyy] [name of copyright owner]
	18	*
	19	* CDDL HEADER END
	20	*/
	21
	22	/*
428870ff	23	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
d96eb2b1	24	* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
9ae529ec	25	* Copyright (c) 2012 by Delphix. All rights reserved.
34dc7c2f BB	26	*/
34dc7c2f BB	27
34dc7c2f BB	28	#include <sys/spa.h>
	29	#include <sys/spa_impl.h>
	30	#include <sys/nvpair.h>
	31	#include <sys/uio.h>
	32	#include <sys/fs/zfs.h>
	33	#include <sys/vdev_impl.h>
	34	#include <sys/zfs_ioctl.h>
	35	#include <sys/utsname.h>
	36	#include <sys/systeminfo.h>
	37	#include <sys/sunddi.h>
9ae529ec	38	#include <sys/zfeature.h>
34dc7c2f BB	39	#ifdef _KERNEL
34dc7c2f BB	40	#include <sys/kobj.h>
d164b209	41	#include <sys/zone.h>
34dc7c2f BB	42	#endif
	43
	44	/*
	45	* Pool configuration repository.
	46	*
	47	* Pool configuration is stored as a packed nvlist on the filesystem. By
	48	* default, all pools are stored in /etc/zfs/zpool.cache and loaded on boot
	49	* (when the ZFS module is loaded). Pools can also have the 'cachefile'
	50	* property set that allows them to be stored in an alternate location until
	51	* the control of external software.
	52	*
	53	* For each cache file, we have a single nvlist which holds all the
	54	* configuration information. When the module loads, we read this information
	55	* from /etc/zfs/zpool.cache and populate the SPA namespace. This namespace is
	56	* maintained independently in spa.c. Whenever the namespace is modified, or
	57	* the configuration of a pool is changed, we call spa_config_sync(), which
	58	* walks through all the active pools and writes the configuration to disk.
	59	*/
	60
	61	static uint64_t spa_config_generation = 1;
	62
	63	/*
	64	* This can be overridden in userland to preserve an alternate namespace for
	65	* userland pools when doing testing.
	66	*/
c28b2279	67	char *spa_config_path = ZPOOL_CACHE;
34dc7c2f BB	68
	69	/*
	70	* Called when the module is first loaded, this routine loads the configuration
	71	* file into the SPA namespace. It does not actually open or load the pools; it
	72	* only populates the namespace.
	73	*/
	74	void
	75	spa_config_load(void)
	76	{
	77	void *buf = NULL;
	78	nvlist_t nvlist, child;
	79	nvpair_t *nvpair;
b128c09f	80	char *pathname;
34dc7c2f BB	81	struct _buf *file;
	82	uint64_t fsize;
	83
	84	/*
	85	* Open the configuration file.
	86	*/
b8d06fca	87	pathname = kmem_alloc(MAXPATHLEN, KM_PUSHPAGE);
b128c09f BB	88
	89	(void) snprintf(pathname, MAXPATHLEN, "%s%s",
	90	(rootdir != NULL) ? "./" : "", spa_config_path);
34dc7c2f BB	91
34dc7c2f BB	92	file = kobj_open_file(pathname);
b128c09f BB	93
	94	kmem_free(pathname, MAXPATHLEN);
	95
34dc7c2f BB	96	if (file == (struct _buf *)-1)
	97	return;
	98
	99	if (kobj_get_filesize(file, &fsize) != 0)
	100	goto out;
	101
b8d06fca	102	buf = kmem_alloc(fsize, KM_PUSHPAGE \| KM_NODEBUG);
34dc7c2f BB	103
	104	/*
	105	* Read the nvlist from the file.
	106	*/
	107	if (kobj_read_file(file, buf, fsize, 0) < 0)
	108	goto out;
	109
	110	/*
	111	* Unpack the nvlist.
	112	*/
b8d06fca	113	if (nvlist_unpack(buf, fsize, &nvlist, KM_PUSHPAGE) != 0)
34dc7c2f BB	114	goto out;
	115
	116	/*
	117	* Iterate over all elements in the nvlist, creating a new spa_t for
	118	* each one with the specified configuration.
