summaryrefslogtreecommitdiffstats
path: root/module
diff options
context:
space:
mode:
authorGeorge Wilson <george.wilson@delphix.com>2018-12-19 07:54:59 -0700
committerBrian Behlendorf <behlendorf1@llnl.gov>2019-01-07 10:37:26 -0800
commit619f09769393d4e0cbaa5f662362138e1c699159 (patch)
tree6a94855931e5bf9e353270c72fba43316a44baa6 /module
parentc87db591967507de027d6bb0c683ffd09dd70105 (diff)
OpenZFS 9102 - zfs should be able to initialize storage devices
PROBLEM ======== The first access to a block incurs a performance penalty on some platforms (e.g. AWS's EBS, VMware VMDKs). Therefore we recommend that volumes are "thick provisioned", where supported by the platform (VMware). This can create a large delay in getting a new virtual machines up and running (or adding storage to an existing Engine). If the thick provision step is omitted, write performance will be suboptimal until all blocks on the LUN have been written. SOLUTION ========= This feature introduces a way to 'initialize' the disks at install or in the background to make sure we don't incur this first read penalty. When an entire LUN is added to ZFS, we make all space available immediately, and allow ZFS to find unallocated space and zero it out. This works with concurrent writes to arbitrary offsets, ensuring that we don't zero out something that has been (or is in the middle of being) written. This scheme can also be applied to existing pools (affecting only free regions on the vdev). Detailed design: - new subcommand:zpool initialize [-cs] <pool> [<vdev> ...] - start, suspend, or cancel initialization - Creates new open-context thread for each vdev - Thread iterates through all metaslabs in this vdev - Each metaslab: - select a metaslab - load the metaslab - mark the metaslab as being zeroed - walk all free ranges within that metaslab and translate them to ranges on the leaf vdev - issue a "zeroing" I/O on the leaf vdev that corresponds to a free range on the metaslab we're working on - continue until all free ranges for this metaslab have been "zeroed" - reset/unmark the metaslab being zeroed - if more metaslabs exist, then repeat above tasks. - if no more metaslabs, then we're done. - progress for the initialization is stored on-disk in the vdev’s leaf zap object. The following information is stored: - the last offset that has been initialized - the state of the initialization process (i.e. active, suspended, or canceled) - the start time for the initialization - progress is reported via the zpool status command and shows information for each of the vdevs that are initializing Porting notes: - Added zfs_initialize_value module parameter to set the pattern written by "zpool initialize". - Added zfs_vdev_{initializing,removal}_{min,max}_active module options. Authored by: George Wilson <george.wilson@delphix.com> Reviewed by: John Wren Kennedy <john.kennedy@delphix.com> Reviewed by: Matthew Ahrens <mahrens@delphix.com> Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com> Reviewed by: Prakash Surya <prakash.surya@delphix.com> Reviewed by: loli10K <ezomori.nozomu@gmail.com> Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov> Approved by: Richard Lowe <richlowe@richlowe.net> Signed-off-by: Tim Chase <tim@chase2k.com> Ported-by: Tim Chase <tim@chase2k.com> OpenZFS-issue: https://www.illumos.org/issues/9102 OpenZFS-commit: https://github.com/openzfs/openzfs/commit/c3963210eb Closes #8230
Diffstat (limited to 'module')
-rw-r--r--module/zfs/Makefile.in1
-rw-r--r--module/zfs/metaslab.c23
-rw-r--r--module/zfs/spa.c155
-rw-r--r--module/zfs/spa_misc.c7
-rw-r--r--module/zfs/vdev.c46
-rw-r--r--module/zfs/vdev_disk.c1
-rw-r--r--module/zfs/vdev_file.c4
-rw-r--r--module/zfs/vdev_indirect.c1
-rw-r--r--module/zfs/vdev_initialize.c819
-rw-r--r--module/zfs/vdev_mirror.c3
-rw-r--r--module/zfs/vdev_missing.c4
-rw-r--r--module/zfs/vdev_queue.c38
-rw-r--r--module/zfs/vdev_raidz.c75
-rw-r--r--module/zfs/vdev_removal.c13
-rw-r--r--module/zfs/vdev_root.c3
-rw-r--r--module/zfs/zfs_ioctl.c86
-rw-r--r--module/zfs/zfs_sysfs.c1
17 files changed, 1254 insertions, 26 deletions
diff --git a/module/zfs/Makefile.in b/module/zfs/Makefile.in
index a243f51d8..193bdc510 100644
--- a/module/zfs/Makefile.in
+++ b/module/zfs/Makefile.in
@@ -89,6 +89,7 @@ $(MODULE)-objs += vdev_file.o
$(MODULE)-objs += vdev_indirect.o
$(MODULE)-objs += vdev_indirect_births.o
$(MODULE)-objs += vdev_indirect_mapping.o
+$(MODULE)-objs += vdev_initialize.o
$(MODULE)-objs += vdev_label.o
$(MODULE)-objs += vdev_mirror.o
$(MODULE)-objs += vdev_missing.o
diff --git a/module/zfs/metaslab.c b/module/zfs/metaslab.c
index 4b5baf6a6..71688b420 100644
--- a/module/zfs/metaslab.c
+++ b/module/zfs/metaslab.c
@@ -635,6 +635,8 @@ metaslab_group_create(metaslab_class_t *mc, vdev_t *vd, int allocators)
mg = kmem_zalloc(sizeof (metaslab_group_t), KM_SLEEP);
mutex_init(&mg->mg_lock, NULL, MUTEX_DEFAULT, NULL);
+ mutex_init(&mg->mg_ms_initialize_lock, NULL, MUTEX_DEFAULT, NULL);
+ cv_init(&mg->mg_ms_initialize_cv, NULL, CV_DEFAULT, NULL);
mg->mg_primaries = kmem_zalloc(allocators * sizeof (metaslab_t *),
KM_SLEEP);
mg->mg_secondaries = kmem_zalloc(allocators * sizeof (metaslab_t *),
@@ -681,6 +683,8 @@ metaslab_group_destroy(metaslab_group_t *mg)
kmem_free(mg->mg_secondaries, mg->mg_allocators *
sizeof (metaslab_t *));
mutex_destroy(&mg->mg_lock);
+ mutex_destroy(&mg->mg_ms_initialize_lock);
+ cv_destroy(&mg->mg_ms_initialize_cv);
for (int i = 0; i < mg->mg_allocators; i++) {
zfs_refcount_destroy(&mg->mg_alloc_queue_depth[i]);
@@ -1502,6 +1506,7 @@ metaslab_init(metaslab_group_t *mg, uint64_t id, uint64_t object, uint64_t txg,
mutex_init(&ms->ms_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&ms->ms_sync_lock, NULL, MUTEX_DEFAULT, NULL);
cv_init(&ms->ms_load_cv, NULL, CV_DEFAULT, NULL);
+
ms->ms_id = id;
ms->ms_start = id << vd->vdev_ms_shift;
ms->ms_size = 1ULL << vd->vdev_ms_shift;
@@ -2686,6 +2691,7 @@ metaslab_sync_done(metaslab_t *msp, uint64_t txg)
* from it in 'metaslab_unload_delay' txgs, then unload it.
*/
if (msp->ms_loaded &&
+ msp->ms_initializing == 0 &&
msp->ms_selected_txg + metaslab_unload_delay < txg) {
for (int t = 1; t < TXG_CONCURRENT_STATES; t++) {
@@ -2967,6 +2973,7 @@ metaslab_block_alloc(metaslab_t *msp, uint64_t size, uint64_t txg)
metaslab_class_t *mc = msp->ms_group->mg_class;
VERIFY(!msp->ms_condensing);
+ VERIFY0(msp->ms_initializing);
start = mc->mc_ops->msop_alloc(msp, size);
if (start != -1ULL) {
@@ -3027,9 +3034,10 @@ find_valid_metaslab(metaslab_group_t *mg, uint64_t activation_weight,
}
/*
- * If the selected metaslab is condensing, skip it.
+ * If the selected metaslab is condensing or being
+ * initialized, skip it.
*/
- if (msp->ms_condensing)
+ if (msp->ms_condensing || msp->ms_initializing > 0)
continue;
*was_active = msp->ms_allocator != -1;
@@ -3190,7 +3198,9 @@ metaslab_group_alloc_normal(metaslab_group_t *mg, zio_alloc_list_t *zal,
/*
* If this metaslab is currently condensing then pick again as
* we can't manipulate this metaslab until it's committed
- * to disk.
+ * to disk. If this metaslab is being initialized, we shouldn't
+ * allocate from it since the allocated region might be
+ * overwritten after allocation.
*/
if (msp->ms_condensing) {
metaslab_trace_add(zal, mg, msp, asize, d,
@@ -3199,6 +3209,13 @@ metaslab_group_alloc_normal(metaslab_group_t *mg, zio_alloc_list_t *zal,
~METASLAB_ACTIVE_MASK);
mutex_exit(&msp->ms_lock);
continue;
+ } else if (msp->ms_initializing > 0) {
+ metaslab_trace_add(zal, mg, msp, asize, d,
+ TRACE_INITIALIZING, allocator);
+ metaslab_passivate(msp, msp->ms_weight &
+ ~METASLAB_ACTIVE_MASK);
+ mutex_exit(&msp->ms_lock);
+ continue;
}
offset = metaslab_block_alloc(msp, asize, txg);
diff --git a/module/zfs/spa.c b/module/zfs/spa.c
index f0683b0b8..622be75f9 100644
--- a/module/zfs/spa.c
+++ b/module/zfs/spa.c
@@ -56,6 +56,7 @@
#include <sys/vdev_removal.h>
#include <sys/vdev_indirect_mapping.h>
#include <sys/vdev_indirect_births.h>
+#include <sys/vdev_initialize.h>
#include <sys/vdev_disk.h>
#include <sys/metaslab.h>
#include <sys/metaslab_impl.h>
@@ -434,8 +435,9 @@ spa_prop_get(spa_t *spa, nvlist_t **nvp)
dp = spa_get_dsl(spa);
dsl_pool_config_enter(dp, FTAG);
- if ((err = dsl_dataset_hold_obj(dp,
- za.za_first_integer, FTAG, &ds))) {
+ err = dsl_dataset_hold_obj(dp,
+ za.za_first_integer, FTAG, &ds);
+ if (err != 0) {
dsl_pool_config_exit(dp, FTAG);
break;
}
@@ -601,7 +603,7 @@ spa_prop_validate(spa_t *spa, nvlist_t *props)
}
error = dmu_objset_hold(strval, FTAG, &os);
- if (error)
+ if (error != 0)
break;
/*
@@ -1218,8 +1220,10 @@ spa_activate(spa_t *spa, int mode)
spa_create_zio_taskqs(spa);
}
- for (size_t i = 0; i < TXG_SIZE; i++)
- spa->spa_txg_zio[i] = zio_root(spa, NULL, NULL, 0);
+ for (size_t i = 0; i < TXG_SIZE; i++) {
+ spa->spa_txg_zio[i] = zio_root(spa, NULL, NULL,
+ ZIO_FLAG_CANFAIL);
+ }
list_create(&spa->spa_config_dirty_list, sizeof (vdev_t),
offsetof(vdev_t, vdev_config_dirty_node));
@@ -1437,6 +1441,11 @@ spa_unload(spa_t *spa)
*/
spa_async_suspend(spa);
+ if (spa->spa_root_vdev) {
+ vdev_initialize_stop_all(spa->spa_root_vdev,
+ VDEV_INITIALIZE_ACTIVE);
+ }
+
/*
* Stop syncing.
