]> git.proxmox.com Git - mirror_zfs.git/blobdiff - module/zfs/spa.c
Fix hung z_zvol tasks during 'zfs receive'
[mirror_zfs.git] / module / zfs / spa.c
index f1f1444f1c94587a41213f7e1c06016ffa4bdf4a..4b6196cc3610105e80d4671a730cbdbb6853934a 100644 (file)
@@ -353,7 +353,7 @@ spa_prop_get(spa_t *spa, nvlist_t **nvp)
                zprop_source_t src = ZPROP_SRC_DEFAULT;
                zpool_prop_t prop;
 
-               if ((prop = zpool_name_to_prop(za.za_name)) == ZPROP_INVAL)
+               if ((prop = zpool_name_to_prop(za.za_name)) == ZPOOL_PROP_INVAL)
                        continue;
 
                switch (za.za_integer_length) {
@@ -440,8 +440,8 @@ spa_prop_validate(spa_t *spa, nvlist_t *props)
                const char *propname = nvpair_name(elem);
                zpool_prop_t prop = zpool_name_to_prop(propname);
 
-               switch ((int)prop) {
-               case ZPROP_INVAL:
+               switch (prop) {
+               case ZPOOL_PROP_INVAL:
                        if (!zpool_prop_feature(propname)) {
                                error = SET_ERROR(EINVAL);
                                break;
@@ -698,7 +698,7 @@ spa_prop_set(spa_t *spa, nvlist_t *nvp)
                    prop == ZPOOL_PROP_READONLY)
                        continue;
 
-               if (prop == ZPOOL_PROP_VERSION || prop == ZPROP_INVAL) {
+               if (prop == ZPOOL_PROP_VERSION || prop == ZPOOL_PROP_INVAL) {
                        uint64_t ver;
 
                        if (prop == ZPOOL_PROP_VERSION) {
@@ -757,10 +757,10 @@ spa_prop_clear_bootfs(spa_t *spa, uint64_t dsobj, dmu_tx_t *tx)
 static int
 spa_change_guid_check(void *arg, dmu_tx_t *tx)
 {
+       ASSERTV(uint64_t *newguid = arg);
        spa_t *spa = dmu_tx_pool(tx)->dp_spa;
        vdev_t *rvd = spa->spa_root_vdev;
        uint64_t vdev_state;
-       ASSERTV(uint64_t *newguid = arg);
 
        spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
        vdev_state = rvd->vdev_state;
@@ -875,7 +875,7 @@ spa_taskqs_init(spa_t *spa, zio_type_t t, zio_taskq_type_t q)
        uint_t count = ztip->zti_count;
        spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q];
        char name[32];
-       uint_t i, flags = 0;
+       uint_t flags = 0;
        boolean_t batch = B_FALSE;
 
        if (mode == ZTI_MODE_NULL) {
@@ -909,7 +909,7 @@ spa_taskqs_init(spa_t *spa, zio_type_t t, zio_taskq_type_t q)
                break;
        }
 
-       for (i = 0; i < count; i++) {
+       for (uint_t i = 0; i < count; i++) {
                taskq_t *tq;
 
                if (count > 1) {
@@ -950,14 +950,13 @@ static void
 spa_taskqs_fini(spa_t *spa, zio_type_t t, zio_taskq_type_t q)
 {
        spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q];
-       uint_t i;
 
        if (tqs->stqs_taskq == NULL) {
                ASSERT3U(tqs->stqs_count, ==, 0);
                return;
        }
 
-       for (i = 0; i < tqs->stqs_count; i++) {
+       for (uint_t i = 0; i < tqs->stqs_count; i++) {
                ASSERT3P(tqs->stqs_taskq[i], !=, NULL);
                taskq_destroy(tqs->stqs_taskq[i]);
        }
@@ -1019,15 +1018,18 @@ spa_taskq_dispatch_sync(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
 static void
 spa_create_zio_taskqs(spa_t *spa)
 {
-       int t, q;
-
-       for (t = 0; t < ZIO_TYPES; t++) {
-               for (q = 0; q < ZIO_TASKQ_TYPES; q++) {
+       for (int t = 0; t < ZIO_TYPES; t++) {
+               for (int q = 0; q < ZIO_TASKQ_TYPES; q++) {
                        spa_taskqs_init(spa, t, q);
                }
        }
 }
 
+/*
+ * Disabled until spa_thread() can be adapted for Linux.
+ */
+#undef HAVE_SPA_THREAD
+
 #if defined(_KERNEL) && defined(HAVE_SPA_THREAD)
 static void
 spa_thread(void *arg)
@@ -1164,6 +1166,8 @@ spa_activate(spa_t *spa, int mode)
            spa_error_entry_compare, sizeof (spa_error_entry_t),
            offsetof(spa_error_entry_t, se_avl));
 
+       spa_keystore_init(&spa->spa_keystore);
+
        /*
         * This taskq is used to perform zvol-minor-related tasks
         * asynchronously. This has several advantages, including easy
@@ -1182,9 +1186,17 @@ spa_activate(spa_t *spa, int mode)
        spa->spa_zvol_taskq = taskq_create("z_zvol", 1, defclsyspri,
            1, INT_MAX, 0);
 
