/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2011, 2017 by Delphix. All rights reserved.
- * Copyright (c) 2015, Nexenta Systems, Inc. All rights reserved.
- * Copyright (c) 2013, 2014, Nexenta Systems, Inc. All rights reserved.
+ * Copyright (c) 2018, Nexenta Systems, Inc. All rights reserved.
* Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
* Copyright 2013 Saso Kiselkov. All rights reserved.
* Copyright (c) 2014 Integros [integros.com]
* Copyright (c) 2016 Actifio, Inc. All rights reserved.
* Copyright (c) 2017 Datto Inc.
* Copyright 2017 Joyent, Inc.
+ * Copyright (c) 2017, Intel Corporation.
*/
/*
#ifdef _KERNEL
#include <sys/fm/protocol.h>
#include <sys/fm/util.h>
-#include <sys/bootprops.h>
#include <sys/callb.h>
-#include <sys/cpupart.h>
-#include <sys/pool.h>
-#include <sys/sysdc.h>
#include <sys/zone.h>
#endif /* _KERNEL */
static void spa_sync_version(void *arg, dmu_tx_t *tx);
static void spa_sync_props(void *arg, dmu_tx_t *tx);
static boolean_t spa_has_active_shared_spare(spa_t *spa);
-static inline int spa_load_impl(spa_t *spa, uint64_t, nvlist_t *config,
- spa_load_state_t state, spa_import_type_t type, boolean_t trust_config,
- char **ereport);
+static int spa_load_impl(spa_t *spa, spa_import_type_t type, char **ereport);
static void spa_vdev_resilver_done(spa_t *spa);
uint_t zio_taskq_batch_pct = 75; /* 1 thread per cpu in pset */
-id_t zio_taskq_psrset_bind = PS_NONE;
boolean_t zio_taskq_sysdc = B_TRUE; /* use SDC scheduling class */
uint_t zio_taskq_basedc = 80; /* base duty cycle */
boolean_t spa_create_process = B_TRUE; /* no process ==> no sysdc */
+/*
+ * Report any spa_load_verify errors found, but do not fail spa_load.
+ * This is used by zdb to analyze non-idle pools.
+ */
+boolean_t spa_load_verify_dryrun = B_FALSE;
+
/*
* This (illegal) pool name is used when temporarily importing a spa_t in order
* to get the vdev stats associated with the imported devices.
*/
#define TRYIMPORT_NAME "$import"
+/*
+ * For debugging purposes: print out vdev tree during pool import.
+ */
+int spa_load_print_vdev_tree = B_FALSE;
+
+/*
+ * A non-zero value for zfs_max_missing_tvds means that we allow importing
+ * pools with missing top-level vdevs. This is strictly intended for advanced
+ * pool recovery cases since missing data is almost inevitable. Pools with
+ * missing devices can only be imported read-only for safety reasons, and their
+ * fail-mode will be automatically set to "continue".
+ *
+ * With 1 missing vdev we should be able to import the pool and mount all
+ * datasets. User data that was not modified after the missing device has been
+ * added should be recoverable. This means that snapshots created prior to the
+ * addition of that device should be completely intact.
+ *
+ * With 2 missing vdevs, some datasets may fail to mount since there are
+ * dataset statistics that are stored as regular metadata. Some data might be
+ * recoverable if those vdevs were added recently.
+ *
+ * With 3 or more missing vdevs, the pool is severely damaged and MOS entries
+ * may be missing entirely. Chances of data recovery are very low. Note that
+ * there are also risks of performing an inadvertent rewind as we might be
+ * missing all the vdevs with the latest uberblocks.
+ */
+unsigned long zfs_max_missing_tvds = 0;
+
+/*
+ * The parameters below are similar to zfs_max_missing_tvds but are only
+ * intended for a preliminary open of the pool with an untrusted config which
+ * might be incomplete or out-dated.
+ *
+ * We are more tolerant for pools opened from a cachefile since we could have
+ * an out-dated cachefile where a device removal was not registered.
+ * We could have set the limit arbitrarily high but in the case where devices
+ * are really missing we would want to return the proper error codes; we chose
+ * SPA_DVAS_PER_BP - 1 so that some copies of the MOS would still be available
+ * and we get a chance to retrieve the trusted config.
+ */
+uint64_t zfs_max_missing_tvds_cachefile = SPA_DVAS_PER_BP - 1;
+
+/*
+ * In the case where config was assembled by scanning device paths (/dev/dsks
+ * by default) we are less tolerant since all the existing devices should have
+ * been detected and we want spa_load to return the right error codes.
+ */
+uint64_t zfs_max_missing_tvds_scan = 0;
+
+/*
+ * Debugging aid that pauses spa_sync() towards the end.
+ */
+boolean_t zfs_pause_spa_sync = B_FALSE;
+
/*
* ==========================================================================
* SPA properties routines
ASSERT(MUTEX_HELD(&spa->spa_props_lock));
if (rvd != NULL) {
- alloc = metaslab_class_get_alloc(spa_normal_class(spa));
- size = metaslab_class_get_space(spa_normal_class(spa));
+ alloc = metaslab_class_get_alloc(mc);
+ alloc += metaslab_class_get_alloc(spa_special_class(spa));
+ alloc += metaslab_class_get_alloc(spa_dedup_class(spa));
+
+ size = metaslab_class_get_space(mc);
+ size += metaslab_class_get_space(spa_special_class(spa));
+ size += metaslab_class_get_space(spa_dedup_class(spa));
+
spa_prop_add_list(*nvp, ZPOOL_PROP_NAME, spa_name(spa), 0, src);
spa_prop_add_list(*nvp, ZPOOL_PROP_SIZE, NULL, size, src);
spa_prop_add_list(*nvp, ZPOOL_PROP_ALLOCATED, NULL, alloc, src);
spa_prop_add_list(*nvp, ZPOOL_PROP_FREE, NULL,
size - alloc, src);
+ spa_prop_add_list(*nvp, ZPOOL_PROP_CHECKPOINT, NULL,
+ spa->spa_checkpoint_info.sci_dspace, src);
spa_prop_add_list(*nvp, ZPOOL_PROP_FRAGMENTATION, NULL,
metaslab_class_fragmentation(mc), src);
spa_prop_add_list(*nvp, ZPOOL_PROP_VERSION, NULL,
version, ZPROP_SRC_LOCAL);
}
+ spa_prop_add_list(*nvp, ZPOOL_PROP_LOAD_GUID,
+ NULL, spa_load_guid(spa), src);
}
if (pool != NULL) {
vdev_t *rvd = spa->spa_root_vdev;
uint64_t vdev_state;
+ if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) {
+ int error = (spa_has_checkpoint(spa)) ?
+ ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT;
+ return (SET_ERROR(error));
+ }
+
spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
vdev_state = rvd->vdev_state;
spa_config_exit(spa, SCL_STATE, FTAG);
uint_t value = ztip->zti_value;
uint_t count = ztip->zti_count;
spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q];
- char name[32];
uint_t flags = 0;
boolean_t batch = B_FALSE;
for (uint_t i = 0; i < count; i++) {
taskq_t *tq;
+ char name[32];
- if (count > 1) {
- (void) snprintf(name, sizeof (name), "%s_%s_%u",
- zio_type_name[t], zio_taskq_types[q], i);
- } else {
- (void) snprintf(name, sizeof (name), "%s_%s",
- zio_type_name[t], zio_taskq_types[q]);
- }
+ (void) snprintf(name, sizeof (name), "%s_%s",
+ zio_type_name[t], zio_taskq_types[q]);
if (zio_taskq_sysdc && spa->spa_proc != &p0) {
if (batch)
static void
spa_thread(void *arg)
{
+ psetid_t zio_taskq_psrset_bind = PS_NONE;
callb_cpr_t cprinfo;
spa_t *spa = arg;
spa->spa_normal_class = metaslab_class_create(spa, zfs_metaslab_ops);
spa->spa_log_class = metaslab_class_create(spa, zfs_metaslab_ops);
+ spa->spa_special_class = metaslab_class_create(spa, zfs_metaslab_ops);
+ spa->spa_dedup_class = metaslab_class_create(spa, zfs_metaslab_ops);
/* Try to create a covering process */
mutex_enter(&spa->spa_proc_lock);
metaslab_class_destroy(spa->spa_log_class);
spa->spa_log_class = NULL;
+ metaslab_class_destroy(spa->spa_special_class);
+ spa->spa_special_class = NULL;
+
+ metaslab_class_destroy(spa->spa_dedup_class);
+ spa->spa_dedup_class = NULL;
+
/*
* If this was part of an import or the open otherwise failed, we may
* still have errors left in the queues. Empty them just in case.
ASSERT(MUTEX_HELD(&spa_namespace_lock));
+ spa_load_note(spa, "UNLOADING");
+
/*
* Stop async tasks.
*/
spa->spa_vdev_removal = NULL;
}
+ if (spa->spa_condense_zthr != NULL) {
+ ASSERT(!zthr_isrunning(spa->spa_condense_zthr));
+ zthr_destroy(spa->spa_condense_zthr);
+ spa->spa_condense_zthr = NULL;
+ }
+
+ if (spa->spa_checkpoint_discard_zthr != NULL) {
+ ASSERT(!zthr_isrunning(spa->spa_checkpoint_discard_zthr));
+ zthr_destroy(spa->spa_checkpoint_discard_zthr);
+ spa->spa_checkpoint_discard_zthr = NULL;
+ }
+
spa_condense_fini(spa);
bpobj_close(&spa->spa_deferred_bpobj);
int i;
vdev_t *vd, *tvd;
+#ifndef _KERNEL
+ /*
+ * zdb opens both the current state of the pool and the
+ * checkpointed state (if present), with a different spa_t.
+ *
+ * As spare vdevs are shared among open pools, we skip loading
+ * them when we load the checkpointed state of the pool.
+ */
+ if (!spa_writeable(spa))
+ return;
+#endif
+
ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);
/*
vdev_t *vd, **oldvdevs, **newvdevs;
spa_aux_vdev_t *sav = &spa->spa_l2cache;
+#ifndef _KERNEL
+ /*
+ * zdb opens both the current state of the pool and the
+ * checkpointed state (if present), with a different spa_t.
+ *
+ * As L2 caches are part of the ARC which is shared among open
+ * pools, we skip loading them when we load the checkpointed
+ * state of the pool.
+ */
+ if (!spa_writeable(spa))
+ return;
+#endif
+
ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);
oldvdevs = sav->sav_vdevs;
return (error);
}
+/*
+ * Concrete top-level vdevs that are not missing and are not logs. At every
+ * spa_sync we write new uberblocks to at least SPA_SYNC_MIN_VDEVS core tvds.
+ */
+static uint64_t
+spa_healthy_core_tvds(spa_t *spa)
+{
+ vdev_t *rvd = spa->spa_root_vdev;
+ uint64_t tvds = 0;
+
+ for (uint64_t i = 0; i < rvd->vdev_children; i++) {
+ vdev_t *vd = rvd->vdev_child[i];
+ if (vd->vdev_islog)
+ continue;
+ if (vdev_is_concrete(vd) && !vdev_is_dead(vd))
+ tvds++;
+ }
+
+ return (tvds);
+}
+
/*
* Checks to see if the given vdev could not be opened, in which case we post a
* sysevent to notify the autoreplace code that the device has been removed.
static void
spa_check_removed(vdev_t *vd)
{
- for (int c = 0; c < vd->vdev_children; c++)
+ for (uint64_t 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) &&
}
}
-static void
-spa_config_valid_zaps(vdev_t *vd, vdev_t *mvd)
-{
- ASSERT3U(vd->vdev_children, ==, mvd->vdev_children);
-
- vd->vdev_top_zap = mvd->vdev_top_zap;
- vd->vdev_leaf_zap = mvd->vdev_leaf_zap;
-
- for (uint64_t i = 0; i < vd->vdev_children; i++) {
- spa_config_valid_zaps(vd->vdev_child[i], mvd->vdev_child[i]);
- }
-}
-
-/*
- * Validate the current config against the MOS config
- */
-static boolean_t
-spa_config_valid(spa_t *spa, nvlist_t *config)
+static int
+spa_check_for_missing_logs(spa_t *spa)
{
- vdev_t *mrvd, *rvd = spa->spa_root_vdev;
- nvlist_t *nv;
-
- VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nv) == 0);
-
- spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
- VERIFY(spa_config_parse(spa, &mrvd, nv, NULL, 0, VDEV_ALLOC_LOAD) == 0);
-
- ASSERT3U(rvd->vdev_children, ==, mrvd->vdev_children);
+ vdev_t *rvd = spa->spa_root_vdev;
/*
* If we're doing a normal import, then build up any additional
- * diagnostic information about missing devices in this config.
+ * diagnostic information about missing log devices.
* We'll pass this up to the user for further processing.
*/
if (!(spa->spa_import_flags & ZFS_IMPORT_MISSING_LOG)) {
KM_SLEEP);
VERIFY(nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_SLEEP) == 0);
- for (int c = 0; c < rvd->vdev_children; c++) {
+ for (uint64_t c = 0; c < rvd->vdev_children; c++) {
vdev_t *tvd = rvd->vdev_child[c];
- vdev_t *mtvd = mrvd->vdev_child[c];
- if (tvd->vdev_ops == &vdev_missing_ops &&
- mtvd->vdev_ops != &vdev_missing_ops &&
- mtvd->vdev_islog)
- child[idx++] = vdev_config_generate(spa, mtvd,
- B_FALSE, 0);
+ /*
+ * We consider a device as missing only if it failed
+ * to open (i.e. offline or faulted is not considered
+ * as missing).