*/
@@ -1452,10 +1461,10 @@ spa_unload(spa_t *spa)
* calling taskq_wait(mg_taskq).
*/
if (spa->spa_root_vdev != NULL) {
- spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
+ spa_config_enter(spa, SCL_ALL, spa, RW_WRITER);
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);
+ spa_config_exit(spa, SCL_ALL, spa);
}
if (spa->spa_mmp.mmp_thread)
@@ -1492,7 +1501,7 @@ spa_unload(spa_t *spa)
bpobj_close(&spa->spa_deferred_bpobj);
- spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
+ spa_config_enter(spa, SCL_ALL, spa, RW_WRITER);
/*
* Close all vdevs.
@@ -1554,7 +1563,7 @@ spa_unload(spa_t *spa)
spa->spa_comment = NULL;
}
- spa_config_exit(spa, SCL_ALL, FTAG);
+ spa_config_exit(spa, SCL_ALL, spa);
}
/*
@@ -4246,6 +4255,9 @@ spa_load_impl(spa_t *spa, spa_import_type_t type, char **ereport)
*/
dsl_pool_clean_tmp_userrefs(spa->spa_dsl_pool);
+ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
+ vdev_initialize_restart(spa->spa_root_vdev);
+ spa_config_exit(spa, SCL_CONFIG, FTAG);
}
spa_load_note(spa, "LOADED");
@@ -5654,6 +5666,18 @@ spa_export_common(char *pool, int new_state, nvlist_t **oldconfig,
}
/*
+ * We're about to export or destroy this pool. Make sure
+ * we stop all initializtion activity here before we
+ * set the spa_final_txg. This will ensure that all
+ * dirty data resulting from the initialization is
+ * committed to disk before we unload the pool.
+ */
+ if (spa->spa_root_vdev != NULL) {
+ vdev_initialize_stop_all(spa->spa_root_vdev,
+ VDEV_INITIALIZE_ACTIVE);
+ }
+
+ /*
* We want this to be reflected on every label,
* so mark them all dirty. spa_unload() will do the
* final sync that pushes these changes out.
@@ -6357,6 +6381,86 @@ spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done)
return (error);
}
+int
+spa_vdev_initialize(spa_t *spa, uint64_t guid, uint64_t cmd_type)
+{
+ /*
+ * We hold the namespace lock through the whole function
+ * to prevent any changes to the pool while we're starting or
+ * stopping initialization. The config and state locks are held so that
+ * we can properly assess the vdev state before we commit to
+ * the initializing operation.
+ */
+ mutex_enter(&spa_namespace_lock);
+ spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER);
+
+ /* Look up vdev and ensure it's a leaf. */
+ vdev_t *vd = spa_lookup_by_guid(spa, guid, B_FALSE);
+ if (vd == NULL || vd->vdev_detached) {
+ spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG);
+ mutex_exit(&spa_namespace_lock);
+ return (SET_ERROR(ENODEV));
+ } else if (!vd->vdev_ops->vdev_op_leaf || !vdev_is_concrete(vd)) {
+ spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG);
+ mutex_exit(&spa_namespace_lock);
+ return (SET_ERROR(EINVAL));
+ } else if (!vdev_writeable(vd)) {
+ spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG);
+ mutex_exit(&spa_namespace_lock);
+ return (SET_ERROR(EROFS));
+ }
+ mutex_enter(&vd->vdev_initialize_lock);
+ spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG);
+
+ /*
+ * When we activate an initialize action we check to see
+ * if the vdev_initialize_thread is NULL. We do this instead
+ * of using the vdev_initialize_state since there might be
+ * a previous initialization process which has completed but
+ * the thread is not exited.
+ */
+ if (cmd_type == POOL_INITIALIZE_DO &&
+ (vd->vdev_initialize_thread != NULL ||
+ vd->vdev_top->vdev_removing)) {
+ mutex_exit(&vd->vdev_initialize_lock);
+ mutex_exit(&spa_namespace_lock);
+ return (SET_ERROR(EBUSY));
+ } else if (cmd_type == POOL_INITIALIZE_CANCEL &&
+ (vd->vdev_initialize_state != VDEV_INITIALIZE_ACTIVE &&
+ vd->vdev_initialize_state != VDEV_INITIALIZE_SUSPENDED)) {
+ mutex_exit(&vd->vdev_initialize_lock);
+ mutex_exit(&spa_namespace_lock);
+ return (SET_ERROR(ESRCH));
+ } else if (cmd_type == POOL_INITIALIZE_SUSPEND &&
+ vd->vdev_initialize_state != VDEV_INITIALIZE_ACTIVE) {
+ mutex_exit(&vd->vdev_initialize_lock);
+ mutex_exit(&spa_namespace_lock);
+ return (SET_ERROR(ESRCH));
+ }
+
+ switch (cmd_type) {
+ case POOL_INITIALIZE_DO:
+ vdev_initialize(vd);
+ break;
+ case POOL_INITIALIZE_CANCEL:
+ vdev_initialize_stop(vd, VDEV_INITIALIZE_CANCELED);
+ break;
+ case POOL_INITIALIZE_SUSPEND:
+ vdev_initialize_stop(vd, VDEV_INITIALIZE_SUSPENDED);
+ break;
+ default:
+ panic("invalid cmd_type %llu", (unsigned long long)cmd_type);
+ }
+ mutex_exit(&vd->vdev_initialize_lock);
+
+ /* Sync out the initializing state */
+ txg_wait_synced(spa->spa_dsl_pool, 0);
+ mutex_exit(&spa_namespace_lock);
+
+ return (0);
+}
+
+
/*
* Split a set of devices from their mirrors, and create a new pool from them.
*/
@@ -6565,6 +6669,19 @@ spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config,
spa_activate(newspa, spa_mode_global);
spa_async_suspend(newspa);
+ for (c = 0; c < children; c++) {
+ if (vml[c] != NULL) {
+ /*
+ * Temporarily stop the initializing activity. We set
+ * the state to ACTIVE so that we know to resume
+ * the initializing once the split has completed.
+ */
+ mutex_enter(&vml[c]->vdev_initialize_lock);
+ vdev_initialize_stop(vml[c], VDEV_INITIALIZE_ACTIVE);
+ mutex_exit(&vml[c]->vdev_initialize_lock);
+ }
+ }
+
newspa->spa_config_source = SPA_CONFIG_SRC_SPLIT;
/* create the new pool from the disks of the original pool */
@@ -6652,6 +6769,10 @@ out:
if (vml[c] != NULL)
vml[c]->vdev_offline = B_FALSE;
}
+
+ /* restart initializing disks as necessary */
+ spa_async_request(spa, SPA_ASYNC_INITIALIZE_RESTART);
+
vdev_reopen(spa->spa_root_vdev);
nvlist_free(spa->spa_config_splitting);
@@ -7025,6 +7146,14 @@ spa_async_thread(void *arg)
!spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_RESILVER_DEFER)))
dsl_resilver_restart(dp, 0);
+ if (tasks & SPA_ASYNC_INITIALIZE_RESTART) {
+ mutex_enter(&spa_namespace_lock);
+ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
+ vdev_initialize_restart(spa->spa_root_vdev);
+ spa_config_exit(spa, SCL_CONFIG, FTAG);
+ mutex_exit(&spa_namespace_lock);
+ }
+
/*
* Let the world know that we're done.
*/
@@ -7677,8 +7806,9 @@ spa_sync(spa_t *spa, uint64_t txg)
* Wait for i/os issued in open context that need to complete
* before this txg syncs.
*/
- VERIFY0(zio_wait(spa->spa_txg_zio[txg & TXG_MASK]));
- spa->spa_txg_zio[txg & TXG_MASK] = zio_root(spa, NULL, NULL, 0);
+ (void) zio_wait(spa->spa_txg_zio[txg & TXG_MASK]);
+ spa->spa_txg_zio[txg & TXG_MASK] = zio_root(spa, NULL, NULL,
+ ZIO_FLAG_CANFAIL);
/*
* Lock out configuration changes.
@@ -7983,7 +8113,8 @@ spa_sync(spa_t *spa, uint64_t txg)
/*
* Update usable space statistics.