+       /*
+        * Taskq dedicated to prefetcher threads: this is used to prevent the
+        * pool traverse code from monopolizing the global (and limited)
+        * system_taskq by inappropriately scheduling long running tasks on it.
+        */
+       spa->spa_prefetch_taskq = taskq_create("z_prefetch", boot_ncpus,
+           defclsyspri, 1, INT_MAX, TASKQ_DYNAMIC);
+
        /*
         * The taskq to upgrade datasets in this pool. Currently used by
-        * feature SPA_FEATURE_USEROBJ_ACCOUNTING.
+        * feature SPA_FEATURE_USEROBJ_ACCOUNTING/SPA_FEATURE_PROJECT_QUOTA.
         */
        spa->spa_upgrade_taskq = taskq_create("z_upgrade", boot_ncpus,
            defclsyspri, 1, INT_MAX, TASKQ_DYNAMIC);
@@ -1196,8 +1208,6 @@ spa_activate(spa_t *spa, int mode)
 static void
 spa_deactivate(spa_t *spa)
 {
-       int t, q;
-
        ASSERT(spa->spa_sync_on == B_FALSE);
        ASSERT(spa->spa_dsl_pool == NULL);
        ASSERT(spa->spa_root_vdev == NULL);
@@ -1211,6 +1221,11 @@ spa_deactivate(spa_t *spa)
                spa->spa_zvol_taskq = NULL;
        }
 
+       if (spa->spa_prefetch_taskq) {
+               taskq_destroy(spa->spa_prefetch_taskq);
+               spa->spa_prefetch_taskq = NULL;
+       }
+
        if (spa->spa_upgrade_taskq) {
                taskq_destroy(spa->spa_upgrade_taskq);
                spa->spa_upgrade_taskq = NULL;
@@ -1224,8 +1239,8 @@ spa_deactivate(spa_t *spa)
 
        taskq_cancel_id(system_delay_taskq, spa->spa_deadman_tqid);
 
-       for (t = 0; t < ZIO_TYPES; t++) {
-               for (q = 0; q < ZIO_TASKQ_TYPES; q++) {
+       for (int t = 0; t < ZIO_TYPES; t++) {
+               for (int q = 0; q < ZIO_TASKQ_TYPES; q++) {
                        spa_taskqs_fini(spa, t, q);
                }
        }
@@ -1241,10 +1256,11 @@ spa_deactivate(spa_t *spa)
         * still have errors left in the queues.  Empty them just in case.
         */
        spa_errlog_drain(spa);
-
        avl_destroy(&spa->spa_errlist_scrub);
        avl_destroy(&spa->spa_errlist_last);
 
+       spa_keystore_fini(&spa->spa_keystore);
+
        spa->spa_state = POOL_STATE_UNINITIALIZED;
 
        mutex_enter(&spa->spa_proc_lock);
@@ -1286,7 +1302,6 @@ spa_config_parse(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent,
        nvlist_t **child;
        uint_t children;
        int error;
-       int c;
 
        if ((error = vdev_alloc(spa, vdp, nv, parent, id, atype)) != 0)
                return (error);
@@ -1306,7 +1321,7 @@ spa_config_parse(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent,
                return (SET_ERROR(EINVAL));
        }
 
-       for (c = 0; c < children; c++) {
+       for (int c = 0; c < children; c++) {
                vdev_t *vd;
                if ((error = spa_config_parse(spa, &vd, child[c], *vdp, c,
                    atype)) != 0) {
@@ -1327,7 +1342,7 @@ spa_config_parse(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent,
 static void
 spa_unload(spa_t *spa)
 {
-       int i, c;
+       int i;
 
        ASSERT(MUTEX_HELD(&spa_namespace_lock));
 
@@ -1352,7 +1367,7 @@ spa_unload(spa_t *spa)
         */
        if (spa->spa_root_vdev != NULL) {
                spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
-               for (c = 0; c < spa->spa_root_vdev->vdev_children; c++)
+               for (int c = 0; c < spa->spa_root_vdev->vdev_children; c++)
                        vdev_metaslab_fini(spa->spa_root_vdev->vdev_child[c]);
                spa_config_exit(spa, SCL_ALL, FTAG);
        }
@@ -1364,7 +1379,7 @@ spa_unload(spa_t *spa)
         * Wait for any outstanding async I/O to complete.
         */
        if (spa->spa_async_zio_root != NULL) {
-               for (i = 0; i < max_ncpus; i++)
+               for (int i = 0; i < max_ncpus; i++)
                        (void) zio_wait(spa->spa_async_zio_root[i]);
                kmem_free(spa->spa_async_zio_root, max_ncpus * sizeof (void *));
                spa->spa_async_zio_root = NULL;
@@ -1561,7 +1576,7 @@ spa_load_spares(spa_t *spa)
 static void
 spa_load_l2cache(spa_t *spa)
 {
-       nvlist_t **l2cache;
+       nvlist_t **l2cache = NULL;
        uint_t nl2cache;
        int i, j, oldnvdevs;
        uint64_t guid;
@@ -1645,7 +1660,9 @@ 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);
+       if (sav->sav_count > 0)
+               l2cache = kmem_alloc(sav->sav_count * sizeof (void *),
+                   KM_SLEEP);
        for (i = 0; i < sav->sav_count; i++)
                l2cache[i] = vdev_config_generate(spa,
                    sav->sav_vdevs[i], B_TRUE, VDEV_CONFIG_L2CACHE);
@@ -1713,9 +1730,7 @@ load_nvlist(spa_t *spa, uint64_t obj, nvlist_t **value)
 static void
 spa_check_removed(vdev_t *vd)
 {
-       int c;
-
-       for (c = 0; c < vd->vdev_children; c++)
+       for (int c = 0; c < vd->vdev_children; c++)
                spa_check_removed(vd->vdev_child[c]);
 
        if (vd->vdev_ops->vdev_op_leaf && vdev_is_dead(vd) &&
@@ -1728,14 +1743,12 @@ spa_check_removed(vdev_t *vd)
 static void
 spa_config_valid_zaps(vdev_t *vd, vdev_t *mvd)
 {
-       uint64_t i;
-
        ASSERT3U(vd->vdev_children, ==, mvd->vdev_children);
 
        vd->vdev_top_zap = mvd->vdev_top_zap;
        vd->vdev_leaf_zap = mvd->vdev_leaf_zap;
 
