X-Git-Url: https://git.camperquake.de/gitweb.cgi?a=blobdiff_plain;f=module%2Fzfs%2Fspa.c;h=b610a0dae6bfd640dd479bba7f49b41eefb63173;hb=37f000c5aa76613e644cf3e5b1ec7bd2df6f7451;hp=606138a3eeef475b6ffc18257a2aacdd2524adda;hpb=bf701a83c5ec192be6d3afe87ebeee45ce9127f4;p=zfs.git

diff --git a/module/zfs/spa.c b/module/zfs/spa.c
index 606138a..b610a0d 100644
--- a/module/zfs/spa.c
+++ b/module/zfs/spa.c
@@ -21,6 +21,8 @@
 
 /*
  * 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.
  */
 
 /*
@@ -40,6 +42,7 @@
 #include <sys/zil.h>
 #include <sys/ddt.h>
 #include <sys/vdev_impl.h>
+#include <sys/vdev_disk.h>
 #include <sys/metaslab.h>
 #include <sys/metaslab_impl.h>
 #include <sys/uberblock_impl.h>
@@ -94,7 +97,7 @@ typedef struct zio_taskq_info {
 } zio_taskq_info_t;
 
 static const char *const zio_taskq_types[ZIO_TASKQ_TYPES] = {
-	"issue", "issue_high", "intr", "intr_high"
+	"iss", "iss_h", "int", "int_h"
 };
 
 /*
@@ -105,8 +108,8 @@ 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 },
 	{ ZTI_FIX(8),	ZTI_NULL,	ZTI_BATCH,	ZTI_NULL },
-	{ ZTI_BATCH,	ZTI_FIX(5),	ZTI_FIX(8),	ZTI_FIX(5) },
-	{ ZTI_FIX(100),	ZTI_NULL,	ZTI_ONE,	ZTI_NULL },
+	{ ZTI_BATCH,	ZTI_FIX(5),	ZTI_FIX(16),	ZTI_FIX(5) },
+	{ ZTI_PCT(100),	ZTI_NULL,	ZTI_ONE,	ZTI_NULL },
 	{ ZTI_ONE,	ZTI_NULL,	ZTI_ONE,	ZTI_NULL },
 	{ ZTI_ONE,	ZTI_NULL,	ZTI_ONE,	ZTI_NULL },
 };
@@ -147,7 +150,7 @@ spa_prop_add_list(nvlist_t *nvl, zpool_prop_t prop, char *strval,
 	const char *propname = zpool_prop_to_name(prop);
 	nvlist_t *propval;
 
-	VERIFY(nvlist_alloc(&propval, NV_UNIQUE_NAME, KM_SLEEP) == 0);
+	VERIFY(nvlist_alloc(&propval, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);
 	VERIFY(nvlist_add_uint64(propval, ZPROP_SOURCE, src) == 0);
 
 	if (strval != NULL)
@@ -165,15 +168,18 @@ spa_prop_add_list(nvlist_t *nvl, zpool_prop_t prop, char *strval,
 static void
 spa_prop_get_config(spa_t *spa, nvlist_t **nvp)
 {
+	vdev_t *rvd = spa->spa_root_vdev;
 	uint64_t size;
 	uint64_t alloc;
+	uint64_t space;
 	uint64_t cap, version;
 	zprop_source_t src = ZPROP_SRC_NONE;
 	spa_config_dirent_t *dp;
+	int c;
 
 	ASSERT(MUTEX_HELD(&spa->spa_props_lock));
 
-	if (spa->spa_root_vdev != NULL) {
+	if (rvd != NULL) {
 		alloc = metaslab_class_get_alloc(spa_normal_class(spa));
 		size = metaslab_class_get_space(spa_normal_class(spa));
 		spa_prop_add_list(*nvp, ZPOOL_PROP_NAME, spa_name(spa), 0, src);
@@ -181,6 +187,15 @@ spa_prop_get_config(spa_t *spa, nvlist_t **nvp)
 		spa_prop_add_list(*nvp, ZPOOL_PROP_ALLOCATED, NULL, alloc, src);
 		spa_prop_add_list(*nvp, ZPOOL_PROP_FREE, NULL,
 		    size - alloc, src);
+
+		space = 0;
+		for (c = 0; c < rvd->vdev_children; c++) {
+			vdev_t *tvd = rvd->vdev_child[c];
+			space += tvd->vdev_max_asize - tvd->vdev_asize;
+		}
+		spa_prop_add_list(*nvp, ZPOOL_PROP_EXPANDSZ, NULL, space,
+		    src);
+
 		spa_prop_add_list(*nvp, ZPOOL_PROP_READONLY, NULL,
 		    (spa_mode(spa) == FREAD), src);
 