*/
- while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, TXG_CLEAN(txg))))
+ while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, TXG_CLEAN(txg)))
+ != NULL)
vdev_sync_done(vd, txg);
spa_update_dspace(spa);
diff --git a/module/zfs/spa_misc.c b/module/zfs/spa_misc.c
index a3ac70f07..dfac92d45 100644
--- a/module/zfs/spa_misc.c
+++ b/module/zfs/spa_misc.c
@@ -38,6 +38,7 @@
#include <sys/zap.h>
#include <sys/zil.h>
#include <sys/vdev_impl.h>
+#include <sys/vdev_initialize.h>
#include <sys/vdev_file.h>
#include <sys/vdev_raidz.h>
#include <sys/metaslab.h>
@@ -1194,6 +1195,12 @@ spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error, char *tag)
if (vd != NULL) {
ASSERT(!vd->vdev_detached || vd->vdev_dtl_sm == NULL);
+ if (vd->vdev_ops->vdev_op_leaf) {
+ mutex_enter(&vd->vdev_initialize_lock);
+ vdev_initialize_stop(vd, VDEV_INITIALIZE_CANCELED);
+ mutex_exit(&vd->vdev_initialize_lock);
+ }
+
spa_config_enter(spa, SCL_ALL, spa, RW_WRITER);
vdev_free(vd);
spa_config_exit(spa, SCL_ALL, spa);
diff --git a/module/zfs/vdev.c b/module/zfs/vdev.c
index 8273e7907..f808f8ee7 100644
--- a/module/zfs/vdev.c
+++ b/module/zfs/vdev.c
@@ -50,6 +50,7 @@
#include <sys/zil.h>
#include <sys/dsl_scan.h>
#include <sys/abd.h>
+#include <sys/vdev_initialize.h>
#include <sys/zvol.h>
#include <sys/zfs_ratelimit.h>
@@ -212,6 +213,14 @@ vdev_getops(const char *type)
return (ops);
}
+/* ARGSUSED */
+void
+vdev_default_xlate(vdev_t *vd, const range_seg_t *in, range_seg_t *res)
+{
+ res->rs_start = in->rs_start;
+ res->rs_end = in->rs_end;
+}
+
/*
* Derive the enumerated alloction bias from string input.
* String origin is either the per-vdev zap or zpool(1M).
@@ -526,6 +535,10 @@ vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid, vdev_ops_t *ops)
mutex_init(&vd->vdev_probe_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&vd->vdev_queue_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&vd->vdev_scan_io_queue_lock, NULL, MUTEX_DEFAULT, NULL);
+ mutex_init(&vd->vdev_initialize_lock, NULL, MUTEX_DEFAULT, NULL);
+ mutex_init(&vd->vdev_initialize_io_lock, NULL, MUTEX_DEFAULT, NULL);
+ cv_init(&vd->vdev_initialize_cv, NULL, CV_DEFAULT, NULL);
+ cv_init(&vd->vdev_initialize_io_cv, NULL, CV_DEFAULT, NULL);
for (int t = 0; t < DTL_TYPES; t++) {
vd->vdev_dtl[t] = range_tree_create(NULL, NULL);
@@ -850,6 +863,7 @@ void
vdev_free(vdev_t *vd)
{
spa_t *spa = vd->vdev_spa;
+ ASSERT3P(vd->vdev_initialize_thread, ==, NULL);
/*
* Scan queues are normally destroyed at the end of a scan. If the
@@ -880,6 +894,7 @@ vdev_free(vdev_t *vd)
ASSERT(vd->vdev_child == NULL);
ASSERT(vd->vdev_guid_sum == vd->vdev_guid);
+ ASSERT(vd->vdev_initialize_thread == NULL);
/*
* Discard allocation state.
@@ -957,6 +972,10 @@ vdev_free(vdev_t *vd)
mutex_destroy(&vd->vdev_stat_lock);
mutex_destroy(&vd->vdev_probe_lock);
mutex_destroy(&vd->vdev_scan_io_queue_lock);
+ mutex_destroy(&vd->vdev_initialize_lock);
+ mutex_destroy(&vd->vdev_initialize_io_lock);
+ cv_destroy(&vd->vdev_initialize_io_cv);
+ cv_destroy(&vd->vdev_initialize_cv);
zfs_ratelimit_fini(&vd->vdev_delay_rl);
zfs_ratelimit_fini(&vd->vdev_checksum_rl);
@@ -3207,7 +3226,8 @@ vdev_sync_done(vdev_t *vd, uint64_t txg)
ASSERT(vdev_is_concrete(vd));
- while ((msp = txg_list_remove(&vd->vdev_ms_list, TXG_CLEAN(txg))))
+ while ((msp = txg_list_remove(&vd->vdev_ms_list, TXG_CLEAN(txg)))
+ != NULL)
metaslab_sync_done(msp, txg);
if (reassess)
@@ -3458,6 +3478,15 @@ vdev_online(spa_t *spa, uint64_t guid, uint64_t flags, vdev_state_t *newstate)
spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE);
}
+ /* Restart initializing if necessary */
+ mutex_enter(&vd->vdev_initialize_lock);
+ if (vdev_writeable(vd) &&
+ vd->vdev_initialize_thread == NULL &&
+ vd->vdev_initialize_state == VDEV_INITIALIZE_ACTIVE) {
+ (void) vdev_initialize(vd);
+ }
+ mutex_exit(&vd->vdev_initialize_lock);
+
if (wasoffline ||
(oldstate < VDEV_STATE_DEGRADED &&
vd->vdev_state >= VDEV_STATE_DEGRADED))
@@ -3848,9 +3877,22 @@ vdev_get_stats_ex(vdev_t *vd, vdev_stat_t *vs, vdev_stat_ex_t *vsx)
vs->vs_timestamp = gethrtime() - vs->vs_timestamp;
vs->vs_state = vd->vdev_state;
vs->vs_rsize = vdev_get_min_asize(vd);
- if (vd->vdev_ops->vdev_op_leaf)
+ if (vd->vdev_ops->vdev_op_leaf) {
vs->vs_rsize += VDEV_LABEL_START_SIZE +
VDEV_LABEL_END_SIZE;
+ /*
+ * Report intializing progress. Since we don't
+ * have the initializing locks held, this is only
+ * an estimate (although a fairly accurate one).
+ */
+ vs->vs_initialize_bytes_done =
+ vd->vdev_initialize_bytes_done;
+ vs->vs_initialize_bytes_est =
+ vd->vdev_initialize_bytes_est;
+ vs->vs_initialize_state = vd->vdev_initialize_state;
+ vs->vs_initialize_action_time =
+ vd->vdev_initialize_action_time;
+ }
/*
* Report expandable space on top-level, non-auxillary devices
* only. The expandable space is reported in terms of metaslab
diff --git a/module/zfs/vdev_disk.c b/module/zfs/vdev_disk.c
index 9c44ba12a..d13f365dd 100644
--- a/module/zfs/vdev_disk.c
+++ b/module/zfs/vdev_disk.c
@@ -890,6 +890,7 @@ vdev_ops_t vdev_disk_ops = {
vdev_disk_hold,
vdev_disk_rele,
NULL,
+ vdev_default_xlate,
VDEV_TYPE_DISK, /* name of this vdev type */
B_TRUE /* leaf vdev */
};
diff --git a/module/zfs/vdev_file.c b/module/zfs/vdev_file.c
index bd7e0bc2e..3551898e0 100644
--- a/module/zfs/vdev_file.c
+++ b/module/zfs/vdev_file.c
@@ -20,7 +20,7 @@
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2015 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2016 by Delphix. All rights reserved.
*/
#include <sys/zfs_context.h>
@@ -254,6 +254,7 @@ vdev_ops_t vdev_file_ops = {
vdev_file_hold,
vdev_file_rele,
NULL,
+ vdev_default_xlate,
VDEV_TYPE_FILE, /* name of this vdev type */
B_TRUE /* leaf vdev */
};
@@ -289,6 +290,7 @@ vdev_ops_t vdev_disk_ops = {
vdev_file_hold,
vdev_file_rele,
NULL,
+ vdev_default_xlate,
VDEV_TYPE_DISK, /* name of this vdev type */
B_TRUE /* leaf vdev */
};
diff --git a/module/zfs/vdev_indirect.c b/module/zfs/vdev_indirect.c
index 724457df4..070d1b8d9 100644
--- a/module/zfs/vdev_indirect.c
+++ b/module/zfs/vdev_indirect.c
@@ -1857,6 +1857,7 @@ vdev_ops_t vdev_indirect_ops = {
NULL,
NULL,
vdev_indirect_remap,
+ NULL,
VDEV_TYPE_INDIRECT, /* name of this vdev type */
B_FALSE /* leaf vdev */
};
diff --git a/module/zfs/vdev_initialize.c b/module/zfs/vdev_initialize.c
new file mode 100644
index 000000000..fcd2c76f9
--- /dev/null
+++ b/module/zfs/vdev_initialize.c
@@ -0,0 +1,819 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License (the "License").
+ * You may not use this file except in compliance with the License.
+ *
+ * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
+ * or http://www.opensolaris.org/os/licensing.
+ * See the License for the specific language governing permissions
+ * and limitations under the License.
+ *
+ * When distributing Covered Code, include this CDDL HEADER in each
+ * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
+ * If applicable, add the following below this CDDL HEADER, with the
+ * fields enclosed by brackets "[]" replaced with your own identifying
+ * information: Portions Copyright [yyyy] [name of copyright owner]
+ *
+ * CDDL HEADER END
+ */
+
+/*
+ * Copyright (c) 2016 by Delphix. All rights reserved.
+ */
+
+#include <sys/spa.h>
+#include <sys/spa_impl.h>
+#include <sys/txg.h>
+#include <sys/vdev_impl.h>
+#include <sys/refcount.h>
+#include <sys/metaslab_impl.h>
+#include <sys/dsl_synctask.h>
+#include <sys/zap.h>
+#include <sys/dmu_tx.h>
+
+/*
+ * Maximum number of metaslabs per group that can be initialized
+ * simultaneously.
+ */
+int max_initialize_ms = 3;
+
+/*
+ * Value that is written to disk during initialization.
+ */
+#ifdef _ILP32
+unsigned long zfs_initialize_value = 0xdeadbeefUL;
+#else
+unsigned long zfs_initialize_value = 0xdeadbeefdeadbeeeULL;
+#endif
+
+/* maximum number of I/Os outstanding per leaf vdev */
+int zfs_initialize_limit = 1;
+
+/* size of initializing writes; default 1MiB, see zfs_remove_max_segment */
+uint64_t zfs_initialize_chunk_size = 1024 * 1024;
+
+static boolean_t
+vdev_initialize_should_stop(vdev_t *vd)
+{
+ return (vd->vdev_initialize_exit_wanted || !vdev_writeable(vd) ||
+ vd->vdev_detached || vd->vdev_top->vdev_removing);
+}
+
+static void
+vdev_initialize_zap_update_sync(void *arg, dmu_tx_t *tx)
+{
+ /*
+ * We pass in the guid instead of the vdev_t since the vdev may
+ * have been freed prior to the sync task being processed. This
+ * happens when a vdev is detached as we call spa_config_vdev_exit(),
+ * stop the intializing thread, schedule the sync task, and free
+ * the vdev. Later when the scheduled sync task is invoked, it would
+ * find that the vdev has been freed.