-       for (i = 0; i < vd->vdev_children; i++) {
+       for (uint64_t i = 0; i < vd->vdev_children; i++) {
                spa_config_valid_zaps(vd->vdev_child[i], mvd->vdev_child[i]);
        }
 }
@@ -1748,7 +1761,6 @@ spa_config_valid(spa_t *spa, nvlist_t *config)
 {
        vdev_t *mrvd, *rvd = spa->spa_root_vdev;
        nvlist_t *nv;
-       int c, i;
 
        VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nv) == 0);
 
@@ -1770,7 +1782,7 @@ spa_config_valid(spa_t *spa, nvlist_t *config)
                    KM_SLEEP);
                VERIFY(nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_SLEEP) == 0);
 
-               for (c = 0; c < rvd->vdev_children; c++) {
+               for (int c = 0; c < rvd->vdev_children; c++) {
                        vdev_t *tvd = rvd->vdev_child[c];
                        vdev_t *mtvd  = mrvd->vdev_child[c];
 
@@ -1787,7 +1799,7 @@ spa_config_valid(spa_t *spa, nvlist_t *config)
                        VERIFY(nvlist_add_nvlist(spa->spa_load_info,
                            ZPOOL_CONFIG_MISSING_DEVICES, nv) == 0);
 
-                       for (i = 0; i < idx; i++)
+                       for (int i = 0; i < idx; i++)
                                nvlist_free(child[i]);
                }
                nvlist_free(nv);
@@ -1799,7 +1811,7 @@ spa_config_valid(spa_t *spa, nvlist_t *config)
         * from the MOS config (mrvd). Check each top-level vdev
         * with the corresponding MOS config top-level (mtvd).
         */
-       for (c = 0; c < rvd->vdev_children; c++) {
+       for (int c = 0; c < rvd->vdev_children; c++) {
                vdev_t *tvd = rvd->vdev_child[c];
                vdev_t *mtvd  = mrvd->vdev_child[c];
 
@@ -1901,14 +1913,13 @@ spa_passivate_log(spa_t *spa)
 {
        vdev_t *rvd = spa->spa_root_vdev;
        boolean_t slog_found = B_FALSE;
-       int c;
 
        ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER));
 
        if (!spa_has_slogs(spa))
                return (B_FALSE);
 
-       for (c = 0; c < rvd->vdev_children; c++) {
+       for (int c = 0; c < rvd->vdev_children; c++) {
                vdev_t *tvd = rvd->vdev_child[c];
                metaslab_group_t *mg = tvd->vdev_mg;
 
@@ -1925,11 +1936,10 @@ static void
 spa_activate_log(spa_t *spa)
 {
        vdev_t *rvd = spa->spa_root_vdev;
-       int c;
 
        ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER));
 
-       for (c = 0; c < rvd->vdev_children; c++) {
+       for (int c = 0; c < rvd->vdev_children; c++) {
                vdev_t *tvd = rvd->vdev_child[c];
                metaslab_group_t *mg = tvd->vdev_mg;
 
@@ -1959,9 +1969,7 @@ spa_offline_log(spa_t *spa)
 static void
 spa_aux_check_removed(spa_aux_vdev_t *sav)
 {
-       int i;
-
-       for (i = 0; i < sav->sav_count; i++)
+       for (int i = 0; i < sav->sav_count; i++)
                spa_check_removed(sav->sav_vdevs[i]);
 }
 
@@ -2003,7 +2011,7 @@ spa_load_verify_done(zio_t *zio)
        }
 
        mutex_enter(&spa->spa_scrub_lock);
-       spa->spa_scrub_inflight--;
+       spa->spa_load_verify_ios--;
        cv_broadcast(&spa->spa_scrub_io_cv);
        mutex_exit(&spa->spa_scrub_lock);
 }
@@ -2021,9 +2029,6 @@ static int
 spa_load_verify_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
     const zbookmark_phys_t *zb, const dnode_phys_t *dnp, void *arg)
 {
-       zio_t *rio;
-       size_t size;
-
        if (bp == NULL || BP_IS_HOLE(bp) || BP_IS_EMBEDDED(bp))
                return (0);
        /*
@@ -2036,13 +2041,13 @@ spa_load_verify_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
        if (!BP_IS_METADATA(bp) && !spa_load_verify_data)
                return (0);
 
-       rio = arg;
-       size = BP_GET_PSIZE(bp);
+       zio_t *rio = arg;
+       size_t size = BP_GET_PSIZE(bp);
 
        mutex_enter(&spa->spa_scrub_lock);
-       while (spa->spa_scrub_inflight >= spa_load_verify_maxinflight)
+       while (spa->spa_load_verify_ios >= spa_load_verify_maxinflight)
                cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock);
-       spa->spa_scrub_inflight++;
+       spa->spa_load_verify_ios++;
        mutex_exit(&spa->spa_scrub_lock);
 
        zio_nowait(zio_read(rio, spa, bp, abd_alloc_for_io(size, B_FALSE), size,
@@ -2089,8 +2094,8 @@ spa_load_verify(spa_t *spa)
 
        if (spa_load_verify_metadata) {
                error = traverse_pool(spa, spa->spa_verify_min_txg,
-                   TRAVERSE_PRE | TRAVERSE_PREFETCH_METADATA,
-                   spa_load_verify_cb, rio);
+                   TRAVERSE_PRE | TRAVERSE_PREFETCH_METADATA |
+                   TRAVERSE_NO_DECRYPT, spa_load_verify_cb, rio);
        }
 