@@ -191,7 +206,7 @@ spa_prop_get_config(spa_t *spa, nvlist_t **nvp)
 		    ddt_get_pool_dedup_ratio(spa), src);
 
 		spa_prop_add_list(*nvp, ZPOOL_PROP_HEALTH, NULL,
-		    spa->spa_root_vdev->vdev_state, src);
+		    rvd->vdev_state, src);
 
 		version = spa_version(spa);
 		if (version == zpool_prop_default_numeric(ZPOOL_PROP_VERSION))
@@ -203,6 +218,11 @@ spa_prop_get_config(spa_t *spa, nvlist_t **nvp)
 
 	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);
@@ -229,7 +249,9 @@ spa_prop_get(spa_t *spa, nvlist_t **nvp)
 	zap_attribute_t za;
 	int err;
 
-	VERIFY(nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_SLEEP) == 0);
+	err = nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_PUSHPAGE);
+	if (err)
+		return err;
 
 	mutex_enter(&spa->spa_props_lock);
 
@@ -241,7 +263,7 @@ spa_prop_get(spa_t *spa, nvlist_t **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);
+		goto out;
 	}
 
 	/*
@@ -279,7 +301,7 @@ spa_prop_get(spa_t *spa, nvlist_t **nvp)
 
 				strval = kmem_alloc(
 				    MAXNAMELEN + strlen(MOS_DIR_NAME) + 1,
-				    KM_SLEEP);
+				    KM_PUSHPAGE);
 				dsl_dataset_name(ds, strval);
 				dsl_dataset_rele(ds, FTAG);
 				rw_exit(&dp->dp_config_rwlock);
@@ -298,7 +320,7 @@ spa_prop_get(spa_t *spa, nvlist_t **nvp)
 
 		case 1:
 			/* string property */
-			strval = kmem_alloc(za.za_num_integers, KM_SLEEP);
+			strval = kmem_alloc(za.za_num_integers, KM_PUSHPAGE);
 			err = zap_lookup(mos, spa->spa_pool_props_object,
 			    za.za_name, 1, za.za_num_integers, strval);
 			if (err) {
@@ -342,7 +364,7 @@ spa_prop_validate(spa_t *spa, nvlist_t *props)
 		char *propname, *strval;
 		uint64_t intval;
 		objset_t *os;
-		char *slash;
+		char *slash, *check;
 
 		propname = nvpair_name(elem);
 
@@ -462,6 +484,20 @@ spa_prop_validate(spa_t *spa, nvlist_t *props)
 				error = EINVAL;
 			break;
 
+		case ZPOOL_PROP_COMMENT:
+			if ((error = nvpair_value_string(elem, &strval)) != 0)
+				break;
+			for (check = strval; *check != '\0'; check++) {
+				if (!isprint(*check)) {
+					error = EINVAL;
+					break;
+				}
+				check++;
+			}
+			if (strlen(strval) > ZPROP_MAX_COMMENT)
+				error = E2BIG;
+			break;
+
 		case ZPOOL_PROP_DEDUPDITTO:
 			if (spa_version(spa) < SPA_VERSION_DEDUP)
 				error = ENOTSUP;
@@ -504,7 +540,7 @@ spa_configfile_set(spa_t *spa, nvlist_t *nvp, boolean_t need_sync)
 		return;
 
 	dp = kmem_alloc(sizeof (spa_config_dirent_t),
-	    KM_SLEEP);
+	    KM_PUSHPAGE);
 
 	if (cachefile[0] == '\0')
 		dp->scd_path = spa_strdup(spa_config_path);
@@ -566,6 +602,43 @@ spa_prop_clear_bootfs(spa_t *spa, uint64_t dsobj, dmu_tx_t *tx)
 }
 
 /*
+ * 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)
+{
+	uint64_t	oldguid, newguid;
+	uint64_t	txg;
+
+	if (!(spa_mode_global & FWRITE))
+		return (EROFS);
+
+	txg = spa_vdev_enter(spa);
+
+	if (spa->spa_root_vdev->vdev_state != VDEV_STATE_HEALTHY)
+		return (spa_vdev_exit(spa, NULL, txg, ENXIO));
+
+	oldguid = spa_guid(spa);
+	newguid = spa_generate_guid(NULL);
+	ASSERT3U(oldguid, !=, newguid);
+
+	spa->spa_root_vdev->vdev_guid = newguid;
+	spa->spa_root_vdev->vdev_guid_sum += (newguid - oldguid);
+
+	vdev_config_dirty(spa->spa_root_vdev);
+
+	spa_event_notify(spa, NULL, FM_EREPORT_ZFS_POOL_REGUID);
+
+	return (spa_vdev_exit(spa, NULL, txg, 0));
+}
+
+/*
  * ==========================================================================
  * SPA state manipulation (open/create/destroy/import/export)
  * ==========================================================================
@@ -674,7 +747,7 @@ spa_create_zio_taskqs(spa_t *spa)
 	}
 }
 
-#ifdef _KERNEL
+#if defined(_KERNEL) && defined(HAVE_SPA_THREAD)
 static void
 spa_thread(void *arg)
 {
@@ -764,6 +837,7 @@ spa_activate(spa_t *spa, int mode)
 	ASSERT(spa->spa_proc == &p0);
 	spa->spa_did = 0;
 