+ */
+ uint64_t guid = *(uint64_t *)arg;
+ uint64_t txg = dmu_tx_get_txg(tx);
+ kmem_free(arg, sizeof (uint64_t));
+
+ vdev_t *vd = spa_lookup_by_guid(tx->tx_pool->dp_spa, guid, B_FALSE);
+ if (vd == NULL || vd->vdev_top->vdev_removing || !vdev_is_concrete(vd))
+ return;
+
+ uint64_t last_offset = vd->vdev_initialize_offset[txg & TXG_MASK];
+ vd->vdev_initialize_offset[txg & TXG_MASK] = 0;
+
+ VERIFY(vd->vdev_leaf_zap != 0);
+
+ objset_t *mos = vd->vdev_spa->spa_meta_objset;
+
+ if (last_offset > 0) {
+ vd->vdev_initialize_last_offset = last_offset;
+ VERIFY0(zap_update(mos, vd->vdev_leaf_zap,
+ VDEV_LEAF_ZAP_INITIALIZE_LAST_OFFSET,
+ sizeof (last_offset), 1, &last_offset, tx));
+ }
+ if (vd->vdev_initialize_action_time > 0) {
+ uint64_t val = (uint64_t)vd->vdev_initialize_action_time;
+ VERIFY0(zap_update(mos, vd->vdev_leaf_zap,
+ VDEV_LEAF_ZAP_INITIALIZE_ACTION_TIME, sizeof (val),
+ 1, &val, tx));
+ }
+
+ uint64_t initialize_state = vd->vdev_initialize_state;
+ VERIFY0(zap_update(mos, vd->vdev_leaf_zap,
+ VDEV_LEAF_ZAP_INITIALIZE_STATE, sizeof (initialize_state), 1,
+ &initialize_state, tx));
+}
+
+static void
+vdev_initialize_change_state(vdev_t *vd, vdev_initializing_state_t new_state)
+{
+ ASSERT(MUTEX_HELD(&vd->vdev_initialize_lock));
+ spa_t *spa = vd->vdev_spa;
+
+ if (new_state == vd->vdev_initialize_state)
+ return;
+
+ /*
+ * Copy the vd's guid, this will be freed by the sync task.
+ */
+ uint64_t *guid = kmem_zalloc(sizeof (uint64_t), KM_SLEEP);
+ *guid = vd->vdev_guid;
+
+ /*
+ * If we're suspending, then preserving the original start time.
+ */
+ if (vd->vdev_initialize_state != VDEV_INITIALIZE_SUSPENDED) {
+ vd->vdev_initialize_action_time = gethrestime_sec();
+ }
+ vd->vdev_initialize_state = new_state;
+
+ dmu_tx_t *tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir);
+ VERIFY0(dmu_tx_assign(tx, TXG_WAIT));
+ dsl_sync_task_nowait(spa_get_dsl(spa), vdev_initialize_zap_update_sync,
+ guid, 2, ZFS_SPACE_CHECK_RESERVED, tx);
+
+ switch (new_state) {
+ case VDEV_INITIALIZE_ACTIVE:
+ spa_history_log_internal(spa, "initialize", tx,
+ "vdev=%s activated", vd->vdev_path);
+ break;
+ case VDEV_INITIALIZE_SUSPENDED:
+ spa_history_log_internal(spa, "initialize", tx,
+ "vdev=%s suspended", vd->vdev_path);
+ break;
+ case VDEV_INITIALIZE_CANCELED:
+ spa_history_log_internal(spa, "initialize", tx,
+ "vdev=%s canceled", vd->vdev_path);
+ break;
+ case VDEV_INITIALIZE_COMPLETE:
+ spa_history_log_internal(spa, "initialize", tx,
+ "vdev=%s complete", vd->vdev_path);
+ break;
+ default:
+ panic("invalid state %llu", (unsigned long long)new_state);
+ }
+
+ dmu_tx_commit(tx);
+}
+
+static void
+vdev_initialize_cb(zio_t *zio)
+{
+ vdev_t *vd = zio->io_vd;
+ mutex_enter(&vd->vdev_initialize_io_lock);
+ if (zio->io_error == ENXIO && !vdev_writeable(vd)) {
+ /*
+ * The I/O failed because the vdev was unavailable; roll the
+ * last offset back. (This works because spa_sync waits on
+ * spa_txg_zio before it runs sync tasks.)
+ */
+ uint64_t *off =
+ &vd->vdev_initialize_offset[zio->io_txg & TXG_MASK];
+ *off = MIN(*off, zio->io_offset);
+ } else {
+ /*
+ * Since initializing is best-effort, we ignore I/O errors and
+ * rely on vdev_probe to determine if the errors are more
+ * critical.
+ */
+ if (zio->io_error != 0)
+ vd->vdev_stat.vs_initialize_errors++;
+
+ vd->vdev_initialize_bytes_done += zio->io_orig_size;
+ }
+ ASSERT3U(vd->vdev_initialize_inflight, >, 0);
+ vd->vdev_initialize_inflight--;
+ cv_broadcast(&vd->vdev_initialize_io_cv);
+ mutex_exit(&vd->vdev_initialize_io_lock);
+
+ spa_config_exit(vd->vdev_spa, SCL_STATE_ALL, vd);
+}
+
+/* Takes care of physical writing and limiting # of concurrent ZIOs. */
+static int
+vdev_initialize_write(vdev_t *vd, uint64_t start, uint64_t size, abd_t *data)
+{
+ spa_t *spa = vd->vdev_spa;
+
+ /* Limit inflight initializing I/Os */
+ mutex_enter(&vd->vdev_initialize_io_lock);
+ while (vd->vdev_initialize_inflight >= zfs_initialize_limit) {
+ cv_wait(&vd->vdev_initialize_io_cv,
+ &vd->vdev_initialize_io_lock);
+ }
+ vd->vdev_initialize_inflight++;
+ mutex_exit(&vd->vdev_initialize_io_lock);
+
+ dmu_tx_t *tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir);
+ VERIFY0(dmu_tx_assign(tx, TXG_WAIT));
+ uint64_t txg = dmu_tx_get_txg(tx);
+
+ spa_config_enter(spa, SCL_STATE_ALL, vd, RW_READER);
+ mutex_enter(&vd->vdev_initialize_lock);
+
+ if (vd->vdev_initialize_offset[txg & TXG_MASK] == 0) {
+ uint64_t *guid = kmem_zalloc(sizeof (uint64_t), KM_SLEEP);
+ *guid = vd->vdev_guid;
+
+ /* This is the first write of this txg. */
+ dsl_sync_task_nowait(spa_get_dsl(spa),
+ vdev_initialize_zap_update_sync, guid, 2,
+ ZFS_SPACE_CHECK_RESERVED, tx);
+ }
+
+ /*
+ * We know the vdev struct will still be around since all
+ * consumers of vdev_free must stop the initialization first.
+ */
+ if (vdev_initialize_should_stop(vd)) {
+ mutex_enter(&vd->vdev_initialize_io_lock);
+ ASSERT3U(vd->vdev_initialize_inflight, >, 0);
+ vd->vdev_initialize_inflight--;
+ mutex_exit(&vd->vdev_initialize_io_lock);
+ spa_config_exit(vd->vdev_spa, SCL_STATE_ALL, vd);
+ mutex_exit(&vd->vdev_initialize_lock);
+ dmu_tx_commit(tx);
+ return (SET_ERROR(EINTR));
+ }
+ mutex_exit(&vd->vdev_initialize_lock);
+
+ vd->vdev_initialize_offset[txg & TXG_MASK] = start + size;
+ zio_nowait(zio_write_phys(spa->spa_txg_zio[txg & TXG_MASK], vd, start,
+ size, data, ZIO_CHECKSUM_OFF, vdev_initialize_cb, NULL,
+ ZIO_PRIORITY_INITIALIZING, ZIO_FLAG_CANFAIL, B_FALSE));
+ /* vdev_initialize_cb releases SCL_STATE_ALL */
+
+ dmu_tx_commit(tx);
+
+ return (0);
+}
+
+/*
+ * Translate a logical range to the physical range for the specified vdev_t.
+ * This function is initially called with a leaf vdev and will walk each
+ * parent vdev until it reaches a top-level vdev. Once the top-level is
+ * reached the physical range is initialized and the recursive function
+ * begins to unwind. As it unwinds it calls the parent's vdev specific
+ * translation function to do the real conversion.
+ */
+void
+vdev_xlate(vdev_t *vd, const range_seg_t *logical_rs, range_seg_t *physical_rs)
+{
+ /*
+ * Walk up the vdev tree
+ */
+ if (vd != vd->vdev_top) {
+ vdev_xlate(vd->vdev_parent, logical_rs, physical_rs);
+ } else {
+ /*
+ * We've reached the top-level vdev, initialize the
+ * physical range to the logical range and start to
+ * unwind.
+ */
+ physical_rs->rs_start = logical_rs->rs_start;
+ physical_rs->rs_end = logical_rs->rs_end;
+ return;
+ }
+
+ vdev_t *pvd = vd->vdev_parent;
+ ASSERT3P(pvd, !=, NULL);
+ ASSERT3P(pvd->vdev_ops->vdev_op_xlate, !=, NULL);
+
+ /*
+ * As this recursive function unwinds, translate the logical
+ * range into its physical components by calling the
+ * vdev specific translate function.
+ */
+ range_seg_t intermediate = { { { 0, 0 } } };
+ pvd->vdev_ops->vdev_op_xlate(vd, physical_rs, &intermediate);
+
+ physical_rs->rs_start = intermediate.rs_start;
+ physical_rs->rs_end = intermediate.rs_end;
+}
+
+/*
+ * Callback to fill each ABD chunk with zfs_initialize_value. len must be
+ * divisible by sizeof (uint64_t), and buf must be 8-byte aligned. The ABD
+ * allocation will guarantee these for us.
