zfs/module/zfs/spa.c

5709 lines
148 KiB
C

/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
*/
/*
* This file contains all the routines used when modifying on-disk SPA state.
* This includes opening, importing, destroying, exporting a pool, and syncing a
* pool.
*/
#include <sys/zfs_context.h>
#include <sys/fm/fs/zfs.h>
#include <sys/spa_impl.h>
#include <sys/zio.h>
#include <sys/zio_checksum.h>
#include <sys/dmu.h>
#include <sys/dmu_tx.h>
#include <sys/zap.h>
#include <sys/zil.h>
#include <sys/ddt.h>
#include <sys/vdev_impl.h>
#include <sys/vdev_disk.h>
#include <sys/metaslab.h>
#include <sys/metaslab_impl.h>
#include <sys/uberblock_impl.h>
#include <sys/txg.h>
#include <sys/avl.h>
#include <sys/dmu_traverse.h>
#include <sys/dmu_objset.h>
#include <sys/unique.h>
#include <sys/dsl_pool.h>
#include <sys/dsl_dataset.h>
#include <sys/dsl_dir.h>
#include <sys/dsl_prop.h>
#include <sys/dsl_synctask.h>
#include <sys/fs/zfs.h>
#include <sys/arc.h>
#include <sys/callb.h>
#include <sys/systeminfo.h>
#include <sys/spa_boot.h>
#include <sys/zfs_ioctl.h>
#include <sys/dsl_scan.h>
#ifdef _KERNEL
#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 */
#include "zfs_prop.h"
#include "zfs_comutil.h"
typedef enum zti_modes {
zti_mode_fixed, /* value is # of threads (min 1) */
zti_mode_online_percent, /* value is % of online CPUs */
zti_mode_batch, /* cpu-intensive; value is ignored */
zti_mode_null, /* don't create a taskq */
zti_nmodes
} zti_modes_t;
#define ZTI_FIX(n) { zti_mode_fixed, (n) }
#define ZTI_PCT(n) { zti_mode_online_percent, (n) }
#define ZTI_BATCH { zti_mode_batch, 0 }
#define ZTI_NULL { zti_mode_null, 0 }
#define ZTI_ONE ZTI_FIX(1)
typedef struct zio_taskq_info {
enum zti_modes zti_mode;
uint_t zti_value;
} zio_taskq_info_t;
static const char *const zio_taskq_types[ZIO_TASKQ_TYPES] = {
"issue", "issue_high", "intr", "intr_high"
};
/*
* Define the taskq threads for the following I/O types:
* NULL, READ, WRITE, FREE, CLAIM, and IOCTL
*/
const zio_taskq_info_t zio_taskqs[ZIO_TYPES][ZIO_TASKQ_TYPES] = {
/* ISSUE ISSUE_HIGH INTR INTR_HIGH */
{ ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL },
{ ZTI_FIX(8), ZTI_NULL, ZTI_BATCH, ZTI_NULL },
{ ZTI_BATCH, ZTI_FIX(5), ZTI_FIX(8), ZTI_FIX(5) },
{ ZTI_FIX(100), ZTI_NULL, ZTI_ONE, ZTI_NULL },
{ ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL },
{ ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL },
};
static dsl_syncfunc_t spa_sync_props;
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 mosconfig,
char **ereport);
uint_t zio_taskq_batch_pct = 100; /* 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 */
/*
* 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"
/*
* ==========================================================================
* SPA properties routines
* ==========================================================================
*/
/*
* Add a (source=src, propname=propval) list to an nvlist.
*/
static void
spa_prop_add_list(nvlist_t *nvl, zpool_prop_t prop, char *strval,
uint64_t intval, zprop_source_t src)
{
const char *propname = zpool_prop_to_name(prop);
nvlist_t *propval;
VERIFY(nvlist_alloc(&propval, NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_uint64(propval, ZPROP_SOURCE, src) == 0);
if (strval != NULL)
VERIFY(nvlist_add_string(propval, ZPROP_VALUE, strval) == 0);
else
VERIFY(nvlist_add_uint64(propval, ZPROP_VALUE, intval) == 0);
VERIFY(nvlist_add_nvlist(nvl, propname, propval) == 0);
nvlist_free(propval);
}
/*
* Get property values from the spa configuration.