+ */
+ if (tvd->vdev_islog &&
+ tvd->vdev_state == VDEV_STATE_CANT_OPEN) {
+ child[idx++] = vdev_config_generate(spa, tvd,
+ B_FALSE, VDEV_CONFIG_MISSING);
+ }
}
- if (idx) {
- VERIFY(nvlist_add_nvlist_array(nv,
- ZPOOL_CONFIG_CHILDREN, child, idx) == 0);
- VERIFY(nvlist_add_nvlist(spa->spa_load_info,
- ZPOOL_CONFIG_MISSING_DEVICES, nv) == 0);
+ if (idx > 0) {
+ fnvlist_add_nvlist_array(nv,
+ ZPOOL_CONFIG_CHILDREN, child, idx);
+ fnvlist_add_nvlist(spa->spa_load_info,
+ ZPOOL_CONFIG_MISSING_DEVICES, nv);
- for (int i = 0; i < idx; i++)
+ for (uint64_t i = 0; i < idx; i++)
nvlist_free(child[i]);
}
nvlist_free(nv);
kmem_free(child, rvd->vdev_children * sizeof (char **));
- }
-
- /*
- * Compare the root vdev tree with the information we have
- * from the MOS config (mrvd). Check each top-level vdev
- * with the corresponding MOS config top-level (mtvd).
- */
- for (int c = 0; c < rvd->vdev_children; c++) {
- vdev_t *tvd = rvd->vdev_child[c];
- vdev_t *mtvd = mrvd->vdev_child[c];
-
- /*
- * Resolve any "missing" vdevs in the current configuration.
- * Also trust the MOS config about any "indirect" vdevs.
- * If we find that the MOS config has more accurate information
- * about the top-level vdev then use that vdev instead.
- */
- if ((tvd->vdev_ops == &vdev_missing_ops &&
- mtvd->vdev_ops != &vdev_missing_ops) ||
- (mtvd->vdev_ops == &vdev_indirect_ops &&
- tvd->vdev_ops != &vdev_indirect_ops)) {
- /*
- * Device specific actions.
- */
- if (mtvd->vdev_islog) {
- if (!(spa->spa_import_flags &
- ZFS_IMPORT_MISSING_LOG)) {
- continue;
- }
+ if (idx > 0) {
+ spa_load_failed(spa, "some log devices are missing");
+ vdev_dbgmsg_print_tree(rvd, 2);
+ return (SET_ERROR(ENXIO));
+ }
+ } else {
+ for (uint64_t c = 0; c < rvd->vdev_children; c++) {
+ vdev_t *tvd = rvd->vdev_child[c];
+ if (tvd->vdev_islog &&
+ tvd->vdev_state == VDEV_STATE_CANT_OPEN) {
spa_set_log_state(spa, SPA_LOG_CLEAR);
- } else if (mtvd->vdev_ops != &vdev_indirect_ops) {
- continue;
- }
-
- /*
- * Swap the missing vdev with the data we were
- * able to obtain from the MOS config.
- */
- vdev_remove_child(rvd, tvd);
- vdev_remove_child(mrvd, mtvd);
-
- vdev_add_child(rvd, mtvd);
- vdev_add_child(mrvd, tvd);
-
- vdev_reopen(rvd);
- } else {
- if (mtvd->vdev_islog) {
- /*
- * Load the slog device's state from the MOS
- * config since it's possible that the label
- * does not contain the most up-to-date
- * information.
- */
- vdev_load_log_state(tvd, mtvd);
- vdev_reopen(tvd);
+ spa_load_note(spa, "some log devices are "
+ "missing, ZIL is dropped.");
+ vdev_dbgmsg_print_tree(rvd, 2);
+ break;
}
-
- /*
- * Per-vdev ZAP info is stored exclusively in the MOS.
- */
- spa_config_valid_zaps(tvd, mtvd);
}
-
- /*
- * Never trust this info from userland; always use what's
- * in the MOS. This prevents it from getting out of sync
- * with the rest of the info in the MOS.
- */
- tvd->vdev_removing = mtvd->vdev_removing;
- tvd->vdev_indirect_config = mtvd->vdev_indirect_config;
}
- vdev_free(mrvd);
- spa_config_exit(spa, SCL_ALL, FTAG);
-
- /*
- * Ensure we were able to validate the config.
- */
- return (rvd->vdev_guid_sum == spa->spa_uberblock.ub_guid_sum);
+ return (0);
}
/*
{
zio_t *rio;
spa_load_error_t sle = { 0 };
- zpool_rewind_policy_t policy;
+ zpool_load_policy_t policy;
boolean_t verify_ok = B_FALSE;
int error = 0;
- zpool_get_rewind_policy(spa->spa_config, &policy);
+ zpool_get_load_policy(spa->spa_config, &policy);
- if (policy.zrp_request & ZPOOL_NEVER_REWIND)
+ if (policy.zlp_rewind & ZPOOL_NEVER_REWIND)
return (0);
dsl_pool_config_enter(spa->spa_dsl_pool, FTAG);
ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE);
if (spa_load_verify_metadata) {
+ if (spa->spa_extreme_rewind) {
+ spa_load_note(spa, "performing a complete scan of the "
+ "pool since extreme rewind is on. This may take "
+ "a very long time.\n (spa_load_verify_data=%u, "
+ "spa_load_verify_metadata=%u)",
+ spa_load_verify_data, spa_load_verify_metadata);
+ }
error = traverse_pool(spa, spa->spa_verify_min_txg,
TRAVERSE_PRE | TRAVERSE_PREFETCH_METADATA |
TRAVERSE_NO_DECRYPT, spa_load_verify_cb, rio);
spa->spa_load_meta_errors = sle.sle_meta_count;
spa->spa_load_data_errors = sle.sle_data_count;
- if (!error && sle.sle_meta_count <= policy.zrp_maxmeta &&
- sle.sle_data_count <= policy.zrp_maxdata) {
+ if (sle.sle_meta_count != 0 || sle.sle_data_count != 0) {
+ spa_load_note(spa, "spa_load_verify found %llu metadata errors "
+ "and %llu data errors", (u_longlong_t)sle.sle_meta_count,
+ (u_longlong_t)sle.sle_data_count);
+ }
+
+ if (spa_load_verify_dryrun ||
+ (!error && sle.sle_meta_count <= policy.zlp_maxmeta &&
+ sle.sle_data_count <= policy.zlp_maxdata)) {
int64_t loss = 0;
verify_ok = B_TRUE;
spa->spa_load_max_txg = spa->spa_uberblock.ub_txg;
}
+ if (spa_load_verify_dryrun)
+ return (0);
+
if (error) {
if (error != ENXIO && error != EIO)
error = SET_ERROR(EIO);
* Find a value in the pool directory object.
*/
static int
-spa_dir_prop(spa_t *spa, const char *name, uint64_t *val)
+spa_dir_prop(spa_t *spa, const char *name, uint64_t *val, boolean_t log_enoent)
{
- return (zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
- name, sizeof (uint64_t), 1, val));
+ int error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
+ name, sizeof (uint64_t), 1, val);
+
+ if (error != 0 && (error != ENOENT || log_enoent)) {
+ spa_load_failed(spa, "couldn't get '%s' value in MOS directory "
+ "[error=%d]", name, error);
+ }
+
+ return (error);
}
static int
return (SET_ERROR(err));
}
+static void
+spa_spawn_aux_threads(spa_t *spa)
+{
+ ASSERT(spa_writeable(spa));
+
+ ASSERT(MUTEX_HELD(&spa_namespace_lock));
+
+ spa_start_indirect_condensing_thread(spa);
+
+ ASSERT3P(spa->spa_checkpoint_discard_zthr, ==, NULL);
+ spa->spa_checkpoint_discard_zthr =
+ zthr_create(spa_checkpoint_discard_thread_check,
+ spa_checkpoint_discard_thread, spa);
+}
+
/*
* Fix up config after a partly-completed split. This is done with the
* ZPOOL_CONFIG_SPLIT nvlist. Both the splitting pool and the split-off
}
static int
-spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type,
- boolean_t mosconfig)
+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;
-
- if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &pool_guid))
- return (SET_ERROR(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.
- */
- if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION,
- &spa->spa_ubsync.ub_version) != 0)
- spa->spa_ubsync.ub_version = SPA_VERSION_INITIAL;
-
- (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG,
- &spa->spa_config_txg);
- if ((state == SPA_LOAD_IMPORT || state == SPA_LOAD_TRYIMPORT) &&
- spa_guid_exists(pool_guid, 0)) {
- error = SET_ERROR(EEXIST);
- } else {
- 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);
- }
-
- nvlist_free(spa->spa_load_info);
- spa->spa_load_info = fnvlist_alloc();
+ spa->spa_load_state = state;
- gethrestime(&spa->spa_loaded_ts);
- error = spa_load_impl(spa, pool_guid, config, state, type,
- mosconfig, &ereport);
- }
+ gethrestime(&spa->spa_loaded_ts);
+ error = spa_load_impl(spa, type, &ereport);
/*
* Don't count references from objsets that are already closed
* and are making their way through the eviction process.
*/
spa_evicting_os_wait(spa);
- spa->spa_minref = refcount_count(&spa->spa_refcount);
+ spa->spa_minref = zfs_refcount_count(&spa->spa_refcount);
if (error) {
if (error != EEXIST) {
spa->spa_loaded_ts.tv_sec = 0;
return (error);
}
-/*
- * Load an existing storage pool, using the pool's builtin spa_config as a
- * source of configuration information.
- */
-__attribute__((always_inline))
-static inline int
-spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
- spa_load_state_t state, spa_import_type_t type, boolean_t trust_config,
- char **ereport)
+static int
+spa_verify_host(spa_t *spa, nvlist_t *mos_config)
+{
+ uint64_t hostid;
+ char *hostname;
+ uint64_t myhostid = 0;
+
+ if (!spa_is_root(spa) && nvlist_lookup_uint64(mos_config,
+ ZPOOL_CONFIG_HOSTID, &hostid) == 0) {
+ hostname = fnvlist_lookup_string(mos_config,
+ ZPOOL_CONFIG_HOSTNAME);
+
+ myhostid = zone_get_hostid(NULL);
+
+ if (hostid != 0 && myhostid != 0 && hostid != myhostid) {
+ cmn_err(CE_WARN, "pool '%s' could not be "
+ "loaded as it was last accessed by "
+ "another system (host: %s hostid: 0x%llx). "
+ "See: http://illumos.org/msg/ZFS-8000-EY",
+ spa_name(spa), hostname, (u_longlong_t)hostid);
+ spa_load_failed(spa, "hostid verification failed: pool "
+ "last accessed by host: %s (hostid: 0x%llx)",
+ hostname, (u_longlong_t)hostid);
+ return (SET_ERROR(EBADF));
+ }
+ }
+
+ return (0);
+}
+
+static int
+spa_ld_parse_config(spa_t *spa, spa_import_type_t type)
{
int error = 0;
- nvlist_t *nvroot = NULL;
- nvlist_t *label;
- vdev_t *rvd;
- uberblock_t *ub = &spa->spa_uberblock;
- uint64_t children, config_cache_txg = spa->spa_config_txg;
- int orig_mode = spa->spa_mode;
+ nvlist_t *nvtree, *nvl, *config = spa->spa_config;
int parse;
- uint64_t obj;
- boolean_t missing_feat_write = B_FALSE;
- boolean_t activity_check = B_FALSE;
+ vdev_t *rvd;
+ uint64_t pool_guid;
+ char *comment;
+
+ /*
+ * Versioning wasn't explicitly added to the label until later, so if
+ * it's not present treat it as the initial version.
+ */
+ if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION,
+ &spa->spa_ubsync.ub_version) != 0)
+ spa->spa_ubsync.ub_version = SPA_VERSION_INITIAL;
+
+ if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &pool_guid)) {
+ spa_load_failed(spa, "invalid config provided: '%s' missing",
+ ZPOOL_CONFIG_POOL_GUID);
+ return (SET_ERROR(EINVAL));
+ }
/*
- * If this is an untrusted config, access the pool in read-only mode.
- * This prevents things like resilvering recently removed devices.
+ * If we are doing an import, ensure that the pool is not already
+ * imported by checking if its pool guid already exists in the
+ * spa namespace.
+ *
+ * The only case that we allow an already imported pool to be
+ * imported again, is when the pool is checkpointed and we want to
+ * look at its checkpointed state from userland tools like zdb.
*/
- if (!trust_config)
- spa->spa_mode = FREAD;
+#ifdef _KERNEL
+ if ((spa->spa_load_state == SPA_LOAD_IMPORT ||
+ spa->spa_load_state == SPA_LOAD_TRYIMPORT) &&
+ spa_guid_exists(pool_guid, 0)) {
+#else
+ if ((spa->spa_load_state == SPA_LOAD_IMPORT ||
+ spa->spa_load_state == SPA_LOAD_TRYIMPORT) &&
+ spa_guid_exists(pool_guid, 0) &&
+ !spa_importing_readonly_checkpoint(spa)) {
+#endif
+ spa_load_failed(spa, "a pool with guid %llu is already open",
+ (u_longlong_t)pool_guid);
+ return (SET_ERROR(EEXIST));
+ }
- ASSERT(MUTEX_HELD(&spa_namespace_lock));
+ spa->spa_config_guid = pool_guid;
- spa->spa_load_state = state;
+ nvlist_free(spa->spa_load_info);
+ spa->spa_load_info = fnvlist_alloc();
- if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot))
- return (SET_ERROR(EINVAL));
+ ASSERT(spa->spa_comment == NULL);
+ if (nvlist_lookup_string(config, ZPOOL_CONFIG_COMMENT, &comment) == 0)
+ spa->spa_comment = spa_strdup(comment);
- parse = (type == SPA_IMPORT_EXISTING ?
- VDEV_ALLOC_LOAD : VDEV_ALLOC_SPLIT);
+ (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG,
+ &spa->spa_config_txg);
+
+ if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT, &nvl) == 0)
+ spa->spa_config_splitting = fnvlist_dup(nvl);
+
+ if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvtree)) {
+ spa_load_failed(spa, "invalid config provided: '%s' missing",
+ ZPOOL_CONFIG_VDEV_TREE);
+ return (SET_ERROR(EINVAL));
+ }
/*
* Create "The Godfather" zio to hold all async IOs
* configuration requires knowing the version number.