	119	*/
	120	mutex_enter(&spa_namespace_lock);
	121	nvpair = NULL;
	122	while ((nvpair = nvlist_next_nvpair(nvlist, nvpair)) != NULL) {
34dc7c2f BB	123	if (nvpair_type(nvpair) != DATA_TYPE_NVLIST)
	124	continue;
	125
	126	VERIFY(nvpair_value_nvlist(nvpair, &child) == 0);
	127
	128	if (spa_lookup(nvpair_name(nvpair)) != NULL)
	129	continue;
428870ff	130	(void) spa_add(nvpair_name(nvpair), child, NULL);
34dc7c2f BB	131	}
	132	mutex_exit(&spa_namespace_lock);
	133
	134	nvlist_free(nvlist);
	135
	136	out:
	137	if (buf != NULL)
	138	kmem_free(buf, fsize);
	139
	140	kobj_close_file(file);
	141	}
	142
34dc7c2f	143	static void
b128c09f	144	spa_config_write(spa_config_dirent_t dp, nvlist_t nvl)
34dc7c2f	145	{
34dc7c2f BB	146	size_t buflen;
	147	char *buf;
	148	vnode_t *vp;
	149	int oflags = FWRITE \| FTRUNC \| FCREAT \| FOFFMAX;
b128c09f BB	150	char *temp;
	151
	152	/*
	153	* If the nvlist is empty (NULL), then remove the old cachefile.
	154	*/
	155	if (nvl == NULL) {
	156	(void) vn_remove(dp->scd_path, UIO_SYSSPACE, RMFILE);
	157	return;
	158	}
34dc7c2f BB	159
	160	/*
	161	* Pack the configuration into a buffer.
	162	*/
b128c09f	163	VERIFY(nvlist_size(nvl, &buflen, NV_ENCODE_XDR) == 0);
34dc7c2f	164
b8d06fca RY	165	buf = kmem_alloc(buflen, KM_PUSHPAGE \| KM_NODEBUG);
b8d06fca RY	166	temp = kmem_zalloc(MAXPATHLEN, KM_PUSHPAGE);
34dc7c2f	167
b128c09f	168	VERIFY(nvlist_pack(nvl, &buf, &buflen, NV_ENCODE_XDR,
b8d06fca	169	KM_PUSHPAGE) == 0);
34dc7c2f BB	170
	171	/*
	172	* Write the configuration to disk. We need to do the traditional
	173	* 'write to temporary file, sync, move over original' to make sure we
	174	* always have a consistent view of the data.
	175	*/
b128c09f	176	(void) snprintf(temp, MAXPATHLEN, "%s.tmp", dp->scd_path);
34dc7c2f	177
b128c09f BB	178	if (vn_open(temp, UIO_SYSSPACE, oflags, 0644, &vp, CRCREAT, 0) == 0) {
	179	if (vn_rdwr(UIO_WRITE, vp, buf, buflen, 0, UIO_SYSSPACE,
	180	0, RLIM64_INFINITY, kcred, NULL) == 0 &&
	181	VOP_FSYNC(vp, FSYNC, kcred, NULL) == 0) {
	182	(void) vn_rename(temp, dp->scd_path, UIO_SYSSPACE);
	183	}
	184	(void) VOP_CLOSE(vp, oflags, 1, 0, kcred, NULL);
34dc7c2f BB	185	}
34dc7c2f BB	186
b128c09f	187	(void) vn_remove(temp, UIO_SYSSPACE, RMFILE);
34dc7c2f	188
34dc7c2f	189	kmem_free(buf, buflen);
b128c09f	190	kmem_free(temp, MAXPATHLEN);
34dc7c2f BB	191	}
	192
	193	/*
b128c09f BB	194	* Synchronize pool configuration to disk. This must be called with the
b128c09f BB	195	* namespace lock held.
34dc7c2f BB	196	*/
34dc7c2f BB	197	void
b128c09f	198	spa_config_sync(spa_t *target, boolean_t removing, boolean_t postsysevent)
34dc7c2f	199	{
b128c09f BB	200	spa_config_dirent_t dp, tdp;
b128c09f BB	201	nvlist_t *nvl;
34dc7c2f BB	202
	203	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	204
45d1cae3	205	if (rootdir == NULL \|\| !(spa_mode_global & FWRITE))
fb5f0bc8 BB	206	return;
fb5f0bc8 BB	207
b128c09f BB	208	/*
	209	* Iterate over all cachefiles for the pool, past or present. When the
	210	* cachefile is changed, the new one is pushed onto this list, allowing
	211	* us to update previous cachefiles that no longer contain this pool.