+ */
+/* ARGSUSED */
+static int
+vdev_initialize_block_fill(void *buf, size_t len, void *unused)
+{
+ ASSERT0(len % sizeof (uint64_t));
+#ifdef _ILP32
+ for (uint64_t i = 0; i < len; i += sizeof (uint32_t)) {
+ *(uint32_t *)((char *)(buf) + i) = zfs_initialize_value;
+ }
+#else
+ for (uint64_t i = 0; i < len; i += sizeof (uint64_t)) {
+ *(uint64_t *)((char *)(buf) + i) = zfs_initialize_value;
+ }
+#endif
+ return (0);
+}
+
+static abd_t *
+vdev_initialize_block_alloc(void)
+{
+ /* Allocate ABD for filler data */
+ abd_t *data = abd_alloc_for_io(zfs_initialize_chunk_size, B_FALSE);
+
+ ASSERT0(zfs_initialize_chunk_size % sizeof (uint64_t));
+ (void) abd_iterate_func(data, 0, zfs_initialize_chunk_size,
+ vdev_initialize_block_fill, NULL);
+
+ return (data);
+}
+
+static void
+vdev_initialize_block_free(abd_t *data)
+{
+ abd_free(data);
+}
+
+static int
+vdev_initialize_ranges(vdev_t *vd, abd_t *data)
+{
+ avl_tree_t *rt = &vd->vdev_initialize_tree->rt_root;
+
+ for (range_seg_t *rs = avl_first(rt); rs != NULL;
+ rs = AVL_NEXT(rt, rs)) {
+ uint64_t size = rs->rs_end - rs->rs_start;
+
+ /* Split range into legally-sized physical chunks */
+ uint64_t writes_required =
+ ((size - 1) / zfs_initialize_chunk_size) + 1;
+
+ for (uint64_t w = 0; w < writes_required; w++) {
+ int error;
+
+ error = vdev_initialize_write(vd,
+ VDEV_LABEL_START_SIZE + rs->rs_start +
+ (w * zfs_initialize_chunk_size),
+ MIN(size - (w * zfs_initialize_chunk_size),
+ zfs_initialize_chunk_size), data);
+ if (error != 0)
+ return (error);
+ }
+ }
+ return (0);
+}
+
+static void
+vdev_initialize_ms_load(metaslab_t *msp)
+{
+ ASSERT(MUTEX_HELD(&msp->ms_lock));
+
+ metaslab_load_wait(msp);
+ if (!msp->ms_loaded)
+ VERIFY0(metaslab_load(msp));
+}
+
+static void
+vdev_initialize_mg_wait(metaslab_group_t *mg)
+{
+ ASSERT(MUTEX_HELD(&mg->mg_ms_initialize_lock));
+ while (mg->mg_initialize_updating) {
+ cv_wait(&mg->mg_ms_initialize_cv, &mg->mg_ms_initialize_lock);
+ }
+}
+
+static void
+vdev_initialize_mg_mark(metaslab_group_t *mg)
+{
+ ASSERT(MUTEX_HELD(&mg->mg_ms_initialize_lock));
+ ASSERT(mg->mg_initialize_updating);
+
+ while (mg->mg_ms_initializing >= max_initialize_ms) {
+ cv_wait(&mg->mg_ms_initialize_cv, &mg->mg_ms_initialize_lock);
+ }
+ mg->mg_ms_initializing++;
+ ASSERT3U(mg->mg_ms_initializing, <=, max_initialize_ms);
+}
+
+/*
+ * Mark the metaslab as being initialized to prevent any allocations
+ * on this metaslab. We must also track how many metaslabs are currently
+ * being initialized within a metaslab group and limit them to prevent
+ * allocation failures from occurring because all metaslabs are being
+ * initialized.
+ */
+static void
+vdev_initialize_ms_mark(metaslab_t *msp)
+{
+ ASSERT(!MUTEX_HELD(&msp->ms_lock));
+ metaslab_group_t *mg = msp->ms_group;
+
+ mutex_enter(&mg->mg_ms_initialize_lock);
+
+ /*
+ * To keep an accurate count of how many threads are initializing
+ * a specific metaslab group, we only allow one thread to mark
+ * the metaslab group at a time. This ensures that the value of
+ * ms_initializing will be accurate when we decide to mark a metaslab
+ * group as being initialized. To do this we force all other threads
+ * to wait till the metaslab's mg_initialize_updating flag is no
+ * longer set.
+ */
+ vdev_initialize_mg_wait(mg);
+ mg->mg_initialize_updating = B_TRUE;
+ if (msp->ms_initializing == 0) {
+ vdev_initialize_mg_mark(mg);
+ }
+ mutex_enter(&msp->ms_lock);
+ msp->ms_initializing++;
+ mutex_exit(&msp->ms_lock);
+
+ mg->mg_initialize_updating = B_FALSE;
+ cv_broadcast(&mg->mg_ms_initialize_cv);
+ mutex_exit(&mg->mg_ms_initialize_lock);
+}
+
+static void
+vdev_initialize_ms_unmark(metaslab_t *msp)
+{
+ ASSERT(!MUTEX_HELD(&msp->ms_lock));
+ metaslab_group_t *mg = msp->ms_group;
+ mutex_enter(&mg->mg_ms_initialize_lock);
+ mutex_enter(&msp->ms_lock);
+ if (--msp->ms_initializing == 0) {
+ mg->mg_ms_initializing--;
+ cv_broadcast(&mg->mg_ms_initialize_cv);
+ }
+ mutex_exit(&msp->ms_lock);
+ mutex_exit(&mg->mg_ms_initialize_lock);
+}
+
+static void
+vdev_initialize_calculate_progress(vdev_t *vd)
+{
+ ASSERT(spa_config_held(vd->vdev_spa, SCL_CONFIG, RW_READER) ||
+ spa_config_held(vd->vdev_spa, SCL_CONFIG, RW_WRITER));
+ ASSERT(vd->vdev_leaf_zap != 0);
+
+ vd->vdev_initialize_bytes_est = 0;
+ vd->vdev_initialize_bytes_done = 0;
+
+ for (uint64_t i = 0; i < vd->vdev_top->vdev_ms_count; i++) {
+ metaslab_t *msp = vd->vdev_top->vdev_ms[i];
+ mutex_enter(&msp->ms_lock);
+
+ uint64_t ms_free = msp->ms_size -
+ space_map_allocated(msp->ms_sm);
+
+ if (vd->vdev_top->vdev_ops == &vdev_raidz_ops)
+ ms_free /= vd->vdev_top->vdev_children;
+
+ /*
+ * Convert the metaslab range to a physical range
+ * on our vdev. We use this to determine if we are
+ * in the middle of this metaslab range.
+ */
+ range_seg_t logical_rs, physical_rs;
+ logical_rs.rs_start = msp->ms_start;
+ logical_rs.rs_end = msp->ms_start + msp->ms_size;
+ vdev_xlate(vd, &logical_rs, &physical_rs);
+
+ if (vd->vdev_initialize_last_offset <= physical_rs.rs_start) {
+ vd->vdev_initialize_bytes_est += ms_free;
+ mutex_exit(&msp->ms_lock);
+ continue;
+ } else if (vd->vdev_initialize_last_offset >
+ physical_rs.rs_end) {
+ vd->vdev_initialize_bytes_done += ms_free;
+ vd->vdev_initialize_bytes_est += ms_free;
+ mutex_exit(&msp->ms_lock);
+ continue;
+ }
+
+ /*
+ * If we get here, we're in the middle of initializing this
+ * metaslab. Load it and walk the free tree for more accurate
+ * progress estimation.
+ */
+ vdev_initialize_ms_load(msp);
+
+ for (range_seg_t *rs = avl_first(&msp->ms_allocatable->rt_root);
+ rs; rs = AVL_NEXT(&msp->ms_allocatable->rt_root, rs)) {
+ logical_rs.rs_start = rs->rs_start;
+ logical_rs.rs_end = rs->rs_end;
+ vdev_xlate(vd, &logical_rs, &physical_rs);
+
+ uint64_t size = physical_rs.rs_end -
+ physical_rs.rs_start;
+ vd->vdev_initialize_bytes_est += size;
+ if (vd->vdev_initialize_last_offset >
+ physical_rs.rs_end) {
+ vd->vdev_initialize_bytes_done += size;
+ } else if (vd->vdev_initialize_last_offset >
+ physical_rs.rs_start &&
+ vd->vdev_initialize_last_offset <
+ physical_rs.rs_end) {
+ vd->vdev_initialize_bytes_done +=
+ vd->vdev_initialize_last_offset -
+ physical_rs.rs_start;
+ }
+ }
+ mutex_exit(&msp->ms_lock);
+ }
+}
+
+static int
+vdev_initialize_load(vdev_t *vd)
+{
+ int err = 0;
+ ASSERT(spa_config_held(vd->vdev_spa, SCL_CONFIG, RW_READER) ||
+ spa_config_held(vd->vdev_spa, SCL_CONFIG, RW_WRITER));
+ ASSERT(vd->vdev_leaf_zap != 0);
+
+ if (vd->vdev_initialize_state == VDEV_INITIALIZE_ACTIVE ||
+ vd->vdev_initialize_state == VDEV_INITIALIZE_SUSPENDED) {
+ err = zap_lookup(vd->vdev_spa->spa_meta_objset,
+ vd->vdev_leaf_zap, VDEV_LEAF_ZAP_INITIALIZE_LAST_OFFSET,
+ sizeof (vd->vdev_initialize_last_offset), 1,
+ &vd->vdev_initialize_last_offset);
+ if (err == ENOENT) {
+ vd->vdev_initialize_last_offset = 0;
+ err = 0;
+ }
+ }
+
+ vdev_initialize_calculate_progress(vd);
+ return (err);
+}
+
+
+/*
+ * Convert the logical range into a physcial range and add it to our
+ * avl tree.