*/
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
- error = spa_config_parse(spa, &rvd, nvroot, NULL, 0, parse);
+ parse = (type == SPA_IMPORT_EXISTING ?
+ VDEV_ALLOC_LOAD : VDEV_ALLOC_SPLIT);
+ error = spa_config_parse(spa, &rvd, nvtree, NULL, 0, parse);
spa_config_exit(spa, SCL_ALL, FTAG);
- if (error != 0)
+ if (error != 0) {
+ spa_load_failed(spa, "unable to parse config [error=%d]",
+ error);
return (error);
+ }
ASSERT(spa->spa_root_vdev == rvd);
ASSERT3U(spa->spa_min_ashift, >=, SPA_MINBLOCKSHIFT);
ASSERT(spa_guid(spa) == pool_guid);
}
- /*
- * Try to open all vdevs, loading each label in the process.
- */
- spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
- error = vdev_open(rvd);
- spa_config_exit(spa, SCL_ALL, FTAG);
- if (error != 0)
- return (error);
+ return (0);
+}
+
+/*
+ * Recursively open all vdevs in the vdev tree. This function is called twice:
+ * first with the untrusted config, then with the trusted config.
+ */
+static int
+spa_ld_open_vdevs(spa_t *spa)
+{
+ int error = 0;
/*
- * We need to validate the vdev labels against the configuration that
- * we have in hand, which is dependent on the setting of mosconfig. If
- * mosconfig is true then we're validating the vdev labels based on
- * that config. Otherwise, we're validating against the cached config
- * (zpool.cache) that was read when we loaded the zfs module, and then
- * later we will recursively call spa_load() and validate against
- * the vdev config.
- *
- * If we're assembling a new pool that's been split off from an
- * existing pool, the labels haven't yet been updated so we skip
- * validation for now.
+ * spa_missing_tvds_allowed defines how many top-level vdevs can be
+ * missing/unopenable for the root vdev to be still considered openable.
*/
- if (type != SPA_IMPORT_ASSEMBLE) {
- spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
- error = vdev_validate(rvd, trust_config);
- spa_config_exit(spa, SCL_ALL, FTAG);
-
- if (error != 0)
- return (error);
-
- if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN)
- return (SET_ERROR(ENXIO));
+ if (spa->spa_trust_config) {
+ spa->spa_missing_tvds_allowed = zfs_max_missing_tvds;
+ } else if (spa->spa_config_source == SPA_CONFIG_SRC_CACHEFILE) {
+ spa->spa_missing_tvds_allowed = zfs_max_missing_tvds_cachefile;
+ } else if (spa->spa_config_source == SPA_CONFIG_SRC_SCAN) {
+ spa->spa_missing_tvds_allowed = zfs_max_missing_tvds_scan;
+ } else {
+ spa->spa_missing_tvds_allowed = 0;
}
- /*
- * Find the best uberblock.
- */
- vdev_uberblock_load(rvd, ub, &label);
+ spa->spa_missing_tvds_allowed =
+ MAX(zfs_max_missing_tvds, spa->spa_missing_tvds_allowed);
+
+ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
+ error = vdev_open(spa->spa_root_vdev);
+ spa_config_exit(spa, SCL_ALL, FTAG);
+
+ if (spa->spa_missing_tvds != 0) {
+ spa_load_note(spa, "vdev tree has %lld missing top-level "
+ "vdevs.", (u_longlong_t)spa->spa_missing_tvds);
+ if (spa->spa_trust_config && (spa->spa_mode & FWRITE)) {
+ /*
+ * Although theoretically we could allow users to open
+ * incomplete pools in RW mode, we'd need to add a lot
+ * of extra logic (e.g. adjust pool space to account
+ * for missing vdevs).
+ * This limitation also prevents users from accidentally
+ * opening the pool in RW mode during data recovery and
+ * damaging it further.
+ */
+ spa_load_note(spa, "pools with missing top-level "
+ "vdevs can only be opened in read-only mode.");
+ error = SET_ERROR(ENXIO);
+ } else {
+ spa_load_note(spa, "current settings allow for maximum "
+ "%lld missing top-level vdevs at this stage.",
+ (u_longlong_t)spa->spa_missing_tvds_allowed);
+ }
+ }
+ if (error != 0) {
+ spa_load_failed(spa, "unable to open vdev tree [error=%d]",
+ error);
+ }
+ if (spa->spa_missing_tvds != 0 || error != 0)
+ vdev_dbgmsg_print_tree(spa->spa_root_vdev, 2);
+
+ return (error);
+}
+
+/*
+ * We need to validate the vdev labels against the configuration that
+ * we have in hand. This function is called twice: first with an untrusted
+ * config, then with a trusted config. The validation is more strict when the
+ * config is trusted.
+ */
+static int
+spa_ld_validate_vdevs(spa_t *spa)
+{
+ int error = 0;
+ vdev_t *rvd = spa->spa_root_vdev;
+
+ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
+ error = vdev_validate(rvd);
+ spa_config_exit(spa, SCL_ALL, FTAG);
+
+ if (error != 0) {
+ spa_load_failed(spa, "vdev_validate failed [error=%d]", error);
+ return (error);
+ }
+
+ if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN) {
+ spa_load_failed(spa, "cannot open vdev tree after invalidating "
+ "some vdevs");
+ vdev_dbgmsg_print_tree(rvd, 2);
+ return (SET_ERROR(ENXIO));
+ }
+
+ return (0);
+}
+
+static void
+spa_ld_select_uberblock_done(spa_t *spa, uberblock_t *ub)
+{
+ spa->spa_state = POOL_STATE_ACTIVE;
+ spa->spa_ubsync = spa->spa_uberblock;
+ spa->spa_verify_min_txg = spa->spa_extreme_rewind ?
+ TXG_INITIAL - 1 : spa_last_synced_txg(spa) - TXG_DEFER_SIZE - 1;
+ spa->spa_first_txg = spa->spa_last_ubsync_txg ?
+ spa->spa_last_ubsync_txg : spa_last_synced_txg(spa) + 1;
+ spa->spa_claim_max_txg = spa->spa_first_txg;
+ spa->spa_prev_software_version = ub->ub_software_version;
+}
+
+static int
+spa_ld_select_uberblock(spa_t *spa, spa_import_type_t type)
+{
+ vdev_t *rvd = spa->spa_root_vdev;
+ nvlist_t *label;
+ uberblock_t *ub = &spa->spa_uberblock;
+ boolean_t activity_check = B_FALSE;
+
+ /*
+ * If we are opening the checkpointed state of the pool by
+ * rewinding to it, at this point we will have written the
+ * checkpointed uberblock to the vdev labels, so searching
+ * the labels will find the right uberblock. However, if
+ * we are opening the checkpointed state read-only, we have
+ * not modified the labels. Therefore, we must ignore the
+ * labels and continue using the spa_uberblock that was set
+ * by spa_ld_checkpoint_rewind.
+ *
+ * Note that it would be fine to ignore the labels when
+ * rewinding (opening writeable) as well. However, if we
+ * crash just after writing the labels, we will end up
+ * searching the labels. Doing so in the common case means
+ * that this code path gets exercised normally, rather than
+ * just in the edge case.
+ */
+ if (ub->ub_checkpoint_txg != 0 &&
+ spa_importing_readonly_checkpoint(spa)) {
+ spa_ld_select_uberblock_done(spa, ub);
+ return (0);
+ }
+
+ /*
+ * Find the best uberblock.
+ */
+ vdev_uberblock_load(rvd, ub, &label);
/*
* If we weren't able to find a single valid uberblock, return failure.
*/
if (ub->ub_txg == 0) {
nvlist_free(label);
+ spa_load_failed(spa, "no valid uberblock found");
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, ENXIO));
}
+ spa_load_note(spa, "using uberblock with txg=%llu",
+ (u_longlong_t)ub->ub_txg);
+
+
/*
* For pools which have the multihost property on determine if the
* 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, label, config);
+ activity_check = spa_activity_check_required(spa, ub, label,
+ spa->spa_config);
if (activity_check) {
if (ub->ub_mmp_magic == MMP_MAGIC && ub->ub_mmp_delay &&
spa_get_hostid() == 0) {
return (spa_vdev_err(rvd, VDEV_AUX_ACTIVE, EREMOTEIO));
}
- error = spa_activity_check(spa, ub, config);
+ int error = spa_activity_check(spa, ub, spa->spa_config);
if (error) {
nvlist_free(label);
return (error);
*/
if (!SPA_VERSION_IS_SUPPORTED(ub->ub_version)) {
nvlist_free(label);
+ spa_load_failed(spa, "version %llu is not supported",
+ (u_longlong_t)ub->ub_version);
return (spa_vdev_err(rvd, VDEV_AUX_VERSION_NEWER, ENOTSUP));
}
* If we weren't able to find what's necessary for reading the
* MOS in the label, return failure.
*/
- if (label == NULL || nvlist_lookup_nvlist(label,
- ZPOOL_CONFIG_FEATURES_FOR_READ, &features) != 0) {
+ if (label == NULL) {
+ spa_load_failed(spa, "label config unavailable");
+ return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA,
+ ENXIO));
+ }
+
+ if (nvlist_lookup_nvlist(label, ZPOOL_CONFIG_FEATURES_FOR_READ,
+ &features) != 0) {
nvlist_free(label);
+ spa_load_failed(spa, "invalid label: '%s' missing",
+ ZPOOL_CONFIG_FEATURES_FOR_READ);
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA,
ENXIO));
}
VERIFY(nvlist_add_nvlist(spa->spa_load_info,
ZPOOL_CONFIG_UNSUP_FEAT, unsup_feat) == 0);
nvlist_free(unsup_feat);
+ spa_load_failed(spa, "some features are unsupported");
return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT,
ENOTSUP));
}
nvlist_free(unsup_feat);
}
- /*
- * If the vdev guid sum doesn't match the uberblock, we have an
- * incomplete configuration. We first check to see if the pool
- * is aware of the complete config (i.e ZPOOL_CONFIG_VDEV_CHILDREN).
- * If it is, defer the vdev_guid_sum check till later so we
- * can handle missing vdevs.
- */
- if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_VDEV_CHILDREN,
- &children) != 0 && trust_config && type != SPA_IMPORT_ASSEMBLE &&
- rvd->vdev_guid_sum != ub->ub_guid_sum)
- return (spa_vdev_err(rvd, VDEV_AUX_BAD_GUID_SUM, ENXIO));
-
if (type != SPA_IMPORT_ASSEMBLE && spa->spa_config_splitting) {
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
- spa_try_repair(spa, config);
+ spa_try_repair(spa, spa->spa_config);
spa_config_exit(spa, SCL_ALL, FTAG);
nvlist_free(spa->spa_config_splitting);
spa->spa_config_splitting = NULL;
/*
* Initialize internal SPA structures.
*/
- spa->spa_state = POOL_STATE_ACTIVE;
- spa->spa_ubsync = spa->spa_uberblock;
- spa->spa_verify_min_txg = spa->spa_extreme_rewind ?
- TXG_INITIAL - 1 : spa_last_synced_txg(spa) - TXG_DEFER_SIZE - 1;
- spa->spa_first_txg = spa->spa_last_ubsync_txg ?
- spa->spa_last_ubsync_txg : spa_last_synced_txg(spa) + 1;
- spa->spa_claim_max_txg = spa->spa_first_txg;
- spa->spa_prev_software_version = ub->ub_software_version;
+ spa_ld_select_uberblock_done(spa, ub);
- /*
- * Everything that we read before we do spa_remove_init() must
- * have been rewritten after the last device removal was initiated.
- * Otherwise we could be reading from indirect vdevs before
- * we have loaded their mappings.
- */
+ return (0);
+}
+
+static int
+spa_ld_open_rootbp(spa_t *spa)
+{
+ int error = 0;
+ vdev_t *rvd = spa->spa_root_vdev;
error = dsl_pool_init(spa, spa->spa_first_txg, &spa->spa_dsl_pool);
- if (error)
+ if (error != 0) {
+ spa_load_failed(spa, "unable to open rootbp in dsl_pool_init "
+ "[error=%d]", error);
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ }
spa->spa_meta_objset = spa->spa_dsl_pool->dp_meta_objset;
- if (spa_dir_prop(spa, DMU_POOL_CONFIG, &spa->spa_config_object) != 0)
+ return (0);
+}
+
+static int
+spa_ld_trusted_config(spa_t *spa, spa_import_type_t type,
+ boolean_t reloading)
+{
+ vdev_t *mrvd, *rvd = spa->spa_root_vdev;
+ nvlist_t *nv, *mos_config, *policy;
+ int error = 0, copy_error;
+ uint64_t healthy_tvds, healthy_tvds_mos;
+ uint64_t mos_config_txg;
+
+ if (spa_dir_prop(spa, DMU_POOL_CONFIG, &spa->spa_config_object, B_TRUE)
+ != 0)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
/*
- * Validate the config, using the MOS config to fill in any
- * information which might be missing. If we fail to validate
- * the config then declare the pool unfit for use. If we're
- * assembling a pool from a split, the log is not transferred
- * over.
+ * If we're assembling a pool from a split, the config provided is
+ * already trusted so there is nothing to do.
*/
- if (type != SPA_IMPORT_ASSEMBLE) {
- 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));
+ if (type == SPA_IMPORT_ASSEMBLE)
+ return (0);
+
+ healthy_tvds = spa_healthy_core_tvds(spa);
+
+ if (load_nvlist(spa, spa->spa_config_object, &mos_config)
+ != 0) {
+ spa_load_failed(spa, "unable to retrieve MOS config");
+ return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ }
- if (!spa_config_valid(spa, mos_config)) {
+ /*
+ * If we are doing an open, pool owner wasn't verified yet, thus do
+ * the verification here.