	212	*/
	213	for (dp = list_head(&target->spa_config_list); dp != NULL;
	214	dp = list_next(&target->spa_config_list, dp)) {
	215	spa_t *spa = NULL;
	216	if (dp->scd_path == NULL)
	217	continue;
	218
	219	/*
	220	* Iterate over all pools, adding any matching pools to 'nvl'.
	221	*/
	222	nvl = NULL;
	223	while ((spa = spa_next(spa)) != NULL) {
	224	if (spa == target && removing)
	225	continue;
	226
	227	mutex_enter(&spa->spa_props_lock);
	228	tdp = list_head(&spa->spa_config_list);
	229	if (spa->spa_config == NULL \|\|
	230	tdp->scd_path == NULL \|\|
	231	strcmp(tdp->scd_path, dp->scd_path) != 0) {
	232	mutex_exit(&spa->spa_props_lock);
	233	continue;
	234	}
	235
	236	if (nvl == NULL)
	237	VERIFY(nvlist_alloc(&nvl, NV_UNIQUE_NAME,
b8d06fca	238	KM_PUSHPAGE) == 0);
b128c09f BB	239
	240	VERIFY(nvlist_add_nvlist(nvl, spa->spa_name,
	241	spa->spa_config) == 0);
	242	mutex_exit(&spa->spa_props_lock);
	243	}
	244
	245	spa_config_write(dp, nvl);
	246	nvlist_free(nvl);
	247	}
34dc7c2f BB	248
34dc7c2f BB	249	/*
b128c09f	250	* Remove any config entries older than the current one.
34dc7c2f	251	*/
b128c09f BB	252	dp = list_head(&target->spa_config_list);
	253	while ((tdp = list_next(&target->spa_config_list, dp)) != NULL) {
	254	list_remove(&target->spa_config_list, tdp);
	255	if (tdp->scd_path != NULL)
	256	spa_strfree(tdp->scd_path);
	257	kmem_free(tdp, sizeof (spa_config_dirent_t));
34dc7c2f BB	258	}
	259
	260	spa_config_generation++;
b128c09f BB	261
b128c09f BB	262	if (postsysevent)
26685276	263	spa_event_notify(target, NULL, FM_EREPORT_ZFS_CONFIG_SYNC);
34dc7c2f BB	264	}
	265
	266	/*
	267	* Sigh. Inside a local zone, we don't have access to /etc/zfs/zpool.cache,
	268	* and we don't want to allow the local zone to see all the pools anyway.
	269	* So we have to invent the ZFS_IOC_CONFIG ioctl to grab the configuration
	270	* information for all pool visible within the zone.
	271	*/
	272	nvlist_t *
	273	spa_all_configs(uint64_t *generation)
	274	{
	275	nvlist_t *pools;
b128c09f	276	spa_t *spa = NULL;
34dc7c2f BB	277
	278	if (*generation == spa_config_generation)
	279	return (NULL);
	280
b8d06fca	281	VERIFY(nvlist_alloc(&pools, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);
34dc7c2f	282
34dc7c2f BB	283	mutex_enter(&spa_namespace_lock);
	284	while ((spa = spa_next(spa)) != NULL) {
	285	if (INGLOBALZONE(curproc) \|\|
	286	zone_dataset_visible(spa_name(spa), NULL)) {
b128c09f	287	mutex_enter(&spa->spa_props_lock);
34dc7c2f BB	288	VERIFY(nvlist_add_nvlist(pools, spa_name(spa),
34dc7c2f BB	289	spa->spa_config) == 0);
b128c09f	290	mutex_exit(&spa->spa_props_lock);
34dc7c2f BB	291	}
34dc7c2f BB	292	}
34dc7c2f	293	*generation = spa_config_generation;
b128c09f	294	mutex_exit(&spa_namespace_lock);
34dc7c2f BB	295
	296	return (pools);
	297	}
	298
	299	void
	300	spa_config_set(spa_t spa, nvlist_t config)
	301	{
b128c09f	302	mutex_enter(&spa->spa_props_lock);
34dc7c2f BB	303	if (spa->spa_config != NULL)
	304	nvlist_free(spa->spa_config);
	305	spa->spa_config = config;
b128c09f	306	mutex_exit(&spa->spa_props_lock);
34dc7c2f BB	307	}
	308
	309	/*
	310	* Generate the pool's configuration based on the current in-core state.
	311	* We infer whether to generate a complete config or just one top-level config
	312	* based on whether vd is the root vdev.