+#ifdef HAVE_SPA_THREAD
 	/* 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,
@@ -784,6 +858,7 @@ spa_activate(spa_t *spa, int mode)
 #endif
 		}
 	}
+#endif /* HAVE_SPA_THREAD */
 	mutex_exit(&spa->spa_proc_lock);
 
 	/* If we didn't create a process, we need to create our taskqs. */
@@ -996,8 +1071,10 @@ spa_unload(spa_t *spa)
 	}
 	spa->spa_spares.sav_count = 0;
 
-	for (i = 0; i < spa->spa_l2cache.sav_count; i++)
+	for (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 *));
@@ -1011,6 +1088,11 @@ spa_unload(spa_t *spa)
 
 	spa->spa_async_suspended = 0;
 
+	if (spa->spa_comment != NULL) {
+		spa_strfree(spa->spa_comment);
+		spa->spa_comment = NULL;
+	}
+
 	spa_config_exit(spa, SCL_ALL, FTAG);
 }
 
@@ -1070,7 +1152,7 @@ spa_load_spares(spa_t *spa)
 	 * active configuration, then we also mark this vdev as an active spare.
 	 */
 	spa->spa_spares.sav_vdevs = kmem_alloc(nspares * sizeof (void *),
-	    KM_SLEEP);
+	    KM_PUSHPAGE);
 	for (i = 0; i < spa->spa_spares.sav_count; i++) {
 		VERIFY(spa_config_parse(spa, &vd, spares[i], NULL, 0,
 		    VDEV_ALLOC_SPARE) == 0);
@@ -1118,7 +1200,7 @@ spa_load_spares(spa_t *spa)
 	    DATA_TYPE_NVLIST_ARRAY) == 0);
 
 	spares = kmem_alloc(spa->spa_spares.sav_count * sizeof (void *),
-	    KM_SLEEP);
+	    KM_PUSHPAGE);
 	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);
@@ -1152,7 +1234,7 @@ spa_load_l2cache(spa_t *spa)
 	if (sav->sav_config != NULL) {
 		VERIFY(nvlist_lookup_nvlist_array(sav->sav_config,
 		    ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0);
-		newvdevs = kmem_alloc(nl2cache * sizeof (void *), KM_SLEEP);
+		newvdevs = kmem_alloc(nl2cache * sizeof (void *), KM_PUSHPAGE);
 	} else {
 		nl2cache = 0;
 	}
@@ -1220,11 +1302,13 @@ spa_load_l2cache(spa_t *spa)
 
 		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);
-			(void) vdev_close(vd);
-			spa_l2cache_remove(vd);
+			vdev_clear_stats(vd);
+			vdev_free(vd);
 		}
 	}
 
@@ -1244,7 +1328,7 @@ spa_load_l2cache(spa_t *spa)
 	VERIFY(nvlist_remove(sav->sav_config, ZPOOL_CONFIG_L2CACHE,
 	    DATA_TYPE_NVLIST_ARRAY) == 0);
 
-	l2cache = kmem_alloc(sav->sav_count * sizeof (void *), KM_SLEEP);
+	l2cache = kmem_alloc(sav->sav_count * sizeof (void *), KM_PUSHPAGE);
 	for (i = 0; i < sav->sav_count; i++)
 		l2cache[i] = vdev_config_generate(spa,
 		    sav->sav_vdevs[i], B_TRUE, VDEV_CONFIG_L2CACHE);
@@ -1266,11 +1350,14 @@ load_nvlist(spa_t *spa, uint64_t obj, nvlist_t **value)
 	int error;
 	*value = NULL;
 
-	VERIFY(0 == dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db));
+	error = dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db);
+	if (error)
+		return (error);
+
 	nvsize = *(uint64_t *)db->db_data;
 	dmu_buf_rele(db, FTAG);
 
-	packed = kmem_alloc(nvsize, KM_SLEEP);
+	packed = kmem_alloc(nvsize, KM_PUSHPAGE | KM_NODEBUG);
 	error = dmu_read(spa->spa_meta_objset, obj, 0, nvsize, packed,
 	    DMU_READ_PREFETCH);
 	if (error == 0)
@@ -1293,8 +1380,9 @@ spa_check_removed(vdev_t *vd)
 		spa_check_removed(vd->vdev_child[c]);
 
 	if (vd->vdev_ops->vdev_op_leaf && vdev_is_dead(vd)) {
-		zfs_post_autoreplace(vd->vdev_spa, vd);
-		spa_event_notify(vd->vdev_spa, vd, ESC_ZFS_VDEV_CHECK);
+		zfs_ereport_post(FM_EREPORT_RESOURCE_AUTOREPLACE,
+		    vd->vdev_spa, vd, NULL, 0, 0);
+		spa_event_notify(vd->vdev_spa, vd, FM_EREPORT_ZFS_DEVICE_CHECK);
 	}
 }
 