+ */
+ if (spa->spa_load_state == SPA_LOAD_OPEN) {
+ error = spa_verify_host(spa, mos_config);
+ if (error != 0) {
nvlist_free(mos_config);
- return (spa_vdev_err(rvd, VDEV_AUX_BAD_GUID_SUM,
- ENXIO));
+ return (error);
}
- nvlist_free(mos_config);
+ }
+
+ nv = fnvlist_lookup_nvlist(mos_config, ZPOOL_CONFIG_VDEV_TREE);
+
+ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
+
+ /*
+ * Build a new vdev tree from the trusted config
+ */
+ VERIFY(spa_config_parse(spa, &mrvd, nv, NULL, 0, VDEV_ALLOC_LOAD) == 0);
+
+ /*
+ * Vdev paths in the MOS may be obsolete. If the untrusted config was
+ * obtained by scanning /dev/dsk, then it will have the right vdev
+ * paths. We update the trusted MOS config with this information.
+ * We first try to copy the paths with vdev_copy_path_strict, which
+ * succeeds only when both configs have exactly the same vdev tree.
+ * If that fails, we fall back to a more flexible method that has a
+ * best effort policy.
+ */
+ copy_error = vdev_copy_path_strict(rvd, mrvd);
+ if (copy_error != 0 || spa_load_print_vdev_tree) {
+ spa_load_note(spa, "provided vdev tree:");
+ vdev_dbgmsg_print_tree(rvd, 2);
+ spa_load_note(spa, "MOS vdev tree:");
+ vdev_dbgmsg_print_tree(mrvd, 2);
+ }
+ if (copy_error != 0) {
+ spa_load_note(spa, "vdev_copy_path_strict failed, falling "
+ "back to vdev_copy_path_relaxed");
+ vdev_copy_path_relaxed(rvd, mrvd);
+ }
+
+ vdev_close(rvd);
+ vdev_free(rvd);
+ spa->spa_root_vdev = mrvd;
+ rvd = mrvd;
+ spa_config_exit(spa, SCL_ALL, FTAG);
+
+ /*
+ * We will use spa_config if we decide to reload the spa or if spa_load
+ * fails and we rewind. We must thus regenerate the config using the
+ * MOS information with the updated paths. ZPOOL_LOAD_POLICY is used to
+ * pass settings on how to load the pool and is not stored in the MOS.
+ * We copy it over to our new, trusted config.
+ */
+ mos_config_txg = fnvlist_lookup_uint64(mos_config,
+ ZPOOL_CONFIG_POOL_TXG);
+ nvlist_free(mos_config);
+ mos_config = spa_config_generate(spa, NULL, mos_config_txg, B_FALSE);
+ if (nvlist_lookup_nvlist(spa->spa_config, ZPOOL_LOAD_POLICY,
+ &policy) == 0)
+ fnvlist_add_nvlist(mos_config, ZPOOL_LOAD_POLICY, policy);
+ spa_config_set(spa, mos_config);
+ spa->spa_config_source = SPA_CONFIG_SRC_MOS;
+
+ /*
+ * Now that we got the config from the MOS, we should be more strict
+ * in checking blkptrs and can make assumptions about the consistency
+ * of the vdev tree. spa_trust_config must be set to true before opening
+ * vdevs in order for them to be writeable.
+ */
+ spa->spa_trust_config = B_TRUE;
+
+ /*
+ * Open and validate the new vdev tree
+ */
+ error = spa_ld_open_vdevs(spa);
+ if (error != 0)
+ return (error);
+
+ error = spa_ld_validate_vdevs(spa);
+ if (error != 0)
+ return (error);
+ if (copy_error != 0 || spa_load_print_vdev_tree) {
+ spa_load_note(spa, "final vdev tree:");
+ vdev_dbgmsg_print_tree(rvd, 2);
+ }
+
+ if (spa->spa_load_state != SPA_LOAD_TRYIMPORT &&
+ !spa->spa_extreme_rewind && zfs_max_missing_tvds == 0) {
/*
- * Now that we've validated the config, check the state of the
- * root vdev. If it can't be opened, it indicates one or
- * more toplevel vdevs are faulted.
+ * Sanity check to make sure that we are indeed loading the
+ * latest uberblock. If we missed SPA_SYNC_MIN_VDEVS tvds
+ * in the config provided and they happened to be the only ones
+ * to have the latest uberblock, we could involuntarily perform
+ * an extreme rewind.
*/
- if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN)
- return (SET_ERROR(ENXIO));
+ healthy_tvds_mos = spa_healthy_core_tvds(spa);
+ if (healthy_tvds_mos - healthy_tvds >=
+ SPA_SYNC_MIN_VDEVS) {
+ spa_load_note(spa, "config provided misses too many "
+ "top-level vdevs compared to MOS (%lld vs %lld). ",
+ (u_longlong_t)healthy_tvds,
+ (u_longlong_t)healthy_tvds_mos);
+ spa_load_note(spa, "vdev tree:");
+ vdev_dbgmsg_print_tree(rvd, 2);
+ if (reloading) {
+ spa_load_failed(spa, "config was already "
+ "provided from MOS. Aborting.");
+ return (spa_vdev_err(rvd,
+ VDEV_AUX_CORRUPT_DATA, EIO));
+ }
+ spa_load_note(spa, "spa must be reloaded using MOS "
+ "config");
+ return (SET_ERROR(EAGAIN));
+ }
+ }
+
+ error = spa_check_for_missing_logs(spa);
+ if (error != 0)
+ return (spa_vdev_err(rvd, VDEV_AUX_BAD_GUID_SUM, ENXIO));
+
+ if (rvd->vdev_guid_sum != spa->spa_uberblock.ub_guid_sum) {
+ spa_load_failed(spa, "uberblock guid sum doesn't match MOS "
+ "guid sum (%llu != %llu)",
+ (u_longlong_t)spa->spa_uberblock.ub_guid_sum,
+ (u_longlong_t)rvd->vdev_guid_sum);
+ return (spa_vdev_err(rvd, VDEV_AUX_BAD_GUID_SUM,
+ ENXIO));
}
+ return (0);
+}
+
+static int
+spa_ld_open_indirect_vdev_metadata(spa_t *spa)
+{
+ int error = 0;
+ vdev_t *rvd = spa->spa_root_vdev;
+
/*
* Everything that we read before spa_remove_init() must be stored
* on concreted vdevs. Therefore we do this as early as possible.
*/
- if (spa_remove_init(spa) != 0)
+ error = spa_remove_init(spa);
+ if (error != 0) {
+ spa_load_failed(spa, "spa_remove_init failed [error=%d]",
+ error);
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ }
+
+ /*
+ * Retrieve information needed to condense indirect vdev mappings.
+ */
+ error = spa_condense_init(spa);
+ if (error != 0) {
+ spa_load_failed(spa, "spa_condense_init failed [error=%d]",
+ error);
+ return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error));
+ }
+
+ return (0);
+}
+
+static int
+spa_ld_check_features(spa_t *spa, boolean_t *missing_feat_writep)
+{
+ int error = 0;
+ vdev_t *rvd = spa->spa_root_vdev;
if (spa_version(spa) >= SPA_VERSION_FEATURES) {
boolean_t missing_feat_read = B_FALSE;
nvlist_t *unsup_feat, *enabled_feat;
if (spa_dir_prop(spa, DMU_POOL_FEATURES_FOR_READ,
- &spa->spa_feat_for_read_obj) != 0) {
+ &spa->spa_feat_for_read_obj, B_TRUE) != 0) {
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
}
if (spa_dir_prop(spa, DMU_POOL_FEATURES_FOR_WRITE,
- &spa->spa_feat_for_write_obj) != 0) {
+ &spa->spa_feat_for_write_obj, B_TRUE) != 0) {
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
}
if (spa_dir_prop(spa, DMU_POOL_FEATURE_DESCRIPTIONS,
- &spa->spa_feat_desc_obj) != 0) {
+ &spa->spa_feat_desc_obj, B_TRUE) != 0) {
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
}
unsup_feat, enabled_feat))
missing_feat_read = B_TRUE;
- if (spa_writeable(spa) || state == SPA_LOAD_TRYIMPORT) {
+ if (spa_writeable(spa) ||
+ spa->spa_load_state == SPA_LOAD_TRYIMPORT) {
if (!spa_features_check(spa, B_TRUE,
unsup_feat, enabled_feat)) {
- missing_feat_write = B_TRUE;
+ *missing_feat_writep = B_TRUE;
}
}
* userland in order to know whether to display the
* abovementioned note.
*/
- if (missing_feat_read || (missing_feat_write &&
+ if (missing_feat_read || (*missing_feat_writep &&
spa_writeable(spa))) {
+ spa_load_failed(spa, "pool uses unsupported features");
return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT,
ENOTSUP));
}
spa->spa_feat_refcount_cache[i] =
SPA_FEATURE_DISABLED;
} else {
+ spa_load_failed(spa, "error getting refcount "
+ "for feature %s [error=%d]",
+ spa_feature_table[i].fi_guid, error);
return (spa_vdev_err(rvd,
VDEV_AUX_CORRUPT_DATA, EIO));
}
if (spa_feature_is_active(spa, SPA_FEATURE_ENABLED_TXG)) {
if (spa_dir_prop(spa, DMU_POOL_FEATURE_ENABLED_TXG,
- &spa->spa_feat_enabled_txg_obj) != 0)
+ &spa->spa_feat_enabled_txg_obj, B_TRUE) != 0)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
}
+ return (0);
+}
+
+static int
+spa_ld_load_special_directories(spa_t *spa)
+{
+ int error = 0;
+ vdev_t *rvd = spa->spa_root_vdev;
+
spa->spa_is_initializing = B_TRUE;
error = dsl_pool_open(spa->spa_dsl_pool);
spa->spa_is_initializing = B_FALSE;
- if (error != 0)
+ if (error != 0) {
+ spa_load_failed(spa, "dsl_pool_open failed [error=%d]", error);
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ }
- if (!trust_config) {
- uint64_t hostid;
- nvlist_t *policy = NULL;
- 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));
-
- if (!spa_is_root(spa) && nvlist_lookup_uint64(mos_config,
- ZPOOL_CONFIG_HOSTID, &hostid) == 0) {
- char *hostname;
- unsigned long myhostid = 0;
-
- VERIFY(nvlist_lookup_string(mos_config,
- ZPOOL_CONFIG_HOSTNAME, &hostname) == 0);
-
- myhostid = spa_get_hostid();
- if (hostid && myhostid && hostid != myhostid) {
- nvlist_free(mos_config);
- return (SET_ERROR(EBADF));
- }
- }
- if (nvlist_lookup_nvlist(spa->spa_config,
- ZPOOL_REWIND_POLICY, &policy) == 0)
- VERIFY(nvlist_add_nvlist(mos_config,
- ZPOOL_REWIND_POLICY, policy) == 0);
-
- spa_config_set(spa, mos_config);
- spa_unload(spa);
- spa_deactivate(spa);
- spa_activate(spa, orig_mode);
+ return (0);
+}
- return (spa_load(spa, state, SPA_IMPORT_EXISTING, B_TRUE));
- }
+static int
+spa_ld_get_props(spa_t *spa)
+{
+ int error = 0;
+ uint64_t obj;
+ vdev_t *rvd = spa->spa_root_vdev;
/* Grab the checksum salt from the MOS. */
error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
(void) random_get_pseudo_bytes(spa->spa_cksum_salt.zcs_bytes,
sizeof (spa->spa_cksum_salt.zcs_bytes));
} else if (error != 0) {
+ spa_load_failed(spa, "unable to retrieve checksum salt from "
+ "MOS [error=%d]", error);
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
}
- if (spa_dir_prop(spa, DMU_POOL_SYNC_BPOBJ, &obj) != 0)
+ if (spa_dir_prop(spa, DMU_POOL_SYNC_BPOBJ, &obj, B_TRUE) != 0)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
error = bpobj_open(&spa->spa_deferred_bpobj, spa->spa_meta_objset, obj);
- if (error != 0)
+ if (error != 0) {
+ spa_load_failed(spa, "error opening deferred-frees bpobj "
+ "[error=%d]", error);
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ }
/*
* Load the bit that tells us to use the new accounting function
* (raid-z deflation). If we have an older pool, this will not
* be present.
*/
- error = spa_dir_prop(spa, DMU_POOL_DEFLATE, &spa->spa_deflate);
+ error = spa_dir_prop(spa, DMU_POOL_DEFLATE, &spa->spa_deflate, B_FALSE);
if (error != 0 && error != ENOENT)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
error = spa_dir_prop(spa, DMU_POOL_CREATION_VERSION,
- &spa->spa_creation_version);
+ &spa->spa_creation_version, B_FALSE);
if (error != 0 && error != ENOENT)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
* Load the persistent error log. If we have an older pool, this will
* not be present.
*/
- error = spa_dir_prop(spa, DMU_POOL_ERRLOG_LAST, &spa->spa_errlog_last);
+ error = spa_dir_prop(spa, DMU_POOL_ERRLOG_LAST, &spa->spa_errlog_last,
+ B_FALSE);
if (error != 0 && error != ENOENT)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
error = spa_dir_prop(spa, DMU_POOL_ERRLOG_SCRUB,
- &spa->spa_errlog_scrub);
+ &spa->spa_errlog_scrub, B_FALSE);
if (error != 0 && error != ENOENT)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
* Load the history object. If we have an older pool, this
* will not be present.