	313	*/
	314	nvlist_t *
	315	spa_config_generate(spa_t spa, vdev_t vd, uint64_t txg, int getstats)
	316	{
	317	nvlist_t config, nvroot;
	318	vdev_t *rvd = spa->spa_root_vdev;
	319	unsigned long hostid = 0;
b128c09f	320	boolean_t locked = B_FALSE;
428870ff	321	uint64_t split_guid;
34dc7c2f	322
b128c09f	323	if (vd == NULL) {
34dc7c2f	324	vd = rvd;
b128c09f BB	325	locked = B_TRUE;
	326	spa_config_enter(spa, SCL_CONFIG \| SCL_STATE, FTAG, RW_READER);
	327	}
	328
	329	ASSERT(spa_config_held(spa, SCL_CONFIG \| SCL_STATE, RW_READER) ==
	330	(SCL_CONFIG \| SCL_STATE));
34dc7c2f BB	331
	332	/*
	333	* If txg is -1, report the current value of spa->spa_config_txg.
	334	*/
	335	if (txg == -1ULL)
	336	txg = spa->spa_config_txg;
	337
b8d06fca	338	VERIFY(nvlist_alloc(&config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);
34dc7c2f BB	339
	340	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_VERSION,
	341	spa_version(spa)) == 0);
	342	VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME,
	343	spa_name(spa)) == 0);
	344	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE,
	345	spa_state(spa)) == 0);
	346	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG,
	347	txg) == 0);
	348	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID,
	349	spa_guid(spa)) == 0);
d96eb2b1 DM	350	VERIFY(spa->spa_comment == NULL \|\| nvlist_add_string(config,
	351	ZPOOL_CONFIG_COMMENT, spa->spa_comment) == 0);
	352
	353
d164b209 BB	354	#ifdef _KERNEL
	355	hostid = zone_get_hostid(NULL);
	356	#else /* _KERNEL */
	357	/*
	358	* We're emulating the system's hostid in userland, so we can't use
	359	* zone_get_hostid().
	360	*/
34dc7c2f	361	(void) ddi_strtoul(hw_serial, NULL, 10, &hostid);
d164b209	362	#endif /* _KERNEL */
34dc7c2f BB	363	if (hostid != 0) {
	364	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_HOSTID,
	365	hostid) == 0);
	366	}
	367	VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_HOSTNAME,
	368	utsname.nodename) == 0);
	369
	370	if (vd != rvd) {
	371	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_TOP_GUID,
	372	vd->vdev_top->vdev_guid) == 0);
	373	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_GUID,
	374	vd->vdev_guid) == 0);
	375	if (vd->vdev_isspare)
	376	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_IS_SPARE,
	377	1ULL) == 0);
	378	if (vd->vdev_islog)
	379	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_IS_LOG,
	380	1ULL) == 0);
	381	vd = vd->vdev_top; /* label contains top config */
428870ff BB	382	} else {
	383	/*
	384	* Only add the (potentially large) split information
	385	* in the mos config, and not in the vdev labels
	386	*/
	387	if (spa->spa_config_splitting != NULL)
	388	VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_SPLIT,
	389	spa->spa_config_splitting) == 0);
34dc7c2f BB	390	}
34dc7c2f BB	391
428870ff BB	392	/*
428870ff BB	393	* Add the top-level config. We even add this on pools which
572e2857	394	* don't support holes in the namespace.
428870ff BB	395	*/
	396	vdev_top_config_generate(spa, config);
	397
	398	/*
	399	* If we're splitting, record the original pool's guid.
	400	*/
	401	if (spa->spa_config_splitting != NULL &&
	402	nvlist_lookup_uint64(spa->spa_config_splitting,
	403	ZPOOL_CONFIG_SPLIT_GUID, &split_guid) == 0) {
	404	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_SPLIT_GUID,
	405	split_guid) == 0);
	406	}
	407
	408	nvroot = vdev_config_generate(spa, vd, getstats, 0);
34dc7c2f BB	409	VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, nvroot) == 0);
	410	nvlist_free(nvroot);
	411
9ae529ec CS	412	/*
	413	* Store what's necessary for reading the MOS in the label.