        (void) zio_wait(rio);
@@ -2296,7 +2301,7 @@ spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type,
                        spa->spa_loaded_ts.tv_nsec = 0;
                }
                if (error != EBADF) {
-                       zfs_ereport_post(ereport, spa, NULL, NULL, 0, 0);
+                       zfs_ereport_post(ereport, spa, NULL, NULL, NULL, 0, 0);
                }
        }
        spa->spa_load_state = error ? SPA_LOAD_ERROR : SPA_LOAD_NONE;
@@ -2316,7 +2321,6 @@ vdev_count_verify_zaps(vdev_t *vd)
 {
        spa_t *spa = vd->vdev_spa;
        uint64_t total = 0;
-       uint64_t i;
 
        if (vd->vdev_top_zap != 0) {
                total++;
@@ -2329,7 +2333,7 @@ vdev_count_verify_zaps(vdev_t *vd)
                    spa->spa_all_vdev_zaps, vd->vdev_leaf_zap));
        }
 
-       for (i = 0; i < vd->vdev_children; i++) {
+       for (uint64_t i = 0; i < vd->vdev_children; i++) {
                total += vdev_count_verify_zaps(vd->vdev_child[i]);
        }
 
@@ -2341,7 +2345,8 @@ vdev_count_verify_zaps(vdev_t *vd)
  * Determine whether the activity check is required.
  */
 static boolean_t
-spa_activity_check_required(spa_t *spa, uberblock_t *ub, nvlist_t *config)
+spa_activity_check_required(spa_t *spa, uberblock_t *ub, nvlist_t *label,
+    nvlist_t *config)
 {
        uint64_t state = 0;
        uint64_t hostid = 0;
@@ -2358,7 +2363,6 @@ spa_activity_check_required(spa_t *spa, uberblock_t *ub, nvlist_t *config)
        }
 
        (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE, &state);
-       (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_HOSTID, &hostid);
 
        /*
         * Disable the MMP activity check - This is used by zdb which
@@ -2384,8 +2388,12 @@ spa_activity_check_required(spa_t *spa, uberblock_t *ub, nvlist_t *config)
 
        /*
         * Allow the activity check to be skipped when importing the pool
-        * on the same host which last imported it.
+        * on the same host which last imported it.  Since the hostid from
+        * configuration may be stale use the one read from the label.
         */
+       if (nvlist_exists(label, ZPOOL_CONFIG_HOSTID))
+               hostid = fnvlist_lookup_uint64(label, ZPOOL_CONFIG_HOSTID);
+
        if (hostid == spa_get_hostid())
                return (B_FALSE);
 
@@ -2545,11 +2553,10 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
        uberblock_t *ub = &spa->spa_uberblock;
        uint64_t children, config_cache_txg = spa->spa_config_txg;
        int orig_mode = spa->spa_mode;
-       int parse, i;
+       int parse;
        uint64_t obj;
        boolean_t missing_feat_write = B_FALSE;
        boolean_t activity_check = B_FALSE;
-       nvlist_t *mos_config;
 
        /*
         * If this is an untrusted config, access the pool in read-only mode.
@@ -2573,7 +2580,7 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
         */
        spa->spa_async_zio_root = kmem_alloc(max_ncpus * sizeof (void *),
            KM_SLEEP);
-       for (i = 0; i < max_ncpus; i++) {
+       for (int i = 0; i < max_ncpus; i++) {
                spa->spa_async_zio_root[i] = zio_root(spa, NULL, NULL,
                    ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE |
                    ZIO_FLAG_GODFATHER);
@@ -2651,7 +2658,7 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
         * pool is truly inactive and can be safely imported.  Prevent
         * hosts which don't have a hostid set from importing the pool.
         */
-       activity_check = spa_activity_check_required(spa, ub, config);
+       activity_check = spa_activity_check_required(spa, ub, label, config);
        if (activity_check) {
                if (ub->ub_mmp_magic == MMP_MAGIC && ub->ub_mmp_delay &&
                    spa_get_hostid() == 0) {
@@ -2712,13 +2719,12 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
         */
        if (ub->ub_version >= SPA_VERSION_FEATURES) {
                nvlist_t *unsup_feat;
-               nvpair_t *nvp;
 
                VERIFY(nvlist_alloc(&unsup_feat, NV_UNIQUE_NAME, KM_SLEEP) ==
                    0);
 
-               for (nvp = nvlist_next_nvpair(spa->spa_label_features, NULL);
-                   nvp != NULL;
+               for (nvpair_t *nvp = nvlist_next_nvpair(spa->spa_label_features,
+                   NULL); nvp != NULL;
                    nvp = nvlist_next_nvpair(spa->spa_label_features, nvp)) {
                        if (!zfeature_is_supported(nvpair_name(nvp))) {
                                VERIFY(nvlist_add_string(unsup_feat,
@@ -2780,7 +2786,6 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
        if (spa_version(spa) >= SPA_VERSION_FEATURES) {
                boolean_t missing_feat_read = B_FALSE;
                nvlist_t *unsup_feat, *enabled_feat;
-               spa_feature_t i;
 
                if (spa_dir_prop(spa, DMU_POOL_FEATURES_FOR_READ,
                    &spa->spa_feat_for_read_obj) != 0) {
@@ -2856,7 +2861,7 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
                 * Load refcounts for ZFS features from disk into an in-memory
                 * cache during SPA initialization.
                 */
-               for (i = 0; i < SPA_FEATURES; i++) {
+               for (spa_feature_t i = 0; i < SPA_FEATURES; i++) {
                        uint64_t refcount;
 
                        error = feature_get_refcount_from_disk(spa,
@@ -2981,6 +2986,7 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
         */
 