*/
- error = spa_dir_prop(spa, DMU_POOL_HISTORY, &spa->spa_history);
+ error = spa_dir_prop(spa, DMU_POOL_HISTORY, &spa->spa_history, B_FALSE);
if (error != 0 && error != ENOENT)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
/* The sentinel is only available in the MOS config. */
nvlist_t *mos_config;
- if (load_nvlist(spa, spa->spa_config_object, &mos_config) != 0)
+ if (load_nvlist(spa, spa->spa_config_object, &mos_config) != 0) {
+ spa_load_failed(spa, "unable to retrieve MOS config");
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ }
error = spa_dir_prop(spa, DMU_POOL_VDEV_ZAP_MAP,
- &spa->spa_all_vdev_zaps);
+ &spa->spa_all_vdev_zaps, B_FALSE);
if (error == ENOENT) {
VERIFY(!nvlist_exists(mos_config,
*/
ASSERT0(vdev_count_verify_zaps(spa->spa_root_vdev));
}
- nvlist_free(mos_config);
+ nvlist_free(mos_config);
+
+ spa->spa_delegation = zpool_prop_default_numeric(ZPOOL_PROP_DELEGATION);
+
+ error = spa_dir_prop(spa, DMU_POOL_PROPS, &spa->spa_pool_props_object,
+ B_FALSE);
+ if (error && error != ENOENT)
+ return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+
+ if (error == 0) {
+ uint64_t autoreplace;
+
+ spa_prop_find(spa, ZPOOL_PROP_BOOTFS, &spa->spa_bootfs);
+ spa_prop_find(spa, ZPOOL_PROP_AUTOREPLACE, &autoreplace);
+ spa_prop_find(spa, ZPOOL_PROP_DELEGATION, &spa->spa_delegation);
+ spa_prop_find(spa, ZPOOL_PROP_FAILUREMODE, &spa->spa_failmode);
+ spa_prop_find(spa, ZPOOL_PROP_AUTOEXPAND, &spa->spa_autoexpand);
+ spa_prop_find(spa, ZPOOL_PROP_MULTIHOST, &spa->spa_multihost);
+ spa_prop_find(spa, ZPOOL_PROP_DEDUPDITTO,
+ &spa->spa_dedup_ditto);
+
+ spa->spa_autoreplace = (autoreplace != 0);
+ }
+
+ /*
+ * If we are importing a pool with missing top-level vdevs,
+ * we enforce that the pool doesn't panic or get suspended on
+ * error since the likelihood of missing data is extremely high.
+ */
+ if (spa->spa_missing_tvds > 0 &&
+ spa->spa_failmode != ZIO_FAILURE_MODE_CONTINUE &&
+ spa->spa_load_state != SPA_LOAD_TRYIMPORT) {
+ spa_load_note(spa, "forcing failmode to 'continue' "
+ "as some top level vdevs are missing");
+ spa->spa_failmode = ZIO_FAILURE_MODE_CONTINUE;
+ }
+
+ return (0);
+}
+
+static int
+spa_ld_open_aux_vdevs(spa_t *spa, spa_import_type_t type)
+{
+ int error = 0;
+ vdev_t *rvd = spa->spa_root_vdev;
+
+ /*
+ * If we're assembling the pool from the split-off vdevs of
+ * an existing pool, we don't want to attach the spares & cache
+ * devices.
+ */
+
+ /*
+ * Load any hot spares for this pool.
+ */
+ error = spa_dir_prop(spa, DMU_POOL_SPARES, &spa->spa_spares.sav_object,
+ B_FALSE);
+ if (error != 0 && error != ENOENT)
+ return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ if (error == 0 && type != SPA_IMPORT_ASSEMBLE) {
+ ASSERT(spa_version(spa) >= SPA_VERSION_SPARES);
+ if (load_nvlist(spa, spa->spa_spares.sav_object,
+ &spa->spa_spares.sav_config) != 0) {
+ spa_load_failed(spa, "error loading spares nvlist");
+ return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ }
+
+ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
+ spa_load_spares(spa);
+ spa_config_exit(spa, SCL_ALL, FTAG);
+ } else if (error == 0) {
+ spa->spa_spares.sav_sync = B_TRUE;
+ }
+
+ /*
+ * Load any level 2 ARC devices for this pool.
+ */
+ error = spa_dir_prop(spa, DMU_POOL_L2CACHE,
+ &spa->spa_l2cache.sav_object, B_FALSE);
+ if (error != 0 && error != ENOENT)
+ return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ if (error == 0 && type != SPA_IMPORT_ASSEMBLE) {
+ ASSERT(spa_version(spa) >= SPA_VERSION_L2CACHE);
+ if (load_nvlist(spa, spa->spa_l2cache.sav_object,
+ &spa->spa_l2cache.sav_config) != 0) {
+ spa_load_failed(spa, "error loading l2cache nvlist");
+ return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ }
+
+ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
+ spa_load_l2cache(spa);
+ spa_config_exit(spa, SCL_ALL, FTAG);
+ } else if (error == 0) {
+ spa->spa_l2cache.sav_sync = B_TRUE;
+ }
+
+ return (0);
+}
+
+static int
+spa_ld_load_vdev_metadata(spa_t *spa)
+{
+ int error = 0;
+ vdev_t *rvd = spa->spa_root_vdev;
+
+ /*
+ * If the 'multihost' property is set, then never allow a pool to
+ * be imported when the system hostid is zero. The exception to
+ * this rule is zdb which is always allowed to access pools.
+ */
+ if (spa_multihost(spa) && spa_get_hostid() == 0 &&
+ (spa->spa_import_flags & ZFS_IMPORT_SKIP_MMP) == 0) {
+ fnvlist_add_uint64(spa->spa_load_info,
+ ZPOOL_CONFIG_MMP_STATE, MMP_STATE_NO_HOSTID);
+ return (spa_vdev_err(rvd, VDEV_AUX_ACTIVE, EREMOTEIO));
+ }
+
+ /*
+ * If the 'autoreplace' property is set, then post a resource notifying
+ * the ZFS DE that it should not issue any faults for unopenable
+ * devices. We also iterate over the vdevs, and post a sysevent for any
+ * unopenable vdevs so that the normal autoreplace handler can take
+ * over.
+ */
+ if (spa->spa_autoreplace && spa->spa_load_state != SPA_LOAD_TRYIMPORT) {
+ spa_check_removed(spa->spa_root_vdev);
+ /*
+ * For the import case, this is done in spa_import(), because
+ * at this point we're using the spare definitions from
+ * the MOS config, not necessarily from the userland config.
+ */
+ if (spa->spa_load_state != SPA_LOAD_IMPORT) {
+ spa_aux_check_removed(&spa->spa_spares);
+ spa_aux_check_removed(&spa->spa_l2cache);
+ }
+ }
+
+ /*
+ * Load the vdev metadata such as metaslabs, DTLs, spacemap object, etc.
+ */
+ error = vdev_load(rvd);
+ if (error != 0) {
+ spa_load_failed(spa, "vdev_load failed [error=%d]", error);
+ return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error));
+ }
+
+ /*
+ * Propagate the leaf DTLs we just loaded all the way up the vdev tree.
+ */
+ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
+ vdev_dtl_reassess(rvd, 0, 0, B_FALSE);
+ spa_config_exit(spa, SCL_ALL, FTAG);
+
+ return (0);
+}
+
+static int
+spa_ld_load_dedup_tables(spa_t *spa)
+{
+ int error = 0;
+ vdev_t *rvd = spa->spa_root_vdev;
+
+ error = ddt_load(spa);
+ if (error != 0) {
+ spa_load_failed(spa, "ddt_load failed [error=%d]", error);
+ return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ }
+
+ return (0);
+}
+
+static int
+spa_ld_verify_logs(spa_t *spa, spa_import_type_t type, char **ereport)
+{
+ vdev_t *rvd = spa->spa_root_vdev;
+
+ if (type != SPA_IMPORT_ASSEMBLE && spa_writeable(spa)) {
+ boolean_t missing = spa_check_logs(spa);
+ if (missing) {
+ if (spa->spa_missing_tvds != 0) {
+ spa_load_note(spa, "spa_check_logs failed "
+ "so dropping the logs");
+ } else {
+ *ereport = FM_EREPORT_ZFS_LOG_REPLAY;
+ spa_load_failed(spa, "spa_check_logs failed");
+ return (spa_vdev_err(rvd, VDEV_AUX_BAD_LOG,
+ ENXIO));
+ }
+ }
+ }
+
+ return (0);
+}
+
+static int
+spa_ld_verify_pool_data(spa_t *spa)
+{
+ int error = 0;
+ vdev_t *rvd = spa->spa_root_vdev;
+
+ /*
+ * We've successfully opened the pool, verify that we're ready
+ * to start pushing transactions.
+ */
+ if (spa->spa_load_state != SPA_LOAD_TRYIMPORT) {
+ error = spa_load_verify(spa);
+ if (error != 0) {
+ spa_load_failed(spa, "spa_load_verify failed "
+ "[error=%d]", error);
+ return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA,
+ error));
+ }
+ }
+
+ return (0);
+}
+
+static void
+spa_ld_claim_log_blocks(spa_t *spa)
+{
+ dmu_tx_t *tx;
+ dsl_pool_t *dp = spa_get_dsl(spa);
+
+ /*
+ * Claim log blocks that haven't been committed yet.
+ * This must all happen in a single txg.
+ * Note: spa_claim_max_txg is updated by spa_claim_notify(),
+ * invoked from zil_claim_log_block()'s i/o done callback.
+ * Price of rollback is that we abandon the log.
+ */
+ spa->spa_claiming = B_TRUE;
+
+ tx = dmu_tx_create_assigned(dp, spa_first_txg(spa));
+ (void) dmu_objset_find_dp(dp, dp->dp_root_dir_obj,
+ zil_claim, tx, DS_FIND_CHILDREN);
+ dmu_tx_commit(tx);
+
+ spa->spa_claiming = B_FALSE;
+
+ spa_set_log_state(spa, SPA_LOG_GOOD);
+}
+
+static void
+spa_ld_check_for_config_update(spa_t *spa, uint64_t config_cache_txg,
+ boolean_t update_config_cache)
+{
+ vdev_t *rvd = spa->spa_root_vdev;
+ int need_update = B_FALSE;
+
+ /*
+ * If the config cache is stale, or we have uninitialized
+ * metaslabs (see spa_vdev_add()), then update the config.
+ *
+ * If this is a verbatim import, trust the current
+ * in-core spa_config and update the disk labels.
+ */
+ if (update_config_cache || config_cache_txg != spa->spa_config_txg ||
+ spa->spa_load_state == SPA_LOAD_IMPORT ||
+ spa->spa_load_state == SPA_LOAD_RECOVER ||
+ (spa->spa_import_flags & ZFS_IMPORT_VERBATIM))
+ need_update = B_TRUE;
+
+ for (int c = 0; c < rvd->vdev_children; c++)
+ if (rvd->vdev_child[c]->vdev_ms_array == 0)
+ need_update = B_TRUE;
+
+ /*
+ * Update the config cache asychronously in case we're the
+ * root pool, in which case the config cache isn't writable yet.
+ */
+ if (need_update)
+ spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE);
+}
+
+static void
+spa_ld_prepare_for_reload(spa_t *spa)
+{
+ int mode = spa->spa_mode;
+ int async_suspended = spa->spa_async_suspended;
+
+ spa_unload(spa);
+ spa_deactivate(spa);
+ spa_activate(spa, mode);
+
+ /*
+ * We save the value of spa_async_suspended as it gets reset to 0 by
+ * spa_unload(). We want to restore it back to the original value before
+ * returning as we might be calling spa_async_resume() later.
+ */
+ spa->spa_async_suspended = async_suspended;
+}
+
+static int
+spa_ld_read_checkpoint_txg(spa_t *spa)
+{
+ uberblock_t checkpoint;
+ int error = 0;
+
+ ASSERT0(spa->spa_checkpoint_txg);
+ ASSERT(MUTEX_HELD(&spa_namespace_lock));
+
+ error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
+ DMU_POOL_ZPOOL_CHECKPOINT, sizeof (uint64_t),
+ sizeof (uberblock_t) / sizeof (uint64_t), &checkpoint);
+
+ if (error == ENOENT)
+ return (0);
+
+ if (error != 0)
+ return (error);
+
+ ASSERT3U(checkpoint.ub_txg, !=, 0);
+ ASSERT3U(checkpoint.ub_checkpoint_txg, !=, 0);
+ ASSERT3U(checkpoint.ub_timestamp, !=, 0);
+ spa->spa_checkpoint_txg = checkpoint.ub_txg;
+ spa->spa_checkpoint_info.sci_timestamp = checkpoint.ub_timestamp;
+
+ return (0);
+}
+
+static int
+spa_ld_mos_init(spa_t *spa, spa_import_type_t type)
+{
+ int error = 0;
+
+ ASSERT(MUTEX_HELD(&spa_namespace_lock));
+ ASSERT(spa->spa_config_source != SPA_CONFIG_SRC_NONE);
+
+ /*
+ * Never trust the config that is provided unless we are assembling
+ * a pool following a split.
+ * This means don't trust blkptrs and the vdev tree in general. This
+ * also effectively puts the spa in read-only mode since
+ * spa_writeable() checks for spa_trust_config to be true.
+ * We will later load a trusted config from the MOS.
+ */
+ if (type != SPA_IMPORT_ASSEMBLE)
+ spa->spa_trust_config = B_FALSE;
+
+ /*
+ * Parse the config provided to create a vdev tree.
+ */
+ error = spa_ld_parse_config(spa, type);
+ if (error != 0)
+ return (error);
+
+ /*
+ * Now that we have the vdev tree, try to open each vdev. This involves
+ * opening the underlying physical device, retrieving its geometry and
+ * probing the vdev with a dummy I/O. The state of each vdev will be set
+ * based on the success of those operations. After this we'll be ready
+ * to read from the vdevs.
+ */
+ error = spa_ld_open_vdevs(spa);
+ if (error != 0)
+ return (error);
+
+ /*
+ * Read the label of each vdev and make sure that the GUIDs stored
+ * there match the GUIDs in the config provided.
+ * If we're assembling a new pool that's been split off from an
+ * existing pool, the labels haven't yet been updated so we skip
+ * validation for now.