*/
static void
spa_prop_get_config(spa_t *spa, nvlist_t **nvp)
{
uint64_t size;
uint64_t alloc;
uint64_t cap, version;
zprop_source_t src = ZPROP_SRC_NONE;
spa_config_dirent_t *dp;
ASSERT(MUTEX_HELD(&spa->spa_props_lock));
if (spa->spa_root_vdev != NULL) {
alloc = metaslab_class_get_alloc(spa_normal_class(spa));
size = metaslab_class_get_space(spa_normal_class(spa));
spa_prop_add_list(*nvp, ZPOOL_PROP_NAME, spa_name(spa), 0, src);
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);
cap = (size == 0) ? 0 : (alloc * 100 / size);
spa_prop_add_list(*nvp, ZPOOL_PROP_CAPACITY, NULL, cap, src);
spa_prop_add_list(*nvp, ZPOOL_PROP_DEDUPRATIO, NULL,
ddt_get_pool_dedup_ratio(spa), src);
spa_prop_add_list(*nvp, ZPOOL_PROP_HEALTH, NULL,
spa->spa_root_vdev->vdev_state, src);
version = spa_version(spa);
if (version == zpool_prop_default_numeric(ZPOOL_PROP_VERSION))
src = ZPROP_SRC_DEFAULT;
else
src = ZPROP_SRC_LOCAL;
spa_prop_add_list(*nvp, ZPOOL_PROP_VERSION, NULL, version, src);
}
spa_prop_add_list(*nvp, ZPOOL_PROP_GUID, NULL, spa_guid(spa), src);
if (spa->spa_root != NULL)
spa_prop_add_list(*nvp, ZPOOL_PROP_ALTROOT, spa->spa_root,
0, ZPROP_SRC_LOCAL);
if ((dp = list_head(&spa->spa_config_list)) != NULL) {
if (dp->scd_path == NULL) {
spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE,
"none", 0, ZPROP_SRC_LOCAL);
} else if (strcmp(dp->scd_path, spa_config_path) != 0) {
spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE,
dp->scd_path, 0, ZPROP_SRC_LOCAL);
}
}
}
/*
* Get zpool property values.
*/
int
spa_prop_get(spa_t *spa, nvlist_t **nvp)
{
objset_t *mos = spa->spa_meta_objset;
zap_cursor_t zc;
zap_attribute_t za;
int err;
err = nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_SLEEP);
if (err)
return err;
mutex_enter(&spa->spa_props_lock);
/*
* Get properties from the spa config.
*/
spa_prop_get_config(spa, nvp);
/* If no pool property object, no more prop to get. */
if (mos == NULL || spa->spa_pool_props_object == 0) {
mutex_exit(&spa->spa_props_lock);
goto out;
}
/*
* Get properties from the MOS pool property object.