+ */
+void
+vdev_initialize_range_add(void *arg, uint64_t start, uint64_t size)
+{
+ vdev_t *vd = arg;
+ range_seg_t logical_rs, physical_rs;
+ logical_rs.rs_start = start;
+ logical_rs.rs_end = start + size;
+
+ ASSERT(vd->vdev_ops->vdev_op_leaf);
+ vdev_xlate(vd, &logical_rs, &physical_rs);
+
+ IMPLY(vd->vdev_top == vd,
+ logical_rs.rs_start == physical_rs.rs_start);
+ IMPLY(vd->vdev_top == vd,
+ logical_rs.rs_end == physical_rs.rs_end);
+
+ /* Only add segments that we have not visited yet */
+ if (physical_rs.rs_end <= vd->vdev_initialize_last_offset)
+ return;
+
+ /* Pick up where we left off mid-range. */
+ if (vd->vdev_initialize_last_offset > physical_rs.rs_start) {
+ zfs_dbgmsg("range write: vd %s changed (%llu, %llu) to "
+ "(%llu, %llu)", vd->vdev_path,
+ (u_longlong_t)physical_rs.rs_start,
+ (u_longlong_t)physical_rs.rs_end,
+ (u_longlong_t)vd->vdev_initialize_last_offset,
+ (u_longlong_t)physical_rs.rs_end);
+ ASSERT3U(physical_rs.rs_end, >,
+ vd->vdev_initialize_last_offset);
+ physical_rs.rs_start = vd->vdev_initialize_last_offset;
+ }
+ ASSERT3U(physical_rs.rs_end, >=, physical_rs.rs_start);
+
+ /*
+ * With raidz, it's possible that the logical range does not live on
+ * this leaf vdev. We only add the physical range to this vdev's if it
+ * has a length greater than 0.
+ */
+ if (physical_rs.rs_end > physical_rs.rs_start) {
+ range_tree_add(vd->vdev_initialize_tree, physical_rs.rs_start,
+ physical_rs.rs_end - physical_rs.rs_start);
+ } else {
+ ASSERT3U(physical_rs.rs_end, ==, physical_rs.rs_start);
+ }
+}
+
+static void
+vdev_initialize_thread(void *arg)
+{
+ vdev_t *vd = arg;
+ spa_t *spa = vd->vdev_spa;
+ int error = 0;
+ uint64_t ms_count = 0;
+
+ ASSERT(vdev_is_concrete(vd));
+ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
+
+ vd->vdev_initialize_last_offset = 0;
+ VERIFY0(vdev_initialize_load(vd));
+
+ abd_t *deadbeef = vdev_initialize_block_alloc();
+
+ vd->vdev_initialize_tree = range_tree_create(NULL, NULL);
+
+ for (uint64_t i = 0; !vd->vdev_detached &&
+ i < vd->vdev_top->vdev_ms_count; i++) {
+ metaslab_t *msp = vd->vdev_top->vdev_ms[i];
+
+ /*
+ * If we've expanded the top-level vdev or it's our
+ * first pass, calculate our progress.
+ */
+ if (vd->vdev_top->vdev_ms_count != ms_count) {
+ vdev_initialize_calculate_progress(vd);
+ ms_count = vd->vdev_top->vdev_ms_count;
+ }
+
+ vdev_initialize_ms_mark(msp);
+ mutex_enter(&msp->ms_lock);
+ vdev_initialize_ms_load(msp);
+
+ range_tree_walk(msp->ms_allocatable, vdev_initialize_range_add,
+ vd);
+ mutex_exit(&msp->ms_lock);
+
+ spa_config_exit(spa, SCL_CONFIG, FTAG);
+ error = vdev_initialize_ranges(vd, deadbeef);
+ vdev_initialize_ms_unmark(msp);
+ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
+
+ range_tree_vacate(vd->vdev_initialize_tree, NULL, NULL);
+ if (error != 0)
+ break;
+ }
+
+ spa_config_exit(spa, SCL_CONFIG, FTAG);
+ mutex_enter(&vd->vdev_initialize_io_lock);
+ while (vd->vdev_initialize_inflight > 0) {
+ cv_wait(&vd->vdev_initialize_io_cv,
+ &vd->vdev_initialize_io_lock);
+ }
+ mutex_exit(&vd->vdev_initialize_io_lock);
+
+ range_tree_destroy(vd->vdev_initialize_tree);
+ vdev_initialize_block_free(deadbeef);
+ vd->vdev_initialize_tree = NULL;
+
+ mutex_enter(&vd->vdev_initialize_lock);
+ if (!vd->vdev_initialize_exit_wanted && vdev_writeable(vd)) {
+ vdev_initialize_change_state(vd, VDEV_INITIALIZE_COMPLETE);
+ }
+ ASSERT(vd->vdev_initialize_thread != NULL ||
+ vd->vdev_initialize_inflight == 0);
+
+ /*
+ * Drop the vdev_initialize_lock while we sync out the
+ * txg since it's possible that a device might be trying to
+ * come online and must check to see if it needs to restart an
+ * initialization. That thread will be holding the spa_config_lock
+ * which would prevent the txg_wait_synced from completing.
+ */
+ mutex_exit(&vd->vdev_initialize_lock);
+ txg_wait_synced(spa_get_dsl(spa), 0);
+ mutex_enter(&vd->vdev_initialize_lock);
+
+ vd->vdev_initialize_thread = NULL;
+ cv_broadcast(&vd->vdev_initialize_cv);
+ mutex_exit(&vd->vdev_initialize_lock);
+}
+
+/*
+ * Initiates a device. Caller must hold vdev_initialize_lock.
+ * Device must be a leaf and not already be initializing.
+ */
+void
+vdev_initialize(vdev_t *vd)
+{
+ ASSERT(MUTEX_HELD(&vd->vdev_initialize_lock));
+ ASSERT(vd->vdev_ops->vdev_op_leaf);
+ ASSERT(vdev_is_concrete(vd));
+ ASSERT3P(vd->vdev_initialize_thread, ==, NULL);
+ ASSERT(!vd->vdev_detached);
+ ASSERT(!vd->vdev_initialize_exit_wanted);
+ ASSERT(!vd->vdev_top->vdev_removing);
+
+ vdev_initialize_change_state(vd, VDEV_INITIALIZE_ACTIVE);
+ vd->vdev_initialize_thread = thread_create(NULL, 0,
+ vdev_initialize_thread, vd, 0, &p0, TS_RUN, maxclsyspri);
+}
+
+/*
+ * Stop initializng a device, with the resultant initialing state being
+ * tgt_state. Blocks until the initializing thread has exited.
+ * Caller must hold vdev_initialize_lock and must not be writing to the spa
+ * config, as the initializing thread may try to enter the config as a reader
+ * before exiting.
+ */
+void
+vdev_initialize_stop(vdev_t *vd, vdev_initializing_state_t tgt_state)
+{
+ ASSERTV(spa_t *spa = vd->vdev_spa);
+ ASSERT(!spa_config_held(spa, SCL_CONFIG | SCL_STATE, RW_WRITER));
+
+ ASSERT(MUTEX_HELD(&vd->vdev_initialize_lock));
+ ASSERT(vd->vdev_ops->vdev_op_leaf);
+ ASSERT(vdev_is_concrete(vd));
+
+ /*
+ * Allow cancel requests to proceed even if the initialize thread
+ * has stopped.
+ */
+ if (vd->vdev_initialize_thread == NULL &&
+ tgt_state != VDEV_INITIALIZE_CANCELED) {
+ return;
+ }
+
+ vdev_initialize_change_state(vd, tgt_state);
+ vd->vdev_initialize_exit_wanted = B_TRUE;
+ while (vd->vdev_initialize_thread != NULL)
+ cv_wait(&vd->vdev_initialize_cv, &vd->vdev_initialize_lock);
+
+ ASSERT3P(vd->vdev_initialize_thread, ==, NULL);
+ vd->vdev_initialize_exit_wanted = B_FALSE;
+}
+
+static void
+vdev_initialize_stop_all_impl(vdev_t *vd, vdev_initializing_state_t tgt_state)
+{
+ if (vd->vdev_ops->vdev_op_leaf && vdev_is_concrete(vd)) {
+ mutex_enter(&vd->vdev_initialize_lock);
+ vdev_initialize_stop(vd, tgt_state);
+ mutex_exit(&vd->vdev_initialize_lock);
+ return;
+ }
+
+ for (uint64_t i = 0; i < vd->vdev_children; i++) {
+ vdev_initialize_stop_all_impl(vd->vdev_child[i], tgt_state);
+ }
+}
+
+/*
+ * Convenience function to stop initializing of a vdev tree and set all
+ * initialize thread pointers to NULL.