        /* The sentinel is only available in the MOS config. */
+       nvlist_t *mos_config;
        if (load_nvlist(spa, spa->spa_config_object, &mos_config) != 0)
                return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
 
@@ -3188,7 +3194,6 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
                dmu_tx_t *tx;
                int need_update = B_FALSE;
                dsl_pool_t *dp = spa_get_dsl(spa);
-               int c;
 
                ASSERT(state != SPA_LOAD_TRYIMPORT);
 
@@ -3235,7 +3240,7 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
                    (spa->spa_import_flags & ZFS_IMPORT_VERBATIM))
                        need_update = B_TRUE;
 
-               for (c = 0; c < rvd->vdev_children; c++)
+               for (int c = 0; c < rvd->vdev_children; c++)
                        if (rvd->vdev_child[c]->vdev_ms_array == 0)
                                need_update = B_TRUE;
 
@@ -3257,7 +3262,7 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
                 * Log the fact that we booted up (so that we can detect if
                 * we rebooted in the middle of an operation).
                 */
-               spa_history_log_version(spa, "open");
+               spa_history_log_version(spa, "open", NULL);
 
                /*
                 * Delete any inconsistent datasets.
@@ -3415,7 +3420,7 @@ spa_open_common(const char *pool, spa_t **spapp, void *tag, nvlist_t *nvpolicy,
         * up calling spa_open() again.  The real fix is to figure out how to
         * avoid dsl_dir_open() calling this in the first place.
         */
-       if (mutex_owner(&spa_namespace_lock) != curthread) {
+       if (MUTEX_NOT_HELD(&spa_namespace_lock)) {
                mutex_enter(&spa_namespace_lock);
                locked = B_TRUE;
        }
@@ -3774,10 +3779,14 @@ spa_get_stats(const char *name, nvlist_t **config,
                            ZPOOL_CONFIG_ERRCOUNT,
                            spa_get_errlog_size(spa)) == 0);
 
-                       if (spa_suspended(spa))
+                       if (spa_suspended(spa)) {
                                VERIFY(nvlist_add_uint64(*config,
                                    ZPOOL_CONFIG_SUSPENDED,
                                    spa->spa_failmode) == 0);
+                               VERIFY(nvlist_add_uint64(*config,
+                                   ZPOOL_CONFIG_SUSPENDED_REASON,
+                                   spa->spa_suspended) == 0);
+                       }
 
                        spa_add_spares(spa, *config);
                        spa_add_l2cache(spa, *config);
@@ -3973,12 +3982,28 @@ spa_l2cache_drop(spa_t *spa)
        }
 }
 
+/*
+ * Verify encryption parameters for spa creation. If we are encrypting, we must
+ * have the encryption feature flag enabled.
+ */
+static int
+spa_create_check_encryption_params(dsl_crypto_params_t *dcp,
+    boolean_t has_encryption)
+{
+       if (dcp->cp_crypt != ZIO_CRYPT_OFF &&
+           dcp->cp_crypt != ZIO_CRYPT_INHERIT &&
+           !has_encryption)
+               return (SET_ERROR(ENOTSUP));
+
+       return (dmu_objset_create_crypt_check(NULL, dcp));
+}
+
 /*
  * Pool Creation
  */
 int
 spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
-    nvlist_t *zplprops)
+    nvlist_t *zplprops, dsl_crypto_params_t *dcp)
 {
        spa_t *spa;
        char *altroot = NULL;
@@ -3989,10 +4014,11 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
        uint64_t txg = TXG_INITIAL;
        nvlist_t **spares, **l2cache;
        uint_t nspares, nl2cache;
-       uint64_t version, obj;
+       uint64_t version, obj, root_dsobj = 0;
        boolean_t has_features;
-       nvpair_t *elem;
-       int c, i;
+       boolean_t has_encryption;
+       spa_feature_t feat;
+       char *feat_name;
        char *poolname;
        nvlist_t *nvl;
 
@@ -4033,10 +4059,28 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
                spa->spa_import_flags |= ZFS_IMPORT_TEMP_NAME;
 
        has_features = B_FALSE;
-       for (elem = nvlist_next_nvpair(props, NULL);
+       has_encryption = B_FALSE;
+       for (nvpair_t *elem = nvlist_next_nvpair(props, NULL);
            elem != NULL; elem = nvlist_next_nvpair(props, elem)) {
-               if (zpool_prop_feature(nvpair_name(elem)))
+               if (zpool_prop_feature(nvpair_name(elem))) {
                        has_features = B_TRUE;
+
+                       feat_name = strchr(nvpair_name(elem), '@') + 1;
+                       VERIFY0(zfeature_lookup_name(feat_name, &feat));
+                       if (feat == SPA_FEATURE_ENCRYPTION)
+                               has_encryption = B_TRUE;
+               }
+       }
+
+       /* verify encryption params, if they were provided */
+       if (dcp != NULL) {
+               error = spa_create_check_encryption_params(dcp, has_encryption);
+               if (error != 0) {
+                       spa_deactivate(spa);
+                       spa_remove(spa);
+                       mutex_exit(&spa_namespace_lock);
+                       return (error);
+               }
        }
 
        if (has_features || nvlist_lookup_uint64(props,
@@ -4056,7 +4100,7 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
         */
        spa->spa_async_zio_root = kmem_alloc(max_ncpus * sizeof (void *),
            KM_SLEEP);
-       for (i = 0; i < max_ncpus; i++) {
+       for (int i = 0; i < max_ncpus; i++) {
                spa->spa_async_zio_root[i] = zio_root(spa, NULL, NULL,
                    ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE |
                    ZIO_FLAG_GODFATHER);
@@ -4079,7 +4123,7 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
            (error = vdev_create(rvd, txg, B_FALSE)) == 0 &&
            (error = spa_validate_aux(spa, nvroot, txg,
            VDEV_ALLOC_ADD)) == 0) {
-               for (c = 0; c < rvd->vdev_children; c++) {
+               for (int c = 0; c < rvd->vdev_children; c++) {
                        vdev_metaslab_set_size(rvd->vdev_child[c]);
                        vdev_expand(rvd->vdev_child[c], txg);
                }
@@ -4126,8 +4170,7 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
        }
 
        spa->spa_is_initializing = B_TRUE;
-       spa->spa_dsl_pool = dp = dsl_pool_create(spa, zplprops, txg);
-       spa->spa_meta_objset = dp->dp_meta_objset;
+       spa->spa_dsl_pool = dp = dsl_pool_create(spa, zplprops, dcp, txg);
        spa->spa_is_initializing = B_FALSE;
 