@@ -1325,8 +1413,8 @@ spa_config_valid(spa_t *spa, nvlist_t *config)
 		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);
+		    KM_PUSHPAGE);
+		VERIFY(nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);
 
 		for (c = 0; c < rvd->vdev_children; c++) {
 			vdev_t *tvd = rvd->vdev_child[c];
@@ -1681,7 +1769,7 @@ spa_try_repair(spa_t *spa, nvlist_t *config)
 	    &glist, &gcount) != 0)
 		return;
 
-	vd = kmem_zalloc(gcount * sizeof (vdev_t *), KM_SLEEP);
+	vd = kmem_zalloc(gcount * sizeof (vdev_t *), KM_PUSHPAGE);
 
 	/* attempt to online all the vdevs & validate */
 	attempt_reopen = B_TRUE;
@@ -1735,6 +1823,7 @@ spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type,
 {
 	nvlist_t *config = spa->spa_config;
 	char *ereport = FM_EREPORT_ZFS_POOL;
+	char *comment;
 	int error;
 	uint64_t pool_guid;
 	nvlist_t *nvl;
@@ -1742,6 +1831,10 @@ spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type,
 	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &pool_guid))
 		return (EINVAL);
 
+	ASSERT(spa->spa_comment == NULL);
+	if (nvlist_lookup_string(config, ZPOOL_CONFIG_COMMENT, &comment) == 0)
+		spa->spa_comment = spa_strdup(comment);
+
 	/*
 	 * Versioning wasn't explicitly added to the label until later, so if
 	 * it's not present treat it as the initial version.
@@ -1757,12 +1850,12 @@ spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type,
 	    spa_guid_exists(pool_guid, 0)) {
 		error = EEXIST;
 	} else {
-		spa->spa_load_guid = pool_guid;
+		spa->spa_config_guid = pool_guid;
 
 		if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT,
 		    &nvl) == 0) {
 			VERIFY(nvlist_dup(nvl, &spa->spa_config_splitting,
-			    KM_SLEEP) == 0);
+			    KM_PUSHPAGE) == 0);
 		}
 
 		gethrestime(&spa->spa_loaded_ts);
@@ -1870,7 +1963,7 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
 	 */
 	if (type != SPA_IMPORT_ASSEMBLE) {
 		spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
-		error = vdev_validate(rvd);
+		error = vdev_validate(rvd, mosconfig);
 		spa_config_exit(spa, SCL_ALL, FTAG);
 
 		if (error != 0)
@@ -1967,7 +2060,7 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
 				cmn_err(CE_WARN, "pool '%s' could not be "
 				    "loaded as it was last accessed by "
 				    "another system (host: %s hostid: 0x%lx). "
-				    "See: http://www.sun.com/msg/ZFS-8000-EY",
+				    "See: http://zfsonlinux.org/msg/ZFS-8000-EY",
 				    spa_name(spa), hostname,
 				    (unsigned long)hostid);
 				return (EBADF);
@@ -2419,7 +2512,7 @@ spa_open_common(const char *pool, spa_t **spapp, void *tag, nvlist_t *nvpolicy,
 			 */
 			if (config != NULL && spa->spa_config) {
 				VERIFY(nvlist_dup(spa->spa_config, config,
-				    KM_SLEEP) == 0);
+				    KM_PUSHPAGE) == 0);
 				VERIFY(nvlist_add_nvlist(*config,
 				    ZPOOL_CONFIG_LOAD_INFO,
 				    spa->spa_load_info) == 0);
@@ -2731,6 +2824,7 @@ spa_validate_aux_devs(spa_t *spa, nvlist_t *nvroot, uint64_t crtxg, int mode,
 		if ((strcmp(config, ZPOOL_CONFIG_L2CACHE) == 0) &&
 		    strcmp(vd->vdev_ops->vdev_op_type, VDEV_TYPE_DISK) != 0) {
 			error = ENOTBLK;
+			vdev_free(vd);
 			goto out;
 		}
 #endif
@@ -2794,13 +2888,13 @@ spa_set_aux_vdevs(spa_aux_vdev_t *sav, nvlist_t **devs, int ndevs,
 		    &olddevs, &oldndevs) == 0);
 
 		newdevs = kmem_alloc(sizeof (void *) *
-		    (ndevs + oldndevs), KM_SLEEP);
+		    (ndevs + oldndevs), KM_PUSHPAGE);
 		for (i = 0; i < oldndevs; i++)
 			VERIFY(nvlist_dup(olddevs[i], &newdevs[i],
-			    KM_SLEEP) == 0);
+			    KM_PUSHPAGE) == 0);
 		for (i = 0; i < ndevs; i++)
 			VERIFY(nvlist_dup(devs[i], &newdevs[i + oldndevs],
-			    KM_SLEEP) == 0);
+			    KM_PUSHPAGE) == 0);
 