*/
for (zap_cursor_init(&zc, mos, spa->spa_pool_props_object);
(err = zap_cursor_retrieve(&zc, &za)) == 0;
zap_cursor_advance(&zc)) {
uint64_t intval = 0;
char *strval = NULL;
zprop_source_t src = ZPROP_SRC_DEFAULT;
zpool_prop_t prop;
if ((prop = zpool_name_to_prop(za.za_name)) == ZPROP_INVAL)
continue;
switch (za.za_integer_length) {
case 8:
/* integer property */
if (za.za_first_integer !=
zpool_prop_default_numeric(prop))
src = ZPROP_SRC_LOCAL;
if (prop == ZPOOL_PROP_BOOTFS) {
dsl_pool_t *dp;
dsl_dataset_t *ds = NULL;
dp = spa_get_dsl(spa);
rw_enter(&dp->dp_config_rwlock, RW_READER);
if ((err = dsl_dataset_hold_obj(dp,
za.za_first_integer, FTAG, &ds))) {
rw_exit(&dp->dp_config_rwlock);
break;
}
strval = kmem_alloc(
MAXNAMELEN + strlen(MOS_DIR_NAME) + 1,
KM_SLEEP);
dsl_dataset_name(ds, strval);
dsl_dataset_rele(ds, FTAG);
rw_exit(&dp->dp_config_rwlock);
} else {
strval = NULL;
intval = za.za_first_integer;
}
spa_prop_add_list(*nvp, prop, strval, intval, src);
if (strval != NULL)
kmem_free(strval,
MAXNAMELEN + strlen(MOS_DIR_NAME) + 1);
break;
case 1:
/* string property */
strval = kmem_alloc(za.za_num_integers, KM_SLEEP);
err = zap_lookup(mos, spa->spa_pool_props_object,
za.za_name, 1, za.za_num_integers, strval);
if (err) {
kmem_free(strval, za.za_num_integers);
break;
}
spa_prop_add_list(*nvp, prop, strval, 0, src);
kmem_free(strval, za.za_num_integers);
break;
default:
break;
}
}
zap_cursor_fini(&zc);
mutex_exit(&spa->spa_props_lock);
out:
if (err && err != ENOENT) {
nvlist_free(*nvp);
*nvp = NULL;
return (err);
}
return (0);
}
/*
* Validate the given pool properties nvlist and modify the list
* for the property values to be set.
*/
static int
spa_prop_validate(spa_t *spa, nvlist_t *props)
{
nvpair_t *elem;
int error = 0, reset_bootfs = 0;
uint64_t objnum = 0;
elem = NULL;
while ((elem = nvlist_next_nvpair(props, elem)) != NULL) {
zpool_prop_t prop;
char *propname, *strval;
uint64_t intval;
objset_t *os;
char *slash;
propname = nvpair_name(elem);
if ((prop = zpool_name_to_prop(propname)) == ZPROP_INVAL)
return (EINVAL);
switch (prop) {
case ZPOOL_PROP_VERSION:
error = nvpair_value_uint64(elem, &intval);
if (!error &&
(intval < spa_version(spa) || intval > SPA_VERSION))
error = EINVAL;
break;
case ZPOOL_PROP_DELEGATION:
case ZPOOL_PROP_AUTOREPLACE:
case ZPOOL_PROP_LISTSNAPS:
case ZPOOL_PROP_AUTOEXPAND:
error = nvpair_value_uint64(elem, &intval);
if (!error && intval > 1)
error = EINVAL;
break;
case ZPOOL_PROP_BOOTFS:
/*
* If the pool version is less than SPA_VERSION_BOOTFS,
* or the pool is still being created (version == 0),
* the bootfs property cannot be set.
*/
if (spa_version(spa) < SPA_VERSION_BOOTFS) {
error = ENOTSUP;
break;
}
/*
* Make sure the vdev config is bootable
*/
if (!vdev_is_bootable(spa->spa_root_vdev)) {
error = ENOTSUP;
break;
}
reset_bootfs = 1;
error = nvpair_value_string(elem, &strval);
if (!error) {
uint64_t compress;
if (strval == NULL || strval[0] == '\0') {
objnum = zpool_prop_default_numeric(
ZPOOL_PROP_BOOTFS);
break;
}
if ((error = dmu_objset_hold(strval,FTAG,&os)))
break;
/* Must be ZPL and not gzip compressed. */
if (dmu_objset_type(os) != DMU_OST_ZFS) {
error = ENOTSUP;
} else if ((error = dsl_prop_get_integer(strval,
zfs_prop_to_name(ZFS_PROP_COMPRESSION),
&compress, NULL)) == 0 &&
!BOOTFS_COMPRESS_VALID(compress)) {
error = ENOTSUP;
} else {
objnum = dmu_objset_id(os);
}
dmu_objset_rele(os, FTAG);
}
break;
case ZPOOL_PROP_FAILUREMODE:
error = nvpair_value_uint64(elem, &intval);
if (!error && (intval < ZIO_FAILURE_MODE_WAIT ||
intval > ZIO_FAILURE_MODE_PANIC))
error = EINVAL;
/*
* This is a special case which only occurs when
* the pool has completely failed. This allows
* the user to change the in-core failmode property
* without syncing it out to disk (I/Os might
* currently be blocked). We do this by returning
* EIO to the caller (spa_prop_set) to trick it
* into thinking we encountered a property validation
* error.