        /*
@@ -4139,6 +4182,15 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
 
        tx = dmu_tx_create_assigned(dp, txg);
 
+       /*
+        * Create the pool's history object.
+        */
+       if (version >= SPA_VERSION_ZPOOL_HISTORY && !spa->spa_history)
+               spa_history_create_obj(spa, tx);
+
+       spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_CREATE);
+       spa_history_log_version(spa, "create", tx);
+
        /*
         * Create the pool config object.
         */
@@ -4152,9 +4204,6 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
                cmn_err(CE_PANIC, "failed to add pool config");
        }
 
-       if (spa_version(spa) >= SPA_VERSION_FEATURES)
-               spa_feature_create_zap_objects(spa, tx);
-
        if (zap_add(spa->spa_meta_objset,
            DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CREATION_VERSION,
            sizeof (uint64_t), 1, &version, tx) != 0) {
@@ -4187,12 +4236,6 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
        VERIFY3U(0, ==, bpobj_open(&spa->spa_deferred_bpobj,
            spa->spa_meta_objset, obj));
 
-       /*
-        * Create the pool's history object.
-        */
-       if (version >= SPA_VERSION_ZPOOL_HISTORY)
-               spa_history_create_obj(spa, tx);
-
        /*
         * Generate some random noise for salted checksums to operate on.
         */
@@ -4215,20 +4258,28 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
 
        dmu_tx_commit(tx);
 
+       /*
+        * If the root dataset is encrypted we will need to create key mappings
+        * for the zio layer before we start to write any data to disk and hold
+        * them until after the first txg has been synced. Waiting for the first
+        * transaction to complete also ensures that our bean counters are
+        * appropriately updated.
+        */
+       if (dp->dp_root_dir->dd_crypto_obj != 0) {
+               root_dsobj = dsl_dir_phys(dp->dp_root_dir)->dd_head_dataset_obj;
+               VERIFY0(spa_keystore_create_mapping_impl(spa, root_dsobj,
+                   dp->dp_root_dir, FTAG));
+       }
+
        spa->spa_sync_on = B_TRUE;
-       txg_sync_start(spa->spa_dsl_pool);
+       txg_sync_start(dp);
        mmp_thread_start(spa);
+       txg_wait_synced(dp, txg);
 
-       /*
-        * We explicitly wait for the first transaction to complete so that our
-        * bean counters are appropriately updated.
-        */
-       txg_wait_synced(spa->spa_dsl_pool, txg);
+       if (dp->dp_root_dir->dd_crypto_obj != 0)
+               VERIFY0(spa_keystore_remove_mapping(spa, root_dsobj, FTAG));
 
        spa_config_sync(spa, B_FALSE, B_TRUE);
-       spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_CREATE);
-
-       spa_history_log_version(spa, "create");
 
        /*
         * Don't count references from objsets that are already closed
@@ -4415,7 +4466,7 @@ spa_import(char *pool, nvlist_t *config, nvlist_t *props, uint64_t flags)
         */
        spa_async_request(spa, SPA_ASYNC_AUTOEXPAND);
 
-       spa_history_log_version(spa, "import");
+       spa_history_log_version(spa, "import", NULL);
 
        spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_IMPORT);
 
@@ -4618,7 +4669,10 @@ spa_export_common(char *pool, int new_state, nvlist_t **oldconfig,
        }
 
 export_spa:
-       spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_DESTROY);
+       if (new_state == POOL_STATE_DESTROYED)
+               spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_DESTROY);
+       else if (new_state == POOL_STATE_EXPORTED)
+               spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_EXPORT);
 
        if (spa->spa_state != POOL_STATE_UNINITIALIZED) {
                spa_unload(spa);
@@ -4688,7 +4742,6 @@ spa_vdev_add(spa_t *spa, nvlist_t *nvroot)
        vdev_t *vd, *tvd;
        nvlist_t **spares, **l2cache;
        uint_t nspares, nl2cache;
-       int c;
 
        ASSERT(spa_writeable(spa));
 
@@ -4725,7 +4778,7 @@ spa_vdev_add(spa_t *spa, nvlist_t *nvroot)
        /*
         * Transfer each new top-level vdev from vd to rvd.
         */
-       for (c = 0; c < vd->vdev_children; c++) {
+       for (int c = 0; c < vd->vdev_children; c++) {
 
                /*
                 * Set the vdev id to the first hole, if one exists.
@@ -4797,12 +4850,12 @@ int
 spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing)
 {
        uint64_t txg, dtl_max_txg;
+       ASSERTV(vdev_t *rvd = spa->spa_root_vdev);
        vdev_t *oldvd, *newvd, *newrootvd, *pvd, *tvd;
        vdev_ops_t *pvops;
        char *oldvdpath, *newvdpath;
        int newvd_isspare;
        int error;
-       ASSERTV(vdev_t *rvd = spa->spa_root_vdev);
 
        ASSERT(spa_writeable(spa));
 