 		VERIFY(nvlist_remove(sav->sav_config, config,
 		    DATA_TYPE_NVLIST_ARRAY) == 0);
@@ -2815,7 +2909,7 @@ spa_set_aux_vdevs(spa_aux_vdev_t *sav, nvlist_t **devs, int ndevs,
 		 * Generate a new dev list.
 		 */
 		VERIFY(nvlist_alloc(&sav->sav_config, NV_UNIQUE_NAME,
-		    KM_SLEEP) == 0);
+		    KM_PUSHPAGE) == 0);
 		VERIFY(nvlist_add_nvlist_array(sav->sav_config, config,
 		    devs, ndevs) == 0);
 	}
@@ -2840,10 +2934,6 @@ spa_l2cache_drop(spa_t *spa)
 		if (spa_l2cache_exists(vd->vdev_guid, &pool) &&
 		    pool != 0ULL && l2arc_vdev_present(vd))
 			l2arc_remove_vdev(vd);
-		if (vd->vdev_isl2cache)
-			spa_l2cache_remove(vd);
-		vdev_clear_stats(vd);
-		(void) vdev_close(vd);
 	}
 }
 
@@ -2945,7 +3035,7 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
 	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
 	    &spares, &nspares) == 0) {
 		VERIFY(nvlist_alloc(&spa->spa_spares.sav_config, NV_UNIQUE_NAME,
-		    KM_SLEEP) == 0);
+		    KM_PUSHPAGE) == 0);
 		VERIFY(nvlist_add_nvlist_array(spa->spa_spares.sav_config,
 		    ZPOOL_CONFIG_SPARES, spares, nspares) == 0);
 		spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
@@ -2960,7 +3050,7 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
 	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
 	    &l2cache, &nl2cache) == 0) {
 		VERIFY(nvlist_alloc(&spa->spa_l2cache.sav_config,
-		    NV_UNIQUE_NAME, KM_SLEEP) == 0);
+		    NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);
 		VERIFY(nvlist_add_nvlist_array(spa->spa_l2cache.sav_config,
 		    ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache) == 0);
 		spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
@@ -3098,7 +3188,7 @@ spa_generate_rootconf(char *devpath, char *devid, uint64_t *guid)
 	/*
 	 * Put this pool's top-level vdevs into a root vdev.
 	 */
-	VERIFY(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, KM_SLEEP) == 0);
+	VERIFY(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);
 	VERIFY(nvlist_add_string(nvroot, ZPOOL_CONFIG_TYPE,
 	    VDEV_TYPE_ROOT) == 0);
 	VERIFY(nvlist_add_uint64(nvroot, ZPOOL_CONFIG_ID, 0ULL) == 0);
@@ -3409,7 +3499,7 @@ spa_import(const char *pool, nvlist_t *config, nvlist_t *props, uint64_t flags)
 			    ZPOOL_CONFIG_SPARES, DATA_TYPE_NVLIST_ARRAY) == 0);
 		else
 			VERIFY(nvlist_alloc(&spa->spa_spares.sav_config,
-			    NV_UNIQUE_NAME, KM_SLEEP) == 0);
+			    NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);
 		VERIFY(nvlist_add_nvlist_array(spa->spa_spares.sav_config,
 		    ZPOOL_CONFIG_SPARES, spares, nspares) == 0);
 		spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
@@ -3424,7 +3514,7 @@ spa_import(const char *pool, nvlist_t *config, nvlist_t *props, uint64_t flags)
 			    ZPOOL_CONFIG_L2CACHE, DATA_TYPE_NVLIST_ARRAY) == 0);
 		else
 			VERIFY(nvlist_alloc(&spa->spa_l2cache.sav_config,
-			    NV_UNIQUE_NAME, KM_SLEEP) == 0);
+			    NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);
 		VERIFY(nvlist_add_nvlist_array(spa->spa_l2cache.sav_config,
 		    ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache) == 0);
 		spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
@@ -3507,7 +3597,7 @@ spa_tryimport(nvlist_t *tryconfig)
 		 * pools are bootable.
 		 */
 		if ((!error || error == EEXIST) && spa->spa_bootfs) {
-			char *tmpname = kmem_alloc(MAXPATHLEN, KM_SLEEP);
+			char *tmpname = kmem_alloc(MAXPATHLEN, KM_PUSHPAGE);
 
 			/*
 			 * We have to play games with the name since the
@@ -3516,7 +3606,7 @@ spa_tryimport(nvlist_t *tryconfig)
 			if (dsl_dsobj_to_dsname(spa_name(spa),
 			    spa->spa_bootfs, tmpname) == 0) {
 				char *cp;
-				char *dsname = kmem_alloc(MAXPATHLEN, KM_SLEEP);
+				char *dsname = kmem_alloc(MAXPATHLEN, KM_PUSHPAGE);
 
 				cp = strchr(tmpname, '/');
 				if (cp == NULL) {
@@ -3639,7 +3729,7 @@ spa_export_common(char *pool, int new_state, nvlist_t **oldconfig,
 		}
 	}
 