*/
if (!error && spa_suspended(spa)) {
spa->spa_failmode = intval;
error = EIO;
}
break;
case ZPOOL_PROP_CACHEFILE:
if ((error = nvpair_value_string(elem, &strval)) != 0)
break;
if (strval[0] == '\0')
break;
if (strcmp(strval, "none") == 0)
break;
if (strval[0] != '/') {
error = EINVAL;
break;
}
slash = strrchr(strval, '/');
ASSERT(slash != NULL);
if (slash[1] == '\0' || strcmp(slash, "/.") == 0 ||
strcmp(slash, "/..") == 0)
error = EINVAL;
break;
case ZPOOL_PROP_DEDUPDITTO:
if (spa_version(spa) < SPA_VERSION_DEDUP)
error = ENOTSUP;
else
error = nvpair_value_uint64(elem, &intval);
if (error == 0 &&
intval != 0 && intval < ZIO_DEDUPDITTO_MIN)
error = EINVAL;
break;
default:
break;
}
if (error)
break;
}
if (!error && reset_bootfs) {
error = nvlist_remove(props,
zpool_prop_to_name(ZPOOL_PROP_BOOTFS), DATA_TYPE_STRING);
if (!error) {
error = nvlist_add_uint64(props,
zpool_prop_to_name(ZPOOL_PROP_BOOTFS), objnum);
}
}
return (error);
}
void
spa_configfile_set(spa_t *spa, nvlist_t *nvp, boolean_t need_sync)
{
char *cachefile;
spa_config_dirent_t *dp;
if (nvlist_lookup_string(nvp, zpool_prop_to_name(ZPOOL_PROP_CACHEFILE),
&cachefile) != 0)
return;
dp = kmem_alloc(sizeof (spa_config_dirent_t),
KM_SLEEP);
if (cachefile[0] == '\0')
dp->scd_path = spa_strdup(spa_config_path);
else if (strcmp(cachefile, "none") == 0)
dp->scd_path = NULL;
else
dp->scd_path = spa_strdup(cachefile);
list_insert_head(&spa->spa_config_list, dp);
if (need_sync)
spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE);
}
int
spa_prop_set(spa_t *spa, nvlist_t *nvp)
{
int error;
nvpair_t *elem;
boolean_t need_sync = B_FALSE;
zpool_prop_t prop;
if ((error = spa_prop_validate(spa, nvp)) != 0)
return (error);
elem = NULL;
while ((elem = nvlist_next_nvpair(nvp, elem)) != NULL) {
if ((prop = zpool_name_to_prop(
nvpair_name(elem))) == ZPROP_INVAL)
return (EINVAL);
if (prop == ZPOOL_PROP_CACHEFILE || prop == ZPOOL_PROP_ALTROOT)
continue;
need_sync = B_TRUE;
break;
}
if (need_sync)
return (dsl_sync_task_do(spa_get_dsl(spa), NULL, spa_sync_props,
spa, nvp, 3));
else
return (0);
}
/*
* If the bootfs property value is dsobj, clear it.
*/
void
spa_prop_clear_bootfs(spa_t *spa, uint64_t dsobj, dmu_tx_t *tx)
{
if (spa->spa_bootfs == dsobj && spa->spa_pool_props_object != 0) {
VERIFY(zap_remove(spa->spa_meta_objset,
spa->spa_pool_props_object,
zpool_prop_to_name(ZPOOL_PROP_BOOTFS), tx) == 0);
spa->spa_bootfs = 0;
}
}
/*
* ==========================================================================
* SPA state manipulation (open/create/destroy/import/export)
* ==========================================================================
*/
static int
spa_error_entry_compare(const void *a, const void *b)
{
spa_error_entry_t *sa = (spa_error_entry_t *)a;
spa_error_entry_t *sb = (spa_error_entry_t *)b;
int ret;
ret = bcmp(&sa->se_bookmark, &sb->se_bookmark,
sizeof (zbookmark_t));
if (ret < 0)
return (-1);
else if (ret > 0)
return (1);
else
return (0);
}
/*
* Utility function which retrieves copies of the current logs and
* re-initializes them in the process.