@@ -5007,12 +5060,12 @@ spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done)
 {
        uint64_t txg;
        int error;
+       ASSERTV(vdev_t *rvd = spa->spa_root_vdev);
        vdev_t *vd, *pvd, *cvd, *tvd;
        boolean_t unspare = B_FALSE;
        uint64_t unspare_guid = 0;
        char *vdpath;
-       int c, t;
-       ASSERTV(vdev_t *rvd = spa->spa_root_vdev);
+
        ASSERT(spa_writeable(spa));
 
        txg = spa_vdev_enter(spa);
@@ -5079,7 +5132,7 @@ spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done)
            vd->vdev_path != NULL) {
                size_t len = strlen(vd->vdev_path);
 
-               for (c = 0; c < pvd->vdev_children; c++) {
+               for (int c = 0; c < pvd->vdev_children; c++) {
                        cvd = pvd->vdev_child[c];
 
                        if (cvd == vd || cvd->vdev_path == NULL)
@@ -5186,7 +5239,7 @@ spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done)
         * prevent vd from being accessed after it's freed.
         */
        vdpath = spa_strdup(vd->vdev_path ? vd->vdev_path : "none");
-       for (t = 0; t < TXG_SIZE; t++)
+       for (int t = 0; t < TXG_SIZE; t++)
                (void) txg_list_remove_this(&tvd->vdev_dtl_list, vd, t);
        vd->vdev_detached = B_TRUE;
        vdev_dirty(tvd, VDD_DTL, vd, txg);
@@ -5534,9 +5587,7 @@ out:
 static nvlist_t *
 spa_nvlist_lookup_by_guid(nvlist_t **nvpp, int count, uint64_t target_guid)
 {
-       int i;
-
-       for (i = 0; i < count; i++) {
+       for (int i = 0; i < count; i++) {
                uint64_t guid;
 
                VERIFY(nvlist_lookup_uint64(nvpp[i], ZPOOL_CONFIG_GUID,
@@ -5554,12 +5605,11 @@ spa_vdev_remove_aux(nvlist_t *config, char *name, nvlist_t **dev, int count,
     nvlist_t *dev_to_remove)
 {
        nvlist_t **newdev = NULL;
-       int i, j;
 
        if (count > 1)
                newdev = kmem_alloc((count - 1) * sizeof (void *), KM_SLEEP);
 
-       for (i = 0, j = 0; i < count; i++) {
+       for (int i = 0, j = 0; i < count; i++) {
                if (dev[i] == dev_to_remove)
                        continue;
                VERIFY(nvlist_dup(dev[i], &newdev[j++], KM_SLEEP) == 0);
@@ -5568,7 +5618,7 @@ spa_vdev_remove_aux(nvlist_t *config, char *name, nvlist_t **dev, int count,
        VERIFY(nvlist_remove(config, name, DATA_TYPE_NVLIST_ARRAY) == 0);
        VERIFY(nvlist_add_nvlist_array(config, name, newdev, count - 1) == 0);
 
-       for (i = 0; i < count - 1; i++)
+       for (int i = 0; i < count - 1; i++)
                nvlist_free(newdev[i]);
 
        if (count > 1)
@@ -5793,9 +5843,8 @@ static vdev_t *
 spa_vdev_resilver_done_hunt(vdev_t *vd)
 {
        vdev_t *newvd, *oldvd;
-       int c;
 
-       for (c = 0; c < vd->vdev_children; c++) {
+       for (int c = 0; c < vd->vdev_children; c++) {
                oldvd = spa_vdev_resilver_done_hunt(vd->vdev_child[c]);
                if (oldvd != NULL)
                        return (oldvd);
@@ -6010,8 +6059,6 @@ spa_scan(spa_t *spa, pool_scan_func_t func)
 static void
 spa_async_remove(spa_t *spa, vdev_t *vd)
 {
-       int c;
-
        if (vd->vdev_remove_wanted) {
                vd->vdev_remove_wanted = B_FALSE;
                vd->vdev_delayed_close = B_FALSE;
@@ -6030,33 +6077,29 @@ spa_async_remove(spa_t *spa, vdev_t *vd)
                vdev_state_dirty(vd->vdev_top);
        }
 
-       for (c = 0; c < vd->vdev_children; c++)
+       for (int c = 0; c < vd->vdev_children; c++)
                spa_async_remove(spa, vd->vdev_child[c]);
 }
 
 static void
 spa_async_probe(spa_t *spa, vdev_t *vd)
 {
-       int c;
-
        if (vd->vdev_probe_wanted) {
                vd->vdev_probe_wanted = B_FALSE;
                vdev_reopen(vd);        /* vdev_open() does the actual probe */
        }
 
-       for (c = 0; c < vd->vdev_children; c++)
+       for (int c = 0; c < vd->vdev_children; c++)
                spa_async_probe(spa, vd->vdev_child[c]);
 }
 
 static void
 spa_async_autoexpand(spa_t *spa, vdev_t *vd)
 {
-       int c;
-
        if (!spa->spa_autoexpand)
                return;
 
-       for (c = 0; c < vd->vdev_children; c++) {
+       for (int c = 0; c < vd->vdev_children; c++) {
                vdev_t *cvd = vd->vdev_child[c];
                spa_async_autoexpand(spa, cvd);
        }
@@ -6068,9 +6111,10 @@ spa_async_autoexpand(spa_t *spa, vdev_t *vd)
 }
 
 static void
-spa_async_thread(spa_t *spa)
+spa_async_thread(void *arg)
 {
-       int tasks, i;
+       spa_t *spa = (spa_t *)arg;
+       int tasks;
 
        ASSERT(spa->spa_sync_on);
 