-	spa_event_notify(spa, NULL, ESC_ZFS_POOL_DESTROY);
+	spa_event_notify(spa, NULL, FM_EREPORT_ZFS_POOL_DESTROY);
 
 	if (spa->spa_state != POOL_STATE_UNINITIALIZED) {
 		spa_unload(spa);
@@ -3839,7 +3929,7 @@ spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing)
 	pvd = oldvd->vdev_parent;
 
 	if ((error = spa_config_parse(spa, &newrootvd, nvroot, NULL, 0,
-	    VDEV_ALLOC_ADD)) != 0)
+	    VDEV_ALLOC_ATTACH)) != 0)
 		return (spa_vdev_exit(spa, NULL, txg, EINVAL));
 
 	if (newrootvd->vdev_children != 1)
@@ -3921,7 +4011,7 @@ spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing)
 	if (strcmp(oldvd->vdev_path, newvd->vdev_path) == 0) {
 		spa_strfree(oldvd->vdev_path);
 		oldvd->vdev_path = kmem_alloc(strlen(newvd->vdev_path) + 5,
-		    KM_SLEEP);
+		    KM_PUSHPAGE);
 		(void) sprintf(oldvd->vdev_path, "%s/%s",
 		    newvd->vdev_path, "old");
 		if (oldvd->vdev_devid != NULL) {
@@ -3970,7 +4060,7 @@ spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing)
 
 	if (newvd->vdev_isspare) {
 		spa_spare_activate(newvd);
-		spa_event_notify(spa, newvd, ESC_ZFS_VDEV_SPARE);
+		spa_event_notify(spa, newvd, FM_EREPORT_ZFS_DEVICE_SPARE);
 	}
 
 	oldvdpath = spa_strdup(oldvd->vdev_path);
@@ -4002,7 +4092,7 @@ spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing)
 	spa_strfree(newvdpath);
 
 	if (spa->spa_bootfs)
-		spa_event_notify(spa, newvd, ESC_ZFS_BOOTFS_VDEV_ATTACH);
+		spa_event_notify(spa, newvd, FM_EREPORT_ZFS_BOOTFS_VDEV_ATTACH);
 
 	return (0);
 }
@@ -4203,7 +4293,7 @@ spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done)
 	vd->vdev_detached = B_TRUE;
 	vdev_dirty(tvd, VDD_DTL, vd, txg);
 
-	spa_event_notify(spa, vd, ESC_ZFS_VDEV_REMOVE);
+	spa_event_notify(spa, vd, FM_EREPORT_ZFS_DEVICE_REMOVE);
 
 	/* hang on to the spa before we release the lock */
 	spa_open_ref(spa, FTAG);
@@ -4316,8 +4406,8 @@ spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config,
 	    nvlist_lookup_nvlist(nvl, ZPOOL_CONFIG_L2CACHE, &tmp) == 0)
 		return (spa_vdev_exit(spa, NULL, txg, EINVAL));
 
-	vml = kmem_zalloc(children * sizeof (vdev_t *), KM_SLEEP);
-	glist = kmem_zalloc(children * sizeof (uint64_t), KM_SLEEP);
+	vml = kmem_zalloc(children * sizeof (vdev_t *), KM_PUSHPAGE);
+	glist = kmem_zalloc(children * sizeof (uint64_t), KM_PUSHPAGE);
 
 	/* then, loop over each vdev and validate it */
 	for (c = 0; c < children; c++) {
@@ -4397,7 +4487,7 @@ spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config,
 	 * Temporarily record the splitting vdevs in the spa config.  This
 	 * will disappear once the config is regenerated.
 	 */
-	VERIFY(nvlist_alloc(&nvl, NV_UNIQUE_NAME, KM_SLEEP) == 0);
+	VERIFY(nvlist_alloc(&nvl, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);
 	VERIFY(nvlist_add_uint64_array(nvl, ZPOOL_CONFIG_SPLIT_LIST,
 	    glist, children) == 0);
 	kmem_free(glist, children * sizeof (uint64_t));
@@ -4444,7 +4534,7 @@ spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config,
 	/* if that worked, generate a real config for the new pool */
 	if (newspa->spa_root_vdev != NULL) {
 		VERIFY(nvlist_alloc(&newspa->spa_config_splitting,
-		    NV_UNIQUE_NAME, KM_SLEEP) == 0);
+		    NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);
 		VERIFY(nvlist_add_uint64(newspa->spa_config_splitting,
 		    ZPOOL_CONFIG_SPLIT_GUID, spa_guid(spa)) == 0);
 		spa_config_set(newspa, spa_config_generate(newspa, NULL, -1ULL,
@@ -4556,12 +4646,12 @@ spa_vdev_remove_aux(nvlist_t *config, char *name, nvlist_t **dev, int count,
 	int i, j;
 
 	if (count > 1)
-		newdev = kmem_alloc((count - 1) * sizeof (void *), KM_SLEEP);
+		newdev = kmem_alloc((count - 1) * sizeof (void *), KM_PUSHPAGE);
 
 	for (i = 0, j = 0; i < count; i++) {
 		if (dev[i] == dev_to_remove)
 			continue;
-		VERIFY(nvlist_dup(dev[i], &newdev[j++], KM_SLEEP) == 0);
+		VERIFY(nvlist_dup(dev[i], &newdev[j++], KM_PUSHPAGE) == 0);
 	}
 