*/
void
spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub)
{
ASSERT(MUTEX_HELD(&spa->spa_errlist_lock));
bcopy(&spa->spa_errlist_last, last, sizeof (avl_tree_t));
bcopy(&spa->spa_errlist_scrub, scrub, sizeof (avl_tree_t));
avl_create(&spa->spa_errlist_scrub,
spa_error_entry_compare, sizeof (spa_error_entry_t),
offsetof(spa_error_entry_t, se_avl));
avl_create(&spa->spa_errlist_last,
spa_error_entry_compare, sizeof (spa_error_entry_t),
offsetof(spa_error_entry_t, se_avl));
}
static taskq_t *
spa_taskq_create(spa_t *spa, const char *name, enum zti_modes mode,
uint_t value)
{
uint_t flags = TASKQ_PREPOPULATE;
boolean_t batch = B_FALSE;
switch (mode) {
case zti_mode_null:
return (NULL); /* no taskq needed */
case zti_mode_fixed:
ASSERT3U(value, >=, 1);
value = MAX(value, 1);
break;
case zti_mode_batch:
batch = B_TRUE;
flags |= TASKQ_THREADS_CPU_PCT;
value = zio_taskq_batch_pct;
break;
case zti_mode_online_percent:
flags |= TASKQ_THREADS_CPU_PCT;
break;
default:
panic("unrecognized mode for %s taskq (%u:%u) in "
"spa_activate()",
name, mode, value);
break;
}
if (zio_taskq_sysdc && spa->spa_proc != &p0) {
if (batch)
flags |= TASKQ_DC_BATCH;
return (taskq_create_sysdc(name, value, 50, INT_MAX,
spa->spa_proc, zio_taskq_basedc, flags));
}
return (taskq_create_proc(name, value, maxclsyspri, 50, INT_MAX,
spa->spa_proc, flags));
}
static void
spa_create_zio_taskqs(spa_t *spa)
{
int t, q;
for (t = 0; t < ZIO_TYPES; t++) {
for (q = 0; q < ZIO_TASKQ_TYPES; q++) {
const zio_taskq_info_t *ztip = &zio_taskqs[t][q];
enum zti_modes mode = ztip->zti_mode;
uint_t value = ztip->zti_value;
char name[32];
(void) snprintf(name, sizeof (name),
"%s_%s", zio_type_name[t], zio_taskq_types[q]);
spa->spa_zio_taskq[t][q] =
spa_taskq_create(spa, name, mode, value);
}
}
}
#if defined(_KERNEL) && defined(HAVE_SPA_THREAD)
static void
spa_thread(void *arg)
{
callb_cpr_t cprinfo;
spa_t *spa = arg;
user_t *pu = PTOU(curproc);
CALLB_CPR_INIT(&cprinfo, &spa->spa_proc_lock, callb_generic_cpr,
spa->spa_name);
ASSERT(curproc != &p0);
(void) snprintf(pu->u_psargs, sizeof (pu->u_psargs),
"zpool-%s", spa->spa_name);
(void) strlcpy(pu->u_comm, pu->u_psargs, sizeof (pu->u_comm));
/* bind this thread to the requested psrset */
if (zio_taskq_psrset_bind != PS_NONE) {
pool_lock();
mutex_enter(&cpu_lock);
mutex_enter(&pidlock);
mutex_enter(&curproc->p_lock);
if (cpupart_bind_thread(curthread, zio_taskq_psrset_bind,
0, NULL, NULL) == 0) {
curthread->t_bind_pset = zio_taskq_psrset_bind;
} else {
cmn_err(CE_WARN,
"Couldn't bind process for zfs pool \"%s\" to "
"pset %d\n", spa->spa_name, zio_taskq_psrset_bind);
}
mutex_exit(&curproc->p_lock);