+ */
+void
+vdev_initialize_stop_all(vdev_t *vd, vdev_initializing_state_t tgt_state)
+{
+ vdev_initialize_stop_all_impl(vd, tgt_state);
+
+ if (vd->vdev_spa->spa_sync_on) {
+ /* Make sure that our state has been synced to disk */
+ txg_wait_synced(spa_get_dsl(vd->vdev_spa), 0);
+ }
+}
+
+void
+vdev_initialize_restart(vdev_t *vd)
+{
+ ASSERT(MUTEX_HELD(&spa_namespace_lock));
+ ASSERT(!spa_config_held(vd->vdev_spa, SCL_ALL, RW_WRITER));
+
+ if (vd->vdev_leaf_zap != 0) {
+ mutex_enter(&vd->vdev_initialize_lock);
+ uint64_t initialize_state = VDEV_INITIALIZE_NONE;
+ int err = zap_lookup(vd->vdev_spa->spa_meta_objset,
+ vd->vdev_leaf_zap, VDEV_LEAF_ZAP_INITIALIZE_STATE,
+ sizeof (initialize_state), 1, &initialize_state);
+ ASSERT(err == 0 || err == ENOENT);
+ vd->vdev_initialize_state = initialize_state;
+
+ uint64_t timestamp = 0;
+ err = zap_lookup(vd->vdev_spa->spa_meta_objset,
+ vd->vdev_leaf_zap, VDEV_LEAF_ZAP_INITIALIZE_ACTION_TIME,
+ sizeof (timestamp), 1, &timestamp);
+ ASSERT(err == 0 || err == ENOENT);
+ vd->vdev_initialize_action_time = (time_t)timestamp;
+
+ if (vd->vdev_initialize_state == VDEV_INITIALIZE_SUSPENDED ||
+ vd->vdev_offline) {
+ /* load progress for reporting, but don't resume */
+ VERIFY0(vdev_initialize_load(vd));
+ } else if (vd->vdev_initialize_state ==
+ VDEV_INITIALIZE_ACTIVE && vdev_writeable(vd)) {
+ vdev_initialize(vd);
+ }
+
+ mutex_exit(&vd->vdev_initialize_lock);
+ }
+
+ for (uint64_t i = 0; i < vd->vdev_children; i++) {
+ vdev_initialize_restart(vd->vdev_child[i]);
+ }
+}
+
+#if defined(_KERNEL)
+EXPORT_SYMBOL(vdev_initialize_restart);
+EXPORT_SYMBOL(vdev_xlate);
+EXPORT_SYMBOL(vdev_initialize_stop_all);
+EXPORT_SYMBOL(vdev_initialize);
+EXPORT_SYMBOL(vdev_initialize_stop);
+
+/* CSTYLED */
+module_param(zfs_initialize_value, ulong, 0644);
+MODULE_PARM_DESC(zfs_initialize_value,
+ "Value written during zpool initialize");
+#endif
diff --git a/module/zfs/vdev_mirror.c b/module/zfs/vdev_mirror.c
index 65357d841..b45c05db2 100644
--- a/module/zfs/vdev_mirror.c
+++ b/module/zfs/vdev_mirror.c
@@ -740,6 +740,7 @@ vdev_ops_t vdev_mirror_ops = {
NULL,
NULL,
NULL,
+ vdev_default_xlate,
VDEV_TYPE_MIRROR, /* name of this vdev type */
B_FALSE /* not a leaf vdev */
};
@@ -755,6 +756,7 @@ vdev_ops_t vdev_replacing_ops = {
NULL,
NULL,
NULL,
+ vdev_default_xlate,
VDEV_TYPE_REPLACING, /* name of this vdev type */
B_FALSE /* not a leaf vdev */
};
@@ -770,6 +772,7 @@ vdev_ops_t vdev_spare_ops = {
NULL,
NULL,
NULL,
+ vdev_default_xlate,
VDEV_TYPE_SPARE, /* name of this vdev type */
B_FALSE /* not a leaf vdev */
};
diff --git a/module/zfs/vdev_missing.c b/module/zfs/vdev_missing.c
index b1c039f16..d85993bff 100644
--- a/module/zfs/vdev_missing.c
+++ b/module/zfs/vdev_missing.c
@@ -24,7 +24,7 @@
*/
/*
- * Copyright (c) 2012, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2012, 2016 by Delphix. All rights reserved.
*/
/*
@@ -90,6 +90,7 @@ vdev_ops_t vdev_missing_ops = {
NULL,
NULL,
NULL,
+ NULL,
VDEV_TYPE_MISSING, /* name of this vdev type */
B_TRUE /* leaf vdev */
};
@@ -105,6 +106,7 @@ vdev_ops_t vdev_hole_ops = {
NULL,
NULL,
NULL,
+ NULL,
VDEV_TYPE_HOLE, /* name of this vdev type */
B_TRUE /* leaf vdev */
};
diff --git a/module/zfs/vdev_queue.c b/module/zfs/vdev_queue.c
index 89cdf7d81..939699cb8 100644
--- a/module/zfs/vdev_queue.c
+++ b/module/zfs/vdev_queue.c
@@ -154,6 +154,8 @@ uint32_t zfs_vdev_scrub_min_active = 1;
uint32_t zfs_vdev_scrub_max_active = 2;
uint32_t zfs_vdev_removal_min_active = 1;
uint32_t zfs_vdev_removal_max_active = 2;
+uint32_t zfs_vdev_initializing_min_active = 1;
+uint32_t zfs_vdev_initializing_max_active = 1;
/*
* When the pool has less than zfs_vdev_async_write_active_min_dirty_percent
@@ -261,6 +263,8 @@ vdev_queue_class_min_active(zio_priority_t p)
return (zfs_vdev_scrub_min_active);
case ZIO_PRIORITY_REMOVAL:
return (zfs_vdev_removal_min_active);
+ case ZIO_PRIORITY_INITIALIZING:
+ return (zfs_vdev_initializing_min_active);
default:
panic("invalid priority %u", p);
return (0);
@@ -331,6 +335,8 @@ vdev_queue_class_max_active(spa_t *spa, zio_priority_t p)
return (zfs_vdev_scrub_max_active);
case ZIO_PRIORITY_REMOVAL:
return (zfs_vdev_removal_max_active);
+ case ZIO_PRIORITY_INITIALIZING:
+ return (zfs_vdev_initializing_max_active);
default:
panic("invalid priority %u", p);
return (0);
@@ -718,8 +724,8 @@ again:
}
/*
- * For LBA-ordered queues (async / scrub), issue the i/o which follows
- * the most recently issued i/o in LBA (offset) order.
+ * For LBA-ordered queues (async / scrub / initializing), issue the
+ * i/o which follows the most recently issued i/o in LBA (offset) order.
*
* For FIFO queues (sync), issue the i/o with the lowest timestamp.
*/
@@ -775,13 +781,15 @@ vdev_queue_io(zio_t *zio)
if (zio->io_priority != ZIO_PRIORITY_SYNC_READ &&
zio->io_priority != ZIO_PRIORITY_ASYNC_READ &&
zio->io_priority != ZIO_PRIORITY_SCRUB &&
- zio->io_priority != ZIO_PRIORITY_REMOVAL)
+ zio->io_priority != ZIO_PRIORITY_REMOVAL &&
+ zio->io_priority != ZIO_PRIORITY_INITIALIZING)
zio->io_priority = ZIO_PRIORITY_ASYNC_READ;
} else {
ASSERT(zio->io_type == ZIO_TYPE_WRITE);
if (zio->io_priority != ZIO_PRIORITY_SYNC_WRITE &&
zio->io_priority != ZIO_PRIORITY_ASYNC_WRITE &&
- zio->io_priority != ZIO_PRIORITY_REMOVAL)
+ zio->io_priority != ZIO_PRIORITY_REMOVAL &&
+ zio->io_priority != ZIO_PRIORITY_INITIALIZING)
zio->io_priority = ZIO_PRIORITY_ASYNC_WRITE;
}
@@ -938,11 +946,29 @@ module_param(zfs_vdev_async_write_min_active, int, 0644);
MODULE_PARM_DESC(zfs_vdev_async_write_min_active,
"Min active async write I/Os per vdev");
+module_param(zfs_vdev_initializing_max_active, int, 0644);
+MODULE_PARM_DESC(zfs_vdev_initializing_max_active,
+ "Max active initializing I/Os per vdev");
+
+module_param(zfs_vdev_initializing_min_active, int, 0644);
+MODULE_PARM_DESC(zfs_vdev_initializing_min_active,
+ "Min active initializing I/Os per vdev");
+
+module_param(zfs_vdev_removal_max_active, int, 0644);
+MODULE_PARM_DESC(zfs_vdev_removal_max_active,
+ "Max active removal I/Os per vdev");
+
+module_param(zfs_vdev_removal_min_active, int, 0644);
+MODULE_PARM_DESC(zfs_vdev_removal_min_active,
+ "Min active removal I/Os per vdev");
+
module_param(zfs_vdev_scrub_max_active, int, 0644);
-MODULE_PARM_DESC(zfs_vdev_scrub_max_active, "Max active scrub I/Os per vdev");
+MODULE_PARM_DESC(zfs_vdev_scrub_max_active,
+ "Max active scrub I/Os per vdev");
module_param(zfs_vdev_scrub_min_active, int, 0644);
-MODULE_PARM_DESC(zfs_vdev_scrub_min_active, "Min active scrub I/Os per vdev");
+MODULE_PARM_DESC(zfs_vdev_scrub_min_active,
+ "Min active scrub I/Os per vdev");
module_param(zfs_vdev_sync_read_max_active, int, 0644);
MODULE_PARM_DESC(zfs_vdev_sync_read_max_active,
diff --git a/module/zfs/vdev_raidz.c b/module/zfs/vdev_raidz.c
index a21baf9c2..d10d89f3e 100644
--- a/module/zfs/vdev_raidz.c
+++ b/module/zfs/vdev_raidz.c
@@ -36,6 +36,10 @@
#include <sys/vdev_raidz.h>
#include <sys/vdev_raidz_impl.h>
+#ifdef ZFS_DEBUG
+#include <sys/vdev_initialize.h> /* vdev_xlate testing */
+#endif
+
/*
* Virtual device vector for RAID-Z.
*
@@ -1627,6 +1631,39 @@ vdev_raidz_child_done(zio_t *zio)
rc->rc_skipped = 0;
}
+static void
+vdev_raidz_io_verify(zio_t *zio, raidz_map_t *rm, int col)
+{
+#ifdef ZFS_DEBUG
+ vdev_t *vd = zio->io_vd;
+ vdev_t *tvd = vd->vdev_top;
+
+ range_seg_t logical_rs, physical_rs;
+ logical_rs.rs_start = zio->io_offset;
+ logical_rs.rs_end = logical_rs.rs_start +
+ vdev_raidz_asize(zio->io_vd, zio->io_size);
+
+ raidz_col_t *rc = &rm->rm_col[col];
+ vdev_t *cvd = vd->vdev_child[rc->rc_devidx];
+
+ vdev_xlate(cvd, &logical_rs, &physical_rs);
+ ASSERT3U(rc->rc_offset, ==, physical_rs.rs_start);
+ ASSERT3U(rc->rc_offset, <, physical_rs.rs_end);
+ /*
+ * It would be nice to assert that rs_end is equal
+ * to rc_offset + rc_size but there might be an
+ * optional I/O at the end that is not accounted in
+ * rc_size.
+ */
+ if (physical_rs.rs_end > rc->rc_offset + rc->rc_size) {
+ ASSERT3U(physical_rs.rs_end, ==, rc->rc_offset +
+ rc->rc_size + (1 << tvd->vdev_ashift));
+ } else {
+ ASSERT3U(physical_rs.rs_end, ==, rc->rc_offset + rc->rc_size);
+ }
+#endif
+}
+
/*
* Start an IO operation on a RAIDZ VDev
*
@@ -1665,6 +1702,12 @@ vdev_raidz_io_start(zio_t *zio)
for (c = 0; c < rm->rm_cols; c++) {
rc = &rm->rm_col[c];
cvd = vd->vdev_child[rc->rc_devidx];
+
+ /*
+ * Verify physical to logical translation.