 	VERIFY(nvlist_remove(config, name, DATA_TYPE_NVLIST_ARRAY) == 0);
@@ -5034,9 +5124,6 @@ spa_async_probe(spa_t *spa, vdev_t *vd)
 static void
 spa_async_autoexpand(spa_t *spa, vdev_t *vd)
 {
-	sysevent_id_t eid;
-	nvlist_t *attr;
-	char *physpath;
 	int c;
 
 	if (!spa->spa_autoexpand)
@@ -5050,17 +5137,7 @@ spa_async_autoexpand(spa_t *spa, vdev_t *vd)
 	if (!vd->vdev_ops->vdev_op_leaf || vd->vdev_physpath == NULL)
 		return;
 
-	physpath = kmem_zalloc(MAXPATHLEN, KM_SLEEP);
-	(void) snprintf(physpath, MAXPATHLEN, "/devices%s", vd->vdev_physpath);
-
-	VERIFY(nvlist_alloc(&attr, NV_UNIQUE_NAME, KM_SLEEP) == 0);
-	VERIFY(nvlist_add_string(attr, DEV_PHYS_PATH, physpath) == 0);
-
-	(void) ddi_log_sysevent(zfs_dip, SUNW_VENDOR, EC_DEV_STATUS,
-	    ESC_DEV_DLE, attr, &eid, DDI_SLEEP);
-
-	nvlist_free(attr);
-	kmem_free(physpath, MAXPATHLEN);
+	spa_event_notify(vd->vdev_spa, vd, FM_EREPORT_ZFS_DEVICE_AUTOEXPAND);
 }
 
 static void
@@ -5229,15 +5306,15 @@ spa_sync_nvlist(spa_t *spa, uint64_t obj, nvlist_t *nv, dmu_tx_t *tx)
 	 * saves us a pre-read to get data we don't actually care about.
 	 */
 	bufsize = P2ROUNDUP(nvsize, SPA_CONFIG_BLOCKSIZE);
-	packed = kmem_alloc(bufsize, KM_SLEEP);
+	packed = vmem_alloc(bufsize, KM_PUSHPAGE);
 
 	VERIFY(nvlist_pack(nv, &packed, &nvsize, NV_ENCODE_XDR,
-	    KM_SLEEP) == 0);
+	    KM_PUSHPAGE) == 0);
 	bzero(packed + nvsize, bufsize - nvsize);
 
 	dmu_write(spa->spa_meta_objset, obj, 0, bufsize, packed, tx);
 
-	kmem_free(packed, bufsize);
+	vmem_free(packed, bufsize);
 
 	VERIFY(0 == dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db));
 	dmu_buf_will_dirty(db, tx);
@@ -5270,11 +5347,11 @@ spa_sync_aux_dev(spa_t *spa, spa_aux_vdev_t *sav, dmu_tx_t *tx,
 		    &sav->sav_object, tx) == 0);
 	}
 
-	VERIFY(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, KM_SLEEP) == 0);
+	VERIFY(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0);
 	if (sav->sav_count == 0) {
 		VERIFY(nvlist_add_nvlist_array(nvroot, config, NULL, 0) == 0);
 	} else {
-		list = kmem_alloc(sav->sav_count * sizeof (void *), KM_SLEEP);
+		list = kmem_alloc(sav->sav_count * sizeof (void *), KM_PUSHPAGE);
 		for (i = 0; i < sav->sav_count; i++)
 			list[i] = vdev_config_generate(spa, sav->sav_vdevs[i],
 			    B_FALSE, VDEV_CONFIG_L2CACHE);
@@ -5365,6 +5442,20 @@ spa_sync_props(void *arg1, void *arg2, dmu_tx_t *tx)
 			 * properties.
 			 */
 			break;
+		case ZPOOL_PROP_COMMENT:
+			VERIFY(nvpair_value_string(elem, &strval) == 0);
+			if (spa->spa_comment != NULL)
+				spa_strfree(spa->spa_comment);
+			spa->spa_comment = spa_strdup(strval);
+			/*
+			 * We need to dirty the configuration on all the vdevs
+			 * so that their labels get updated.  It's unnecessary
+			 * to do this for pool creation since the vdev's
+			 * configuratoin has already been dirtied.
+			 */
+			if (tx->tx_txg != TXG_INITIAL)
+				vdev_config_dirty(spa->spa_root_vdev);
+			break;
 		default:
 			/*
 			 * Set pool property values in the poolprops mos object.
@@ -5858,8 +5949,7 @@ spa_has_active_shared_spare(spa_t *spa)
 }
 