+ */
+ if (type != SPA_IMPORT_ASSEMBLE) {
+ error = spa_ld_validate_vdevs(spa);
+ if (error != 0)
+ return (error);
+ }
+
+ /*
+ * Read all vdev labels to find the best uberblock (i.e. latest,
+ * unless spa_load_max_txg is set) and store it in spa_uberblock. We
+ * get the list of features required to read blkptrs in the MOS from
+ * the vdev label with the best uberblock and verify that our version
+ * of zfs supports them all.
+ */
+ error = spa_ld_select_uberblock(spa, type);
+ if (error != 0)
+ return (error);
+
+ /*
+ * Pass that uberblock to the dsl_pool layer which will open the root
+ * blkptr. This blkptr points to the latest version of the MOS and will
+ * allow us to read its contents.
+ */
+ error = spa_ld_open_rootbp(spa);
+ if (error != 0)
+ return (error);
+
+ return (0);
+}
+
+static int
+spa_ld_checkpoint_rewind(spa_t *spa)
+{
+ uberblock_t checkpoint;
+ int error = 0;
+
+ ASSERT(MUTEX_HELD(&spa_namespace_lock));
+ ASSERT(spa->spa_import_flags & ZFS_IMPORT_CHECKPOINT);
+
+ error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
+ DMU_POOL_ZPOOL_CHECKPOINT, sizeof (uint64_t),
+ sizeof (uberblock_t) / sizeof (uint64_t), &checkpoint);
+
+ if (error != 0) {
+ spa_load_failed(spa, "unable to retrieve checkpointed "
+ "uberblock from the MOS config [error=%d]", error);
+
+ if (error == ENOENT)
+ error = ZFS_ERR_NO_CHECKPOINT;
+
+ return (error);
+ }
+
+ ASSERT3U(checkpoint.ub_txg, <, spa->spa_uberblock.ub_txg);
+ ASSERT3U(checkpoint.ub_txg, ==, checkpoint.ub_checkpoint_txg);
+
+ /*
+ * We need to update the txg and timestamp of the checkpointed
+ * uberblock to be higher than the latest one. This ensures that
+ * the checkpointed uberblock is selected if we were to close and
+ * reopen the pool right after we've written it in the vdev labels.
+ * (also see block comment in vdev_uberblock_compare)
+ */
+ checkpoint.ub_txg = spa->spa_uberblock.ub_txg + 1;
+ checkpoint.ub_timestamp = gethrestime_sec();
+
+ /*
+ * Set current uberblock to be the checkpointed uberblock.
+ */
+ spa->spa_uberblock = checkpoint;
+
+ /*
+ * If we are doing a normal rewind, then the pool is open for
+ * writing and we sync the "updated" checkpointed uberblock to
+ * disk. Once this is done, we've basically rewound the whole
+ * pool and there is no way back.
+ *
+ * There are cases when we don't want to attempt and sync the
+ * checkpointed uberblock to disk because we are opening a
+ * pool as read-only. Specifically, verifying the checkpointed
+ * state with zdb, and importing the checkpointed state to get
+ * a "preview" of its content.
+ */
+ if (spa_writeable(spa)) {
+ vdev_t *rvd = spa->spa_root_vdev;
+
+ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
+ vdev_t *svd[SPA_SYNC_MIN_VDEVS] = { NULL };
+ int svdcount = 0;
+ int children = rvd->vdev_children;
+ int c0 = spa_get_random(children);
+
+ for (int c = 0; c < children; c++) {
+ vdev_t *vd = rvd->vdev_child[(c0 + c) % children];
+
+ /* Stop when revisiting the first vdev */
+ if (c > 0 && svd[0] == vd)
+ break;
+
+ if (vd->vdev_ms_array == 0 || vd->vdev_islog ||
+ !vdev_is_concrete(vd))
+ continue;
+
+ svd[svdcount++] = vd;
+ if (svdcount == SPA_SYNC_MIN_VDEVS)
+ break;
+ }
+ error = vdev_config_sync(svd, svdcount, spa->spa_first_txg);
+ if (error == 0)
+ spa->spa_last_synced_guid = rvd->vdev_guid;
+ spa_config_exit(spa, SCL_ALL, FTAG);
+
+ if (error != 0) {
+ spa_load_failed(spa, "failed to write checkpointed "
+ "uberblock to the vdev labels [error=%d]", error);
+ return (error);
+ }
+ }
+
+ return (0);
+}
+
+static int
+spa_ld_mos_with_trusted_config(spa_t *spa, spa_import_type_t type,
+ boolean_t *update_config_cache)
+{
+ int error;
/*
- * If we're assembling the pool from the split-off vdevs of
- * an existing pool, we don't want to attach the spares & cache
- * devices.
+ * Parse the config for pool, open and validate vdevs,
+ * select an uberblock, and use that uberblock to open
+ * the MOS.
*/
+ error = spa_ld_mos_init(spa, type);
+ if (error != 0)
+ return (error);
/*
- * Load any hot spares for this pool.
+ * Retrieve the trusted config stored in the MOS and use it to create
+ * a new, exact version of the vdev tree, then reopen all vdevs.
*/
- error = spa_dir_prop(spa, DMU_POOL_SPARES, &spa->spa_spares.sav_object);
- if (error != 0 && error != ENOENT)
- return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
- if (error == 0 && type != SPA_IMPORT_ASSEMBLE) {
- ASSERT(spa_version(spa) >= SPA_VERSION_SPARES);
- if (load_nvlist(spa, spa->spa_spares.sav_object,
- &spa->spa_spares.sav_config) != 0)
- return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ error = spa_ld_trusted_config(spa, type, B_FALSE);
+ if (error == EAGAIN) {
+ if (update_config_cache != NULL)
+ *update_config_cache = B_TRUE;
- spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
- spa_load_spares(spa);
- spa_config_exit(spa, SCL_ALL, FTAG);
- } else if (error == 0) {
- spa->spa_spares.sav_sync = B_TRUE;
- }
+ /*
+ * Redo the loading process with the trusted config if it is
+ * too different from the untrusted config.
+ */
+ spa_ld_prepare_for_reload(spa);
+ spa_load_note(spa, "RELOADING");
+ error = spa_ld_mos_init(spa, type);
+ if (error != 0)
+ return (error);
- /*
- * Load any level 2 ARC devices for this pool.
- */
- error = spa_dir_prop(spa, DMU_POOL_L2CACHE,
- &spa->spa_l2cache.sav_object);
- if (error != 0 && error != ENOENT)
- return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
- if (error == 0 && type != SPA_IMPORT_ASSEMBLE) {
- ASSERT(spa_version(spa) >= SPA_VERSION_L2CACHE);
- if (load_nvlist(spa, spa->spa_l2cache.sav_object,
- &spa->spa_l2cache.sav_config) != 0)
- return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ error = spa_ld_trusted_config(spa, type, B_TRUE);
+ if (error != 0)
+ return (error);
- spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
- spa_load_l2cache(spa);
- spa_config_exit(spa, SCL_ALL, FTAG);
- } else if (error == 0) {
- spa->spa_l2cache.sav_sync = B_TRUE;
+ } else if (error != 0) {
+ return (error);
}
- spa->spa_delegation = zpool_prop_default_numeric(ZPOOL_PROP_DELEGATION);
+ return (0);
+}
- error = spa_dir_prop(spa, DMU_POOL_PROPS, &spa->spa_pool_props_object);
- if (error && error != ENOENT)
- return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+/*
+ * Load an existing storage pool, using the config provided. This config
+ * describes which vdevs are part of the pool and is later validated against
+ * partial configs present in each vdev's label and an entire copy of the
+ * config stored in the MOS.
+ */
+static int
+spa_load_impl(spa_t *spa, spa_import_type_t type, char **ereport)
+{
+ int error = 0;
+ boolean_t missing_feat_write = B_FALSE;
+ boolean_t checkpoint_rewind =
+ (spa->spa_import_flags & ZFS_IMPORT_CHECKPOINT);
+ boolean_t update_config_cache = B_FALSE;
- if (error == 0) {
- uint64_t autoreplace = 0;
+ ASSERT(MUTEX_HELD(&spa_namespace_lock));
+ ASSERT(spa->spa_config_source != SPA_CONFIG_SRC_NONE);
- spa_prop_find(spa, ZPOOL_PROP_BOOTFS, &spa->spa_bootfs);
- spa_prop_find(spa, ZPOOL_PROP_AUTOREPLACE, &autoreplace);
- spa_prop_find(spa, ZPOOL_PROP_DELEGATION, &spa->spa_delegation);
- spa_prop_find(spa, ZPOOL_PROP_FAILUREMODE, &spa->spa_failmode);
- spa_prop_find(spa, ZPOOL_PROP_AUTOEXPAND, &spa->spa_autoexpand);
- spa_prop_find(spa, ZPOOL_PROP_MULTIHOST, &spa->spa_multihost);
- spa_prop_find(spa, ZPOOL_PROP_DEDUPDITTO,
- &spa->spa_dedup_ditto);
+ spa_load_note(spa, "LOADING");
- spa->spa_autoreplace = (autoreplace != 0);
- }
+ error = spa_ld_mos_with_trusted_config(spa, type, &update_config_cache);
+ if (error != 0)
+ return (error);
/*
- * If the 'multihost' property is set, then never allow a pool to
- * be imported when the system hostid is zero. The exception to
- * this rule is zdb which is always allowed to access pools.
+ * If we are rewinding to the checkpoint then we need to repeat
+ * everything we've done so far in this function but this time
+ * selecting the checkpointed uberblock and using that to open
+ * the MOS.
*/
- if (spa_multihost(spa) && spa_get_hostid() == 0 &&
- (spa->spa_import_flags & ZFS_IMPORT_SKIP_MMP) == 0) {
- fnvlist_add_uint64(spa->spa_load_info,
- ZPOOL_CONFIG_MMP_STATE, MMP_STATE_NO_HOSTID);
- return (spa_vdev_err(rvd, VDEV_AUX_ACTIVE, EREMOTEIO));
- }
+ if (checkpoint_rewind) {
+ /*
+ * If we are rewinding to the checkpoint update config cache
+ * anyway.
+ */
+ update_config_cache = B_TRUE;
- /*
- * If the 'autoreplace' property is set, then post a resource notifying
- * the ZFS DE that it should not issue any faults for unopenable
- * devices. We also iterate over the vdevs, and post a sysevent for any
- * unopenable vdevs so that the normal autoreplace handler can take
- * over.
- */
- if (spa->spa_autoreplace && state != SPA_LOAD_TRYIMPORT) {
- spa_check_removed(spa->spa_root_vdev);
/*
- * For the import case, this is done in spa_import(), because
- * at this point we're using the spare definitions from
- * the MOS config, not necessarily from the userland config.
+ * Extract the checkpointed uberblock from the current MOS
+ * and use this as the pool's uberblock from now on. If the
+ * pool is imported as writeable we also write the checkpoint
+ * uberblock to the labels, making the rewind permanent.
*/
- if (state != SPA_LOAD_IMPORT) {
- spa_aux_check_removed(&spa->spa_spares);
- spa_aux_check_removed(&spa->spa_l2cache);
- }
+ error = spa_ld_checkpoint_rewind(spa);
+ if (error != 0)
+ return (error);
+
+ /*
+ * Redo the loading process process again with the
+ * checkpointed uberblock.
+ */
+ spa_ld_prepare_for_reload(spa);
+ spa_load_note(spa, "LOADING checkpointed uberblock");
+ error = spa_ld_mos_with_trusted_config(spa, type, NULL);
+ if (error != 0)
+ return (error);
}
/*
- * Load the vdev state for all toplevel vdevs.
+ * Retrieve the checkpoint txg if the pool has a checkpoint.
*/
- error = vdev_load(rvd);
- if (error != 0) {
- return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error));
- }
+ error = spa_ld_read_checkpoint_txg(spa);
+ if (error != 0)
+ return (error);
- error = spa_condense_init(spa);
- if (error != 0) {
- return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error));
- }
+ /*
+ * Retrieve the mapping of indirect vdevs. Those vdevs were removed
+ * from the pool and their contents were re-mapped to other vdevs. Note
+ * that everything that we read before this step must have been
+ * rewritten on concrete vdevs after the last device removal was
+ * initiated. Otherwise we could be reading from indirect vdevs before
+ * we have loaded their mappings.
+ */
+ error = spa_ld_open_indirect_vdev_metadata(spa);
+ if (error != 0)
+ return (error);
/*
- * Propagate the leaf DTLs we just loaded all the way up the tree.
+ * Retrieve the full list of active features from the MOS and check if
+ * they are all supported.
*/
- spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
- vdev_dtl_reassess(rvd, 0, 0, B_FALSE);
- spa_config_exit(spa, SCL_ALL, FTAG);
+ error = spa_ld_check_features(spa, &missing_feat_write);
+ if (error != 0)
+ return (error);
/*
- * Load the DDTs (dedup tables).
+ * Load several special directories from the MOS needed by the dsl_pool
+ * layer.
*/
- error = ddt_load(spa);
+ error = spa_ld_load_special_directories(spa);
if (error != 0)
- return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+ return (error);
- spa_update_dspace(spa);
+ /*
+ * Retrieve pool properties from the MOS.
+ */
+ error = spa_ld_get_props(spa);
+ if (error != 0)
+ return (error);
- if (type != SPA_IMPORT_ASSEMBLE && spa_writeable(spa) &&
- spa_check_logs(spa)) {
- *ereport = FM_EREPORT_ZFS_LOG_REPLAY;
- return (spa_vdev_err(rvd, VDEV_AUX_BAD_LOG, ENXIO));
- }
+ /*
+ * Retrieve the list of auxiliary devices - cache devices and spares -
+ * and open them.
+ */
+ error = spa_ld_open_aux_vdevs(spa, type);
+ if (error != 0)
+ return (error);
+
+ /*
+ * Load the metadata for all vdevs. Also check if unopenable devices
+ * should be autoreplaced.