	414	*/
	415	VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_FEATURES_FOR_READ,
	416	spa->spa_label_features) == 0);
	417
428870ff BB	418	if (getstats && spa_load_state(spa) == SPA_LOAD_NONE) {
	419	ddt_histogram_t *ddh;
	420	ddt_stat_t *dds;
	421	ddt_object_t *ddo;
	422
b8d06fca	423	ddh = kmem_zalloc(sizeof (ddt_histogram_t), KM_PUSHPAGE);
428870ff BB	424	ddt_get_dedup_histogram(spa, ddh);
	425	VERIFY(nvlist_add_uint64_array(config,
	426	ZPOOL_CONFIG_DDT_HISTOGRAM,
	427	(uint64_t )ddh, sizeof (ddh) / sizeof (uint64_t)) == 0);
	428	kmem_free(ddh, sizeof (ddt_histogram_t));
	429
b8d06fca	430	ddo = kmem_zalloc(sizeof (ddt_object_t), KM_PUSHPAGE);
428870ff BB	431	ddt_get_dedup_object_stats(spa, ddo);
	432	VERIFY(nvlist_add_uint64_array(config,
	433	ZPOOL_CONFIG_DDT_OBJ_STATS,
	434	(uint64_t )ddo, sizeof (ddo) / sizeof (uint64_t)) == 0);
	435	kmem_free(ddo, sizeof (ddt_object_t));
	436
b8d06fca	437	dds = kmem_zalloc(sizeof (ddt_stat_t), KM_PUSHPAGE);
428870ff BB	438	ddt_get_dedup_stats(spa, dds);
	439	VERIFY(nvlist_add_uint64_array(config,
	440	ZPOOL_CONFIG_DDT_STATS,
	441	(uint64_t )dds, sizeof (dds) / sizeof (uint64_t)) == 0);
	442	kmem_free(dds, sizeof (ddt_stat_t));
	443	}
	444
b128c09f BB	445	if (locked)
	446	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	447
34dc7c2f BB	448	return (config);
	449	}
	450
34dc7c2f BB	451	/*
	452	* Update all disk labels, generate a fresh config based on the current
	453	* in-core state, and sync the global config cache (do not sync the config
	454	* cache if this is a booting rootpool).
	455	*/
	456	void
45d1cae3	457	spa_config_update(spa_t *spa, int what)
34dc7c2f BB	458	{
	459	vdev_t *rvd = spa->spa_root_vdev;
	460	uint64_t txg;
	461	int c;
	462
	463	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	464
b128c09f	465	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
34dc7c2f BB	466	txg = spa_last_synced_txg(spa) + 1;
	467	if (what == SPA_CONFIG_UPDATE_POOL) {
	468	vdev_config_dirty(rvd);
	469	} else {
	470	/*
	471	* If we have top-level vdevs that were added but have
	472	* not yet been prepared for allocation, do that now.
	473	* (It's safe now because the config cache is up to date,
	474	* so it will be able to translate the new DVAs.)
	475	* See comments in spa_vdev_add() for full details.
	476	*/
	477	for (c = 0; c < rvd->vdev_children; c++) {
	478	vdev_t *tvd = rvd->vdev_child[c];
9babb374 BB	479	if (tvd->vdev_ms_array == 0)
	480	vdev_metaslab_set_size(tvd);
	481	vdev_expand(tvd, txg);
34dc7c2f BB	482	}
34dc7c2f BB	483	}
b128c09f	484	spa_config_exit(spa, SCL_ALL, FTAG);
34dc7c2f BB	485
	486	/*
	487	* Wait for the mosconfig to be regenerated and synced.
	488	*/
	489	txg_wait_synced(spa->spa_dsl_pool, txg);
	490
	491	/*
	492	* Update the global config cache to reflect the new mosconfig.
	493	*/
45d1cae3	494	if (!spa->spa_is_root)
b128c09f	495	spa_config_sync(spa, B_FALSE, what != SPA_CONFIG_UPDATE_POOL);
34dc7c2f BB	496
34dc7c2f BB	497	if (what == SPA_CONFIG_UPDATE_POOL)
45d1cae3	498	spa_config_update(spa, SPA_CONFIG_UPDATE_VDEVS);
34dc7c2f	499	}
c28b2279 BB	500
	501	#if defined(_KERNEL) && defined(HAVE_SPL)
	502	EXPORT_SYMBOL(spa_config_sync);
	503	EXPORT_SYMBOL(spa_config_load);
	504	EXPORT_SYMBOL(spa_all_configs);
	505	EXPORT_SYMBOL(spa_config_set);
	506	EXPORT_SYMBOL(spa_config_generate);
	507	EXPORT_SYMBOL(spa_config_update);
	508
	509	module_param(spa_config_path, charp, 0444);
	510	MODULE_PARM_DESC(spa_config_path, "SPA config file (/etc/zfs/zpool.cache)");
	511	#endif