 /*
- * Post a sysevent corresponding to the given event.  The 'name' must be one of
- * the event definitions in sys/sysevent/eventdefs.h.  The payload will be
+ * Post a FM_EREPORT_ZFS_* event from sys/fm/fs/zfs.h.  The payload will be
  * filled in from the spa and (optionally) the vdev.  This doesn't do anything
  * in the userland libzpool, as we don't want consumers to misinterpret ztest
  * or zdb as real changes.
@@ -5868,49 +5958,65 @@ void
 spa_event_notify(spa_t *spa, vdev_t *vd, const char *name)
 {
 #ifdef _KERNEL
-	sysevent_t		*ev;
-	sysevent_attr_list_t	*attr = NULL;
-	sysevent_value_t	value;
-	sysevent_id_t		eid;
-
-	ev = sysevent_alloc(EC_ZFS, (char *)name, SUNW_KERN_PUB "zfs",
-	    SE_SLEEP);
-
-	value.value_type = SE_DATA_TYPE_STRING;
-	value.value.sv_string = spa_name(spa);
-	if (sysevent_add_attr(&attr, ZFS_EV_POOL_NAME, &value, SE_SLEEP) != 0)
-		goto done;
-
-	value.value_type = SE_DATA_TYPE_UINT64;
-	value.value.sv_uint64 = spa_guid(spa);
-	if (sysevent_add_attr(&attr, ZFS_EV_POOL_GUID, &value, SE_SLEEP) != 0)
-		goto done;
-
-	if (vd) {
-		value.value_type = SE_DATA_TYPE_UINT64;
-		value.value.sv_uint64 = vd->vdev_guid;
-		if (sysevent_add_attr(&attr, ZFS_EV_VDEV_GUID, &value,
-		    SE_SLEEP) != 0)
-			goto done;
-
-		if (vd->vdev_path) {
-			value.value_type = SE_DATA_TYPE_STRING;
-			value.value.sv_string = vd->vdev_path;
-			if (sysevent_add_attr(&attr, ZFS_EV_VDEV_PATH,
-			    &value, SE_SLEEP) != 0)
-				goto done;
-		}
-	}
-
-	if (sysevent_attach_attributes(ev, attr) != 0)
-		goto done;
-	attr = NULL;
-
-	(void) log_sysevent(ev, SE_SLEEP, &eid);
-
-done:
-	if (attr)
-		sysevent_free_attr(attr);
-	sysevent_free(ev);
+	zfs_ereport_post(name, spa, vd, NULL, 0, 0);
 #endif
 }
+
+#if defined(_KERNEL) && defined(HAVE_SPL)
+/* state manipulation functions */
+EXPORT_SYMBOL(spa_open);
+EXPORT_SYMBOL(spa_open_rewind);
+EXPORT_SYMBOL(spa_get_stats);
+EXPORT_SYMBOL(spa_create);
+EXPORT_SYMBOL(spa_import_rootpool);
+EXPORT_SYMBOL(spa_import);
+EXPORT_SYMBOL(spa_tryimport);
+EXPORT_SYMBOL(spa_destroy);
+EXPORT_SYMBOL(spa_export);
+EXPORT_SYMBOL(spa_reset);
+EXPORT_SYMBOL(spa_async_request);
+EXPORT_SYMBOL(spa_async_suspend);
+EXPORT_SYMBOL(spa_async_resume);
+EXPORT_SYMBOL(spa_inject_addref);
+EXPORT_SYMBOL(spa_inject_delref);
+EXPORT_SYMBOL(spa_scan_stat_init);
+EXPORT_SYMBOL(spa_scan_get_stats);
+
+/* device maniion */
+EXPORT_SYMBOL(spa_vdev_add);
+EXPORT_SYMBOL(spa_vdev_attach);
+EXPORT_SYMBOL(spa_vdev_detach);
+EXPORT_SYMBOL(spa_vdev_remove);
+EXPORT_SYMBOL(spa_vdev_setpath);
+EXPORT_SYMBOL(spa_vdev_setfru);
+EXPORT_SYMBOL(spa_vdev_split_mirror);
+
+/* spare statech is global across all pools) */
+EXPORT_SYMBOL(spa_spare_add);
+EXPORT_SYMBOL(spa_spare_remove);
+EXPORT_SYMBOL(spa_spare_exists);
+EXPORT_SYMBOL(spa_spare_activate);
+
+/* L2ARC statech is global across all pools) */
+EXPORT_SYMBOL(spa_l2cache_add);
+EXPORT_SYMBOL(spa_l2cache_remove);
+EXPORT_SYMBOL(spa_l2cache_exists);
+EXPORT_SYMBOL(spa_l2cache_activate);
+EXPORT_SYMBOL(spa_l2cache_drop);
+
+/* scanning */
+EXPORT_SYMBOL(spa_scan);
+EXPORT_SYMBOL(spa_scan_stop);
+
+/* spa syncing */
+EXPORT_SYMBOL(spa_sync); /* only for DMU use */
+EXPORT_SYMBOL(spa_sync_allpools);
+
+/* properties */
+EXPORT_SYMBOL(spa_prop_set);
+EXPORT_SYMBOL(spa_prop_get);
+EXPORT_SYMBOL(spa_prop_clear_bootfs);
+
+/* asynchronous event notification */
+EXPORT_SYMBOL(spa_event_notify);
+#endif