+ */
+ error = spa_ld_load_vdev_metadata(spa);
+ if (error != 0)
+ return (error);
+
+ error = spa_ld_load_dedup_tables(spa);
+ if (error != 0)
+ return (error);
+
+ /*
+ * Verify the logs now to make sure we don't have any unexpected errors
+ * when we claim log blocks later.
+ */
+ error = spa_ld_verify_logs(spa, type, ereport);
+ if (error != 0)
+ return (error);
if (missing_feat_write) {
- ASSERT(state == SPA_LOAD_TRYIMPORT);
+ ASSERT(spa->spa_load_state == SPA_LOAD_TRYIMPORT);
/*
* At this point, we know that we can open the pool in
* read-only mode but not read-write mode. We now have enough
* information and can return to userland.
*/
- return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT, ENOTSUP));
+ return (spa_vdev_err(spa->spa_root_vdev, VDEV_AUX_UNSUP_FEAT,
+ ENOTSUP));
}
/*
- * We've successfully opened the pool, verify that we're ready
- * to start pushing transactions.
+ * Traverse the last txgs to make sure the pool was left off in a safe
+ * state. When performing an extreme rewind, we verify the whole pool,
+ * which can take a very long time.
*/
- if (state != SPA_LOAD_TRYIMPORT) {
- if ((error = spa_load_verify(spa)))
- return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA,
- error));
- }
+ error = spa_ld_verify_pool_data(spa);
+ if (error != 0)
+ return (error);
- if (spa_writeable(spa) && (state == SPA_LOAD_RECOVER ||
+ /*
+ * Calculate the deflated space for the pool. This must be done before
+ * we write anything to the pool because we'd need to update the space
+ * accounting using the deflated sizes.
+ */
+ spa_update_dspace(spa);
+
+ /*
+ * We have now retrieved all the information we needed to open the
+ * pool. If we are importing the pool in read-write mode, a few
+ * additional steps must be performed to finish the import.
+ */
+ if (spa_writeable(spa) && (spa->spa_load_state == SPA_LOAD_RECOVER ||
spa->spa_load_max_txg == UINT64_MAX)) {
- dmu_tx_t *tx;
- int need_update = B_FALSE;
- dsl_pool_t *dp = spa_get_dsl(spa);
+ uint64_t config_cache_txg = spa->spa_config_txg;
+
+ ASSERT(spa->spa_load_state != SPA_LOAD_TRYIMPORT);
/*
- * We must check this before we start the sync thread, because
- * we only want to start a condense thread for condense
- * operations that were in progress when the pool was
- * imported. Once we start syncing, spa_sync() could
- * initiate a condense (and start a thread for it). In
- * that case it would be wrong to start a second
- * condense thread.
+ * In case of a checkpoint rewind, log the original txg
+ * of the checkpointed uberblock.
*/
- boolean_t condense_in_progress =
- (spa->spa_condensing_indirect != NULL);
-
- ASSERT(state != SPA_LOAD_TRYIMPORT);
+ if (checkpoint_rewind) {
+ spa_history_log_internal(spa, "checkpoint rewind",
+ NULL, "rewound state to txg=%llu",
+ (u_longlong_t)spa->spa_uberblock.ub_checkpoint_txg);
+ }
/*
- * Claim log blocks that haven't been committed yet.
- * This must all happen in a single txg.
- * Note: spa_claim_max_txg is updated by spa_claim_notify(),
- * invoked from zil_claim_log_block()'s i/o done callback.
- * Price of rollback is that we abandon the log.
+ * Traverse the ZIL and claim all blocks.
*/
- spa->spa_claiming = B_TRUE;
-
- tx = dmu_tx_create_assigned(dp, spa_first_txg(spa));
- (void) dmu_objset_find_dp(dp, dp->dp_root_dir_obj,
- zil_claim, tx, DS_FIND_CHILDREN);
- dmu_tx_commit(tx);
-
- spa->spa_claiming = B_FALSE;
+ spa_ld_claim_log_blocks(spa);
- spa_set_log_state(spa, SPA_LOG_GOOD);
+ /*
+ * Kick-off the syncing thread.
+ */
spa->spa_sync_on = B_TRUE;
txg_sync_start(spa->spa_dsl_pool);
mmp_thread_start(spa);
* Wait for all claims to sync. We sync up to the highest
* claimed log block birth time so that claimed log blocks
* don't appear to be from the future. spa_claim_max_txg
- * will have been set for us by either zil_check_log_chain()
- * (invoked from spa_check_logs()) or zil_claim() above.
+ * will have been set for us by ZIL traversal operations
+ * performed above.
*/
txg_wait_synced(spa->spa_dsl_pool, spa->spa_claim_max_txg);
/*
- * If the config cache is stale, or we have uninitialized
- * metaslabs (see spa_vdev_add()), then update the config.
- *
- * If this is a verbatim import, trust the current
- * in-core spa_config and update the disk labels.
- */
- if (config_cache_txg != spa->spa_config_txg ||
- state == SPA_LOAD_IMPORT ||
- state == SPA_LOAD_RECOVER ||
- (spa->spa_import_flags & ZFS_IMPORT_VERBATIM))
- need_update = B_TRUE;
-
- for (int c = 0; c < rvd->vdev_children; c++)
- if (rvd->vdev_child[c]->vdev_ms_array == 0)
- need_update = B_TRUE;
-
- /*
- * Update the config cache asychronously in case we're the
- * root pool, in which case the config cache isn't writable yet.
+ * Check if we need to request an update of the config. On the
+ * next sync, we would update the config stored in vdev labels
+ * and the cachefile (by default /etc/zfs/zpool.cache).
*/
- if (need_update)
- spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE);
+ spa_ld_check_for_config_update(spa, config_cache_txg,
+ update_config_cache);
/*
* Check all DTLs to see if anything needs resilvering.
*/
if (!dsl_scan_resilvering(spa->spa_dsl_pool) &&
- vdev_resilver_needed(rvd, NULL, NULL))
+ vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL))
spa_async_request(spa, SPA_ASYNC_RESILVER);
/*
*/
dsl_pool_clean_tmp_userrefs(spa->spa_dsl_pool);
- /*
- * Note: unlike condensing, we don't need an analogous
- * "removal_in_progress" dance because no other thread
- * can start a removal while we hold the spa_namespace_lock.
- */
spa_restart_removal(spa);
- if (condense_in_progress)
- spa_condense_indirect_restart(spa);
+ spa_spawn_aux_threads(spa);
}
+ spa_load_note(spa, "LOADED");
+
return (0);
}
static int
-spa_load_retry(spa_t *spa, spa_load_state_t state, int mosconfig)
+spa_load_retry(spa_t *spa, spa_load_state_t state)
{
int mode = spa->spa_mode;
spa_activate(spa, mode);
spa_async_suspend(spa);
- return (spa_load(spa, state, SPA_IMPORT_EXISTING, mosconfig));
+ spa_load_note(spa, "spa_load_retry: rewind, max txg: %llu",
+ (u_longlong_t)spa->spa_load_max_txg);
+
+ return (spa_load(spa, state, SPA_IMPORT_EXISTING));
}
/*
* spa_load().
*/
static int
-spa_load_best(spa_t *spa, spa_load_state_t state, int mosconfig,
- uint64_t max_request, int rewind_flags)
+spa_load_best(spa_t *spa, spa_load_state_t state, uint64_t max_request,
+ int rewind_flags)
{
nvlist_t *loadinfo = NULL;
nvlist_t *config = NULL;
spa->spa_extreme_rewind = B_TRUE;
}
- load_error = rewind_error = spa_load(spa, state, SPA_IMPORT_EXISTING,
- mosconfig);
+ load_error = rewind_error = spa_load(spa, state, SPA_IMPORT_EXISTING);
if (load_error == 0)
return (0);
+ if (load_error == ZFS_ERR_NO_CHECKPOINT) {
+ /*
+ * When attempting checkpoint-rewind on a pool with no
+ * checkpoint, we should not attempt to load uberblocks
+ * from previous txgs when spa_load fails.
+ */
+ ASSERT(spa->spa_import_flags & ZFS_IMPORT_CHECKPOINT);
+ return (load_error);
+ }
if (spa->spa_root_vdev != NULL)
config = spa_config_generate(spa, NULL, -1ULL, B_TRUE);
spa->spa_uberblock.ub_txg <= spa->spa_load_max_txg) {
if (spa->spa_load_max_txg < safe_rewind_txg)
spa->spa_extreme_rewind = B_TRUE;
- rewind_error = spa_load_retry(spa, state, mosconfig);
+ rewind_error = spa_load_retry(spa, state);
}
spa->spa_extreme_rewind = B_FALSE;
}
if (spa->spa_state == POOL_STATE_UNINITIALIZED) {
- zpool_rewind_policy_t policy;
+ zpool_load_policy_t policy;
firstopen = B_TRUE;
- zpool_get_rewind_policy(nvpolicy ? nvpolicy : spa->spa_config,
+ zpool_get_load_policy(nvpolicy ? nvpolicy : spa->spa_config,
&policy);
- if (policy.zrp_request & ZPOOL_DO_REWIND)
+ if (policy.zlp_rewind & ZPOOL_DO_REWIND)
state = SPA_LOAD_RECOVER;
spa_activate(spa, spa_mode_global);
if (state != SPA_LOAD_RECOVER)
spa->spa_last_ubsync_txg = spa->spa_load_txg = 0;
+ spa->spa_config_source = SPA_CONFIG_SRC_CACHEFILE;
- error = spa_load_best(spa, state, B_FALSE, policy.zrp_txg,
- policy.zrp_request);
+ zfs_dbgmsg("spa_open_common: opening %s", pool);
+ error = spa_load_best(spa, state, policy.zlp_txg,
+ policy.zlp_rewind);
if (error == EBADF) {
/*
!has_encryption)
return (SET_ERROR(ENOTSUP));
- return (dmu_objset_create_crypt_check(NULL, dcp));
+ return (dmu_objset_create_crypt_check(NULL, dcp, NULL));
}
/*
char *poolname;
nvlist_t *nvl;
- if (nvlist_lookup_string(props, "tname", &poolname) != 0)
+ if (props == NULL ||
+ nvlist_lookup_string(props, "tname", &poolname) != 0)
poolname = (char *)pool;
/*
(error = vdev_create(rvd, txg, B_FALSE)) == 0 &&
(error = spa_validate_aux(spa, nvroot, txg,
VDEV_ALLOC_ADD)) == 0) {
- for (int c = 0; c < rvd->vdev_children; c++) {
- vdev_metaslab_set_size(rvd->vdev_child[c]);
- vdev_expand(rvd->vdev_child[c], txg);
+ /*
+ * instantiate the metaslab groups (this will dirty the vdevs)
+ * we can no longer error exit past this point
+ */
+ for (int c = 0; error == 0 && c < rvd->vdev_children; c++) {
+ vdev_t *vd = rvd->vdev_child[c];
+
+ vdev_metaslab_set_size(vd);
+ vdev_expand(vd, txg);
}
}
if (dp->dp_root_dir->dd_crypto_obj != 0)
VERIFY0(spa_keystore_remove_mapping(spa, root_dsobj, FTAG));
+ spa_spawn_aux_threads(spa);
+
spa_write_cachefile(spa, B_FALSE, B_TRUE);
/*
* and are making their way through the eviction process.
*/
spa_evicting_os_wait(spa);
- spa->spa_minref = refcount_count(&spa->spa_refcount);
+ spa->spa_minref = zfs_refcount_count(&spa->spa_refcount);
spa->spa_load_state = SPA_LOAD_NONE;
mutex_exit(&spa_namespace_lock);
spa_t *spa;
char *altroot = NULL;
spa_load_state_t state = SPA_LOAD_IMPORT;
- zpool_rewind_policy_t policy;
+ zpool_load_policy_t policy;
uint64_t mode = spa_mode_global;
uint64_t readonly = B_FALSE;
int error;
spa_write_cachefile(spa, B_FALSE, B_TRUE);
spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_IMPORT);
-
+ zfs_dbgmsg("spa_import: verbatim import of %s", pool);
mutex_exit(&spa_namespace_lock);
return (0);
}
*/
spa_async_suspend(spa);
- zpool_get_rewind_policy(config, &policy);
- if (policy.zrp_request & ZPOOL_DO_REWIND)
+ zpool_get_load_policy(config, &policy);
+ if (policy.zlp_rewind & ZPOOL_DO_REWIND)
state = SPA_LOAD_RECOVER;
- /*
- * Pass off the heavy lifting to spa_load(). Pass TRUE for mosconfig
- * because the user-supplied config is actually the one to trust when
- * doing an import.
- */
- if (state != SPA_LOAD_RECOVER)
- spa->spa_last_ubsync_txg = spa->spa_load_txg = 0;
+ spa->spa_config_source = SPA_CONFIG_SRC_TRYIMPORT;
- error = spa_load_best(spa, state, B_TRUE, policy.zrp_txg,
- policy.zrp_request);
+ if (state != SPA_LOAD_RECOVER) {
+ spa->spa_last_ubsync_txg = spa->spa_load_txg = 0;
+ zfs_dbgmsg("spa_import: importing %s", pool);
+ } else {
+ zfs_dbgmsg("spa_import: importing %s, max_txg=%lld "
+ "(RECOVERY MODE)", pool, (longlong_t)policy.zlp_txg);
+ }
+ error = spa_load_best(spa, state, policy.zlp_txg, policy.zlp_rewind);
/*
* Propagate anything learned while loading the pool and pass it
spa_tryimport(nvlist_t *tryconfig)
{
nvlist_t *config = NULL;
- char *poolname;
+ char *poolname, *cachefile;
spa_t *spa;
uint64_t state;
int error;
+ zpool_load_policy_t policy;
if (nvlist_lookup_string(tryconfig, ZPOOL_CONFIG_POOL_NAME, &poolname))
return (NULL);
spa_activate(spa, FREAD);
/*
- * Pass off the heavy lifting to spa_load().