@@ -6108,9 +6152,9 @@ spa_async_thread(spa_t *spa)
        if (tasks & SPA_ASYNC_REMOVE) {
                spa_vdev_state_enter(spa, SCL_NONE);
                spa_async_remove(spa, spa->spa_root_vdev);
-               for (i = 0; i < spa->spa_l2cache.sav_count; i++)
+               for (int i = 0; i < spa->spa_l2cache.sav_count; i++)
                        spa_async_remove(spa, spa->spa_l2cache.sav_vdevs[i]);
-               for (i = 0; i < spa->spa_spares.sav_count; i++)
+               for (int i = 0; i < spa->spa_spares.sav_count; i++)
                        spa_async_remove(spa, spa->spa_spares.sav_vdevs[i]);
                (void) spa_vdev_state_exit(spa, NULL, 0);
        }
@@ -6348,7 +6392,6 @@ static void
 spa_avz_build(vdev_t *vd, uint64_t avz, dmu_tx_t *tx)
 {
        spa_t *spa = vd->vdev_spa;
-       uint64_t i;
 
        if (vd->vdev_top_zap != 0) {
                VERIFY0(zap_add_int(spa->spa_meta_objset, avz,
@@ -6358,7 +6401,7 @@ spa_avz_build(vdev_t *vd, uint64_t avz, dmu_tx_t *tx)
                VERIFY0(zap_add_int(spa->spa_meta_objset, avz,
                    vd->vdev_leaf_zap, tx));
        }
-       for (i = 0; i < vd->vdev_children; i++) {
+       for (uint64_t i = 0; i < vd->vdev_children; i++) {
                spa_avz_build(vd->vdev_child[i], avz, tx);
        }
 }
@@ -6385,15 +6428,15 @@ spa_sync_config_object(spa_t *spa, dmu_tx_t *tx)
            spa->spa_all_vdev_zaps != 0);
 
        if (spa->spa_avz_action == AVZ_ACTION_REBUILD) {
-               zap_cursor_t zc;
-               zap_attribute_t za;
-
                /* Make and build the new AVZ */
                uint64_t new_avz = zap_create(spa->spa_meta_objset,
                    DMU_OTN_ZAP_METADATA, DMU_OT_NONE, 0, tx);
                spa_avz_build(spa->spa_root_vdev, new_avz, tx);
 
                /* Diff old AVZ with new one */
+               zap_cursor_t zc;
+               zap_attribute_t za;
+
                for (zap_cursor_init(&zc, spa->spa_meta_objset,
                    spa->spa_all_vdev_zaps);
                    zap_cursor_retrieve(&zc, &za) == 0;
@@ -6515,9 +6558,8 @@ spa_sync_props(void *arg, dmu_tx_t *tx)
                zprop_type_t proptype;
                spa_feature_t fid;
 
-               prop = zpool_name_to_prop(nvpair_name(elem));
-               switch ((int)prop) {
-               case ZPROP_INVAL:
+               switch (prop = zpool_name_to_prop(nvpair_name(elem))) {
+               case ZPOOL_PROP_INVAL:
                        /*
                         * We checked this earlier in spa_prop_validate().
                         */
@@ -6727,15 +6769,12 @@ spa_sync(spa_t *spa, uint64_t txg)
        dsl_pool_t *dp = spa->spa_dsl_pool;
        objset_t *mos = spa->spa_meta_objset;
        bplist_t *free_bpl = &spa->spa_free_bplist[txg & TXG_MASK];
-       metaslab_class_t *mc;
        vdev_t *rvd = spa->spa_root_vdev;
        vdev_t *vd;
        dmu_tx_t *tx;
        int error;
        uint32_t max_queue_depth = zfs_vdev_async_write_max_active *
            zfs_vdev_queue_depth_pct / 100;
-       uint64_t queue_depth_total;
-       int c;
 
        VERIFY(spa_writeable(spa));
 
@@ -6810,8 +6849,8 @@ spa_sync(spa_t *spa, uint64_t txg)
         * The max queue depth will not change in the middle of syncing
         * out this txg.
         */
-       queue_depth_total = 0;
-       for (c = 0; c < rvd->vdev_children; c++) {
+       uint64_t queue_depth_total = 0;
+       for (int c = 0; c < rvd->vdev_children; c++) {
                vdev_t *tvd = rvd->vdev_child[c];
                metaslab_group_t *mg = tvd->vdev_mg;
 
@@ -6828,7 +6867,7 @@ spa_sync(spa_t *spa, uint64_t txg)
                mg->mg_max_alloc_queue_depth = max_queue_depth;
                queue_depth_total += mg->mg_max_alloc_queue_depth;
        }
-       mc = spa_normal_class(spa);
+       metaslab_class_t *mc = spa_normal_class(spa);
        ASSERT0(refcount_count(&mc->mc_alloc_slots));
        mc->mc_alloc_max_slots = queue_depth_total;
        mc->mc_alloc_throttle_enabled = zio_dva_throttle_enabled;
@@ -6941,7 +6980,7 @@ spa_sync(spa_t *spa, uint64_t txg)
                        int children = rvd->vdev_children;
                        int c0 = spa_get_random(children);
 
-                       for (c = 0; c < children; c++) {
+                       for (int c = 0; c < children; c++) {
                                vd = rvd->vdev_child[(c0 + c) % children];
                                if (vd->vdev_ms_array == 0 || vd->vdev_islog)
                                        continue;
@@ -6962,7 +7001,7 @@ spa_sync(spa_t *spa, uint64_t txg)
 
                if (error == 0)
                        break;
-               zio_suspend(spa, NULL);
+               zio_suspend(spa, NULL, ZIO_SUSPEND_IOERR);
                zio_resume_wait(spa);
        }
        dmu_tx_commit(tx);