+ */
+ vdev_raidz_io_verify(zio, rm, c);
+
zio_nowait(zio_vdev_child_io(zio, NULL, cvd,
rc->rc_offset, rc->rc_abd, rc->rc_size,
zio->io_type, zio->io_priority, 0,
@@ -2323,6 +2366,37 @@ vdev_raidz_need_resilver(vdev_t *vd, uint64_t offset, size_t psize)
return (B_FALSE);
}
+static void
+vdev_raidz_xlate(vdev_t *cvd, const range_seg_t *in, range_seg_t *res)
+{
+ vdev_t *raidvd = cvd->vdev_parent;
+ ASSERT(raidvd->vdev_ops == &vdev_raidz_ops);
+
+ uint64_t width = raidvd->vdev_children;
+ uint64_t tgt_col = cvd->vdev_id;
+ uint64_t ashift = raidvd->vdev_top->vdev_ashift;
+
+ /* make sure the offsets are block-aligned */
+ ASSERT0(in->rs_start % (1 << ashift));
+ ASSERT0(in->rs_end % (1 << ashift));
+ uint64_t b_start = in->rs_start >> ashift;
+ uint64_t b_end = in->rs_end >> ashift;
+
+ uint64_t start_row = 0;
+ if (b_start > tgt_col) /* avoid underflow */
+ start_row = ((b_start - tgt_col - 1) / width) + 1;
+
+ uint64_t end_row = 0;
+ if (b_end > tgt_col)
+ end_row = ((b_end - tgt_col - 1) / width) + 1;
+
+ res->rs_start = start_row << ashift;
+ res->rs_end = end_row << ashift;
+
+ ASSERT3U(res->rs_start, <=, in->rs_start);
+ ASSERT3U(res->rs_end - res->rs_start, <=, in->rs_end - in->rs_start);
+}
+
vdev_ops_t vdev_raidz_ops = {
vdev_raidz_open,
vdev_raidz_close,
@@ -2334,6 +2408,7 @@ vdev_ops_t vdev_raidz_ops = {
NULL,
NULL,
NULL,
+ vdev_raidz_xlate,
VDEV_TYPE_RAIDZ, /* name of this vdev type */
B_FALSE /* not a leaf vdev */
};
diff --git a/module/zfs/vdev_removal.c b/module/zfs/vdev_removal.c
index a706bc2a4..d0824aa84 100644
--- a/module/zfs/vdev_removal.c
+++ b/module/zfs/vdev_removal.c
@@ -44,6 +44,7 @@
#include <sys/vdev_indirect_births.h>
#include <sys/vdev_indirect_mapping.h>
#include <sys/abd.h>
+#include <sys/vdev_initialize.h>
#include <sys/trace_vdev.h>
/*
@@ -1186,6 +1187,7 @@ vdev_remove_complete(spa_t *spa)
txg_wait_synced(spa->spa_dsl_pool, 0);
txg = spa_vdev_enter(spa);
vdev_t *vd = vdev_lookup_top(spa, spa->spa_vdev_removal->svr_vdev_id);
+ ASSERT3P(vd->vdev_initialize_thread, ==, NULL);
sysevent_t *ev = spa_event_create(spa, vd, NULL,
ESC_ZFS_VDEV_REMOVE_DEV);
@@ -1896,6 +1898,9 @@ spa_vdev_remove_log(vdev_t *vd, uint64_t *txg)
spa_vdev_config_exit(spa, NULL, *txg, 0, FTAG);
+ /* Stop initializing */
+ (void) vdev_initialize_stop_all(vd, VDEV_INITIALIZE_CANCELED);
+
*txg = spa_vdev_config_enter(spa);
sysevent_t *ev = spa_event_create(spa, vd, NULL,
@@ -2072,6 +2077,13 @@ spa_vdev_remove_top(vdev_t *vd, uint64_t *txg)
*/
error = spa_reset_logs(spa);
+ /*
+ * We stop any initializing that is currently in progress but leave
+ * the state as "active". This will allow the initializing to resume
+ * if the removal is canceled sometime later.
+ */
+ vdev_initialize_stop_all(vd, VDEV_INITIALIZE_ACTIVE);
+
*txg = spa_vdev_config_enter(spa);
/*
@@ -2083,6 +2095,7 @@ spa_vdev_remove_top(vdev_t *vd, uint64_t *txg)
if (error != 0) {
metaslab_group_activate(mg);
+ spa_async_request(spa, SPA_ASYNC_INITIALIZE_RESTART);
return (error);
}
diff --git a/module/zfs/vdev_root.c b/module/zfs/vdev_root.c
index 9f86cbfa4..e40b7ce8e 100644
--- a/module/zfs/vdev_root.c
+++ b/module/zfs/vdev_root.c
@@ -24,7 +24,7 @@
*/
/*
- * Copyright (c) 2012, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2012, 2016 by Delphix. All rights reserved.
*/
#include <sys/zfs_context.h>
@@ -150,6 +150,7 @@ vdev_ops_t vdev_root_ops = {
NULL,
NULL,
NULL,
+ NULL,
VDEV_TYPE_ROOT, /* name of this vdev type */
B_FALSE /* not a leaf vdev */
};
diff --git a/module/zfs/zfs_ioctl.c b/module/zfs/zfs_ioctl.c
index a71da2837..3c36502d8 100644
--- a/module/zfs/zfs_ioctl.c
+++ b/module/zfs/zfs_ioctl.c
@@ -202,6 +202,8 @@
#include <sys/zio_checksum.h>
#include <sys/vdev_removal.h>
#include <sys/zfs_sysfs.h>
+#include <sys/vdev_impl.h>
+#include <sys/vdev_initialize.h>
#include <linux/miscdevice.h>
#include <linux/slab.h>
@@ -3843,6 +3845,85 @@ zfs_ioc_destroy(zfs_cmd_t *zc)
}
/*
+ * innvl: {
+ * vdevs: {
+ * guid 1, guid 2, ...
+ * },
+ * func: POOL_INITIALIZE_{CANCEL|DO|SUSPEND}
+ * }
+ *
+ * outnvl: {
+ * [func: EINVAL (if provided command type didn't make sense)],
+ * [vdevs: {
+ * guid1: errno, (see function body for possible errnos)
+ * ...
+ * }]
+ * }
+ *
+ */
+static const zfs_ioc_key_t zfs_keys_pool_initialize[] = {
+ {ZPOOL_INITIALIZE_COMMAND, DATA_TYPE_UINT64, 0},
+ {ZPOOL_INITIALIZE_VDEVS, DATA_TYPE_NVLIST, 0}
+};
+
+static int
+zfs_ioc_pool_initialize(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl)
+{
+ spa_t *spa;
+ int error;
+
+ error = spa_open(poolname, &spa, FTAG);
+ if (error != 0)
+ return (error);
+
+ uint64_t cmd_type;
+ if (nvlist_lookup_uint64(innvl, ZPOOL_INITIALIZE_COMMAND,
+ &cmd_type) != 0) {
+ spa_close(spa, FTAG);
+ return (SET_ERROR(EINVAL));
+ }
+ if (!(cmd_type == POOL_INITIALIZE_CANCEL ||
+ cmd_type == POOL_INITIALIZE_DO ||
+ cmd_type == POOL_INITIALIZE_SUSPEND)) {
+ spa_close(spa, FTAG);
+ return (SET_ERROR(EINVAL));
+ }
+
+ nvlist_t *vdev_guids;
+ if (nvlist_lookup_nvlist(innvl, ZPOOL_INITIALIZE_VDEVS,
+ &vdev_guids) != 0) {
+ spa_close(spa, FTAG);
+ return (SET_ERROR(EINVAL));
+ }
+
+ nvlist_t *vdev_errlist = fnvlist_alloc();
+ int total_errors = 0;
+
+ for (nvpair_t *pair = nvlist_next_nvpair(vdev_guids, NULL);
+ pair != NULL; pair = nvlist_next_nvpair(vdev_guids, pair)) {
+ uint64_t vdev_guid = fnvpair_value_uint64(pair);
+
+ error = spa_vdev_initialize(spa, vdev_guid, cmd_type);
+ if (error != 0) {
+ char guid_as_str[MAXNAMELEN];
+
+ (void) snprintf(guid_as_str, sizeof (guid_as_str),
+ "%llu", (unsigned long long)vdev_guid);
+ fnvlist_add_int64(vdev_errlist, guid_as_str, error);
+ total_errors++;
+ }
+ }
+ if (fnvlist_size(vdev_errlist) > 0) {
+ fnvlist_add_nvlist(outnvl, ZPOOL_INITIALIZE_VDEVS,
+ vdev_errlist);
+ }
+ fnvlist_free(vdev_errlist);
+
+ spa_close(spa, FTAG);
+ return (total_errors > 0 ? EINVAL : 0);
+}
+
+/*
* fsname is name of dataset to rollback (to most recent snapshot)
*
* innvl may contain name of expected target snapshot
@@ -6453,6 +6534,11 @@ zfs_ioctl_init(void)
zfs_keys_pool_discard_checkpoint,
ARRAY_SIZE(zfs_keys_pool_discard_checkpoint));
+ zfs_ioctl_register("initialize", ZFS_IOC_POOL_INITIALIZE,
+ zfs_ioc_pool_initialize, zfs_secpolicy_config, POOL_NAME,
+ POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE,
+ zfs_keys_pool_initialize, ARRAY_SIZE(zfs_keys_pool_initialize));
+
/* IOCTLS that use the legacy function signature */
zfs_ioctl_register_legacy(ZFS_IOC_POOL_FREEZE, zfs_ioc_pool_freeze,
diff --git a/module/zfs/zfs_sysfs.c b/module/zfs/zfs_sysfs.c
index b17c91f65..87c4ac117 100644
--- a/module/zfs/zfs_sysfs.c
+++ b/module/zfs/zfs_sysfs.c
@@ -358,6 +358,7 @@ pool_property_show(struct kobject *kobj, struct attribute *attr, char *buf)
*/
static const char *zfs_features[] = {
/* --> Add new kernel features here (post ZoL 0.8.0) */
+ "vdev_initialize"
};
#define ZFS_FEATURE_COUNT ARRAY_SIZE(zfs_features)