- * Pass TRUE for mosconfig because the user-supplied config
- * is actually the one to trust when doing an import.
+ * Rewind pool if a max txg was provided.
*/
- error = spa_load(spa, SPA_LOAD_TRYIMPORT, SPA_IMPORT_EXISTING, B_TRUE);
+ zpool_get_load_policy(spa->spa_config, &policy);
+ if (policy.zlp_txg != UINT64_MAX) {
+ spa->spa_load_max_txg = policy.zlp_txg;
+ spa->spa_extreme_rewind = B_TRUE;
+ zfs_dbgmsg("spa_tryimport: importing %s, max_txg=%lld",
+ poolname, (longlong_t)policy.zlp_txg);
+ } else {
+ zfs_dbgmsg("spa_tryimport: importing %s", poolname);
+ }
+
+ if (nvlist_lookup_string(tryconfig, ZPOOL_CONFIG_CACHEFILE, &cachefile)
+ == 0) {
+ zfs_dbgmsg("spa_tryimport: using cachefile '%s'", cachefile);
+ spa->spa_config_source = SPA_CONFIG_SRC_CACHEFILE;
+ } else {
+ spa->spa_config_source = SPA_CONFIG_SRC_SCAN;
+ }
+
+ error = spa_load(spa, SPA_LOAD_TRYIMPORT, SPA_IMPORT_EXISTING);
/*
* If 'tryconfig' was at least parsable, return the current config.
for (int c = 0; c < vd->vdev_children; c++) {
tvd = vd->vdev_child[c];
if (spa->spa_vdev_removal != NULL &&
- tvd->vdev_ashift !=
- spa->spa_vdev_removal->svr_vdev->vdev_ashift) {
+ tvd->vdev_ashift != spa->spa_max_ashift) {
return (spa_vdev_exit(spa, vd, txg, EINVAL));
}
/* Fail if top level vdev is raidz */
oldvd = spa_lookup_by_guid(spa, guid, B_FALSE);
- if (spa->spa_vdev_removal != NULL ||
- spa->spa_removing_phys.sr_prev_indirect_vdev != -1) {
- return (spa_vdev_exit(spa, NULL, txg, EBUSY));
+ ASSERT(MUTEX_HELD(&spa_namespace_lock));
+ if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) {
+ error = (spa_has_checkpoint(spa)) ?
+ ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT;
+ return (spa_vdev_exit(spa, NULL, txg, error));
}
+ if (spa->spa_vdev_removal != NULL)
+ return (spa_vdev_exit(spa, NULL, txg, EBUSY));
+
if (oldvd == NULL)
return (spa_vdev_exit(spa, NULL, txg, ENODEV));
vd = spa_lookup_by_guid(spa, guid, B_FALSE);
+ /*
+ * Besides being called directly from the userland through the
+ * ioctl interface, spa_vdev_detach() can be potentially called
+ * at the end of spa_vdev_resilver_done().
+ *
+ * In the regular case, when we have a checkpoint this shouldn't
+ * happen as we never empty the DTLs of a vdev during the scrub
+ * [see comment in dsl_scan_done()]. Thus spa_vdev_resilvering_done()
+ * should never get here when we have a checkpoint.
+ *
+ * That said, even in a case when we checkpoint the pool exactly
+ * as spa_vdev_resilver_done() calls this function everything
+ * should be fine as the resilver will return right away.
+ */
+ ASSERT(MUTEX_HELD(&spa_namespace_lock));
+ if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) {
+ error = (spa_has_checkpoint(spa)) ?
+ ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT;
+ return (spa_vdev_exit(spa, NULL, txg, error));
+ }
+
if (vd == NULL)
return (spa_vdev_exit(spa, NULL, txg, ENODEV));
txg = spa_vdev_enter(spa);
+ ASSERT(MUTEX_HELD(&spa_namespace_lock));
+ if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) {
+ error = (spa_has_checkpoint(spa)) ?
+ ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT;
+ return (spa_vdev_exit(spa, NULL, txg, error));
+ }
+
/* clear the log and flush everything up to now */
activate_slog = spa_passivate_log(spa);
(void) spa_vdev_config_exit(spa, NULL, txg, 0, FTAG);
break;
}
- if (vdev_dtl_required(vml[c])) {
+ if (vdev_dtl_required(vml[c]) ||
+ vdev_resilver_needed(vml[c], NULL, NULL)) {
error = SET_ERROR(EBUSY);
break;
}
spa_activate(newspa, spa_mode_global);
spa_async_suspend(newspa);
+ newspa->spa_config_source = SPA_CONFIG_SRC_SPLIT;
+
/* create the new pool from the disks of the original pool */
- error = spa_load(newspa, SPA_LOAD_IMPORT, SPA_IMPORT_ASSEMBLE, B_TRUE);
+ error = spa_load(newspa, SPA_LOAD_IMPORT, SPA_IMPORT_ASSEMBLE);
if (error)
goto out;
mutex_enter(&spa_namespace_lock);
old_space = metaslab_class_get_space(spa_normal_class(spa));
+ old_space += metaslab_class_get_space(spa_special_class(spa));
+ old_space += metaslab_class_get_space(spa_dedup_class(spa));
+
spa_config_update(spa, SPA_CONFIG_UPDATE_POOL);
+
new_space = metaslab_class_get_space(spa_normal_class(spa));
+ new_space += metaslab_class_get_space(spa_special_class(spa));
+ new_space += metaslab_class_get_space(spa_dedup_class(spa));
mutex_exit(&spa_namespace_lock);
/*
{
mutex_enter(&spa->spa_async_lock);
spa->spa_async_suspended++;
- while (spa->spa_async_thread != NULL ||
- spa->spa_condense_thread != NULL)
+ while (spa->spa_async_thread != NULL)
cv_wait(&spa->spa_async_cv, &spa->spa_async_lock);
mutex_exit(&spa->spa_async_lock);
spa_vdev_remove_suspend(spa);
+
+ zthr_t *condense_thread = spa->spa_condense_zthr;
+ if (condense_thread != NULL && zthr_isrunning(condense_thread))
+ VERIFY0(zthr_cancel(condense_thread));
+
+ zthr_t *discard_thread = spa->spa_checkpoint_discard_zthr;
+ if (discard_thread != NULL && zthr_isrunning(discard_thread))
+ VERIFY0(zthr_cancel(discard_thread));
}
void
spa->spa_async_suspended--;
mutex_exit(&spa->spa_async_lock);
spa_restart_removal(spa);
+
+ zthr_t *condense_thread = spa->spa_condense_zthr;
+ if (condense_thread != NULL && !zthr_isrunning(condense_thread))
+ zthr_resume(condense_thread);
+
+ zthr_t *discard_thread = spa->spa_checkpoint_discard_zthr;
+ if (discard_thread != NULL && !zthr_isrunning(discard_thread))
+ zthr_resume(discard_thread);
}
static boolean_t
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 *normal = spa_normal_class(spa);
+ metaslab_class_t *special = spa_special_class(spa);
+ metaslab_class_t *dedup = spa_dedup_class(spa);
vdev_t *rvd = spa->spa_root_vdev;
vdev_t *vd;
dmu_tx_t *tx;
spa->spa_syncing_txg = txg;
spa->spa_sync_pass = 0;
- mutex_enter(&spa->spa_alloc_lock);
- VERIFY0(avl_numnodes(&spa->spa_alloc_tree));
- mutex_exit(&spa->spa_alloc_lock);
+ for (int i = 0; i < spa->spa_alloc_count; i++) {
+ mutex_enter(&spa->spa_alloc_locks[i]);
+ VERIFY0(avl_numnodes(&spa->spa_alloc_trees[i]));
+ mutex_exit(&spa->spa_alloc_locks[i]);
+ }
/*
* If there are any pending vdev state changes, convert them
* The max queue depth will not change in the middle of syncing
* out this txg.
*/
- uint64_t queue_depth_total = 0;
+ uint64_t slots_per_allocator = 0;
for (int c = 0; c < rvd->vdev_children; c++) {
vdev_t *tvd = rvd->vdev_child[c];
metaslab_group_t *mg = tvd->vdev_mg;
+ metaslab_class_t *mc;
+
+ if (mg == NULL || !metaslab_group_initialized(mg))
+ continue;
- if (mg == NULL || mg->mg_class != spa_normal_class(spa) ||
- !metaslab_group_initialized(mg))
+ mc = mg->mg_class;
+ if (mc != normal && mc != special && mc != dedup)
continue;
/*
* allocations look at mg_max_alloc_queue_depth, and async
* allocations all happen from spa_sync().
*/
- ASSERT0(refcount_count(&mg->mg_alloc_queue_depth));
+ for (int i = 0; i < spa->spa_alloc_count; i++)
+ ASSERT0(zfs_refcount_count(
+ &(mg->mg_alloc_queue_depth[i])));
mg->mg_max_alloc_queue_depth = max_queue_depth;
- queue_depth_total += mg->mg_max_alloc_queue_depth;
+
+ for (int i = 0; i < spa->spa_alloc_count; i++) {
+ mg->mg_cur_max_alloc_queue_depth[i] =
+ zfs_vdev_def_queue_depth;
+ }
+ slots_per_allocator += zfs_vdev_def_queue_depth;
}
- 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;
- ASSERT3U(mc->mc_alloc_max_slots, <=,
- max_queue_depth * rvd->vdev_children);
+ for (int i = 0; i < spa->spa_alloc_count; i++) {
+ ASSERT0(zfs_refcount_count(&normal->mc_alloc_slots[i]));
+ ASSERT0(zfs_refcount_count(&special->mc_alloc_slots[i]));
+ ASSERT0(zfs_refcount_count(&dedup->mc_alloc_slots[i]));
+ normal->mc_alloc_max_slots[i] = slots_per_allocator;
+ special->mc_alloc_max_slots[i] = slots_per_allocator;
+ dedup->mc_alloc_max_slots[i] = slots_per_allocator;
+ }
+ normal->mc_alloc_throttle_enabled = zio_dva_throttle_enabled;
+ special->mc_alloc_throttle_enabled = zio_dva_throttle_enabled;
+ dedup->mc_alloc_throttle_enabled = zio_dva_throttle_enabled;
for (int c = 0; c < rvd->vdev_children; c++) {
vdev_t *vd = rvd->vdev_child[c];
txg));
ASSERT(txg_list_empty(&dp->dp_dirty_dirs, txg));
ASSERT(txg_list_empty(&dp->dp_sync_tasks, txg));
+ ASSERT(txg_list_empty(&dp->dp_early_sync_tasks,
+ txg));
break;
}
spa_sync_deferred_frees(spa, tx);
spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
if (list_is_empty(&spa->spa_config_dirty_list)) {
- vdev_t *svd[SPA_DVAS_PER_BP];
+ vdev_t *svd[SPA_SYNC_MIN_VDEVS] = { NULL };
int svdcount = 0;
int children = rvd->vdev_children;
int c0 = spa_get_random(children);
for (int c = 0; c < children; c++) {
vd = rvd->vdev_child[(c0 + c) % children];
+
+ /* Stop when revisiting the first vdev */
+ if (c > 0 && svd[0] == vd)
+ break;
+
if (vd->vdev_ms_array == 0 || vd->vdev_islog ||
!vdev_is_concrete(vd))
continue;
+
svd[svdcount++] = vd;
- if (svdcount == SPA_DVAS_PER_BP)
+ if (svdcount == SPA_SYNC_MIN_VDEVS)
break;
}
error = vdev_config_sync(svd, svdcount, txg);
dsl_pool_sync_done(dp, txg);
- mutex_enter(&spa->spa_alloc_lock);
- VERIFY0(avl_numnodes(&spa->spa_alloc_tree));
- mutex_exit(&spa->spa_alloc_lock);
+ for (int i = 0; i < spa->spa_alloc_count; i++) {
+ mutex_enter(&spa->spa_alloc_locks[i]);
+ VERIFY0(avl_numnodes(&spa->spa_alloc_trees[i]));
+ mutex_exit(&spa->spa_alloc_locks[i]);
+ }
/*
* Update usable space statistics.
ASSERT(txg_list_empty(&dp->dp_dirty_dirs, txg));
ASSERT(txg_list_empty(&spa->spa_vdev_txg_list, txg));
+ while (zfs_pause_spa_sync)
+ delay(1);
+
spa->spa_sync_pass = 0;
/*
spa_event_post(spa_event_create(spa, vd, hist_nvl, name));
}
-#if defined(_KERNEL) && defined(HAVE_SPL)
+#if defined(_KERNEL)
/* state manipulation functions */
EXPORT_SYMBOL(spa_open);
EXPORT_SYMBOL(spa_open_rewind);
EXPORT_SYMBOL(spa_event_notify);
#endif
-#if defined(_KERNEL) && defined(HAVE_SPL)
+#if defined(_KERNEL)
module_param(spa_load_verify_maxinflight, int, 0644);
MODULE_PARM_DESC(spa_load_verify_maxinflight,
"Max concurrent traversal I/Os while verifying pool during import -X");
MODULE_PARM_DESC(spa_load_verify_data,
"Set to traverse data on pool import");
+module_param(spa_load_print_vdev_tree, int, 0644);
+MODULE_PARM_DESC(spa_load_print_vdev_tree,
+ "Print vdev tree to zfs_dbgmsg during pool import");
+
/* CSTYLED */
module_param(zio_taskq_batch_pct, uint, 0444);
MODULE_PARM_DESC(zio_taskq_batch_pct,
"Percentage of CPUs to run an IO worker thread");
+/* BEGIN CSTYLED */
+module_param(zfs_max_missing_tvds, ulong, 0644);
+MODULE_PARM_DESC(zfs_max_missing_tvds,
+ "Allow importing pool with up to this number of missing top-level vdevs"
+ " (in read-only mode)");
+/* END CSTYLED */
+
#endif
projects / mirror_zfs.git / blobdiff