mutex_exit(&pidlock);
mutex_exit(&cpu_lock);
pool_unlock();
}
if (zio_taskq_sysdc) {
sysdc_thread_enter(curthread, 100, 0);
}
spa->spa_proc = curproc;
spa->spa_did = curthread->t_did;
spa_create_zio_taskqs(spa);
mutex_enter(&spa->spa_proc_lock);
ASSERT(spa->spa_proc_state == SPA_PROC_CREATED);
spa->spa_proc_state = SPA_PROC_ACTIVE;
cv_broadcast(&spa->spa_proc_cv);
CALLB_CPR_SAFE_BEGIN(&cprinfo);
while (spa->spa_proc_state == SPA_PROC_ACTIVE)
cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock);
CALLB_CPR_SAFE_END(&cprinfo, &spa->spa_proc_lock);
ASSERT(spa->spa_proc_state == SPA_PROC_DEACTIVATE);
spa->spa_proc_state = SPA_PROC_GONE;
spa->spa_proc = &p0;
cv_broadcast(&spa->spa_proc_cv);
CALLB_CPR_EXIT(&cprinfo); /* drops spa_proc_lock */
mutex_enter(&curproc->p_lock);
lwp_exit();
}
#endif
/*
* Activate an uninitialized pool.
*/
static void
spa_activate(spa_t *spa, int mode)
{
ASSERT(spa->spa_state == POOL_STATE_UNINITIALIZED);
spa->spa_state = POOL_STATE_ACTIVE;
spa->spa_mode = mode;
spa->spa_normal_class = metaslab_class_create(spa, zfs_metaslab_ops);
spa->spa_log_class = metaslab_class_create(spa, zfs_metaslab_ops);
/* Try to create a covering process */
mutex_enter(&spa->spa_proc_lock);
ASSERT(spa->spa_proc_state == SPA_PROC_NONE);
ASSERT(spa->spa_proc == &p0);
spa->spa_did = 0;
#ifdef HAVE_SPA_THREAD
/* Only create a process if we're going to be around a while. */
if (spa_create_process && strcmp(spa->spa_name, TRYIMPORT_NAME) != 0) {
if (newproc(spa_thread, (caddr_t)spa, syscid, maxclsyspri,
NULL, 0) == 0) {
spa->spa_proc_state = SPA_PROC_CREATED;
while (spa->spa_proc_state == SPA_PROC_CREATED) {
cv_wait(&spa->spa_proc_cv,
&spa->spa_proc_lock);
}
ASSERT(spa->spa_proc_state == SPA_PROC_ACTIVE);
ASSERT(spa->spa_proc != &p0);
ASSERT(spa->spa_did != 0);
} else {
#ifdef _KERNEL
cmn_err(CE_WARN,
"Couldn't create process for zfs pool \"%s\"\n",
spa->spa_name);
#endif
}
}
#endif /* HAVE_SPA_THREAD */
mutex_exit(&spa->spa_proc_lock);
/* If we didn't create a process, we need to create our taskqs. */
if (spa->spa_proc == &p0) {
spa_create_zio_taskqs(spa);
}
list_create(&spa->spa_config_dirty_list, sizeof (vdev_t),
offsetof(vdev_t, vdev_config_dirty_node));
list_create(&spa->spa_state_dirty_list, sizeof (vdev_t),
offsetof(vdev_t, vdev_state_dirty_node));
txg_list_create(&spa->spa_vdev_txg_list,
offsetof(struct vdev, vdev_txg_node));
avl_create(&spa->spa_errlist_scrub,
spa_error_entry_compare, sizeof (spa_error_entry_t),
offsetof(spa_error_entry_t, se_avl));
avl_create(&spa->spa_errlist_last,
spa_error_entry_compare, sizeof (spa_error_entry_t),
offsetof(spa_error_entry_t, se_avl));
}
/*
* Opposite of spa_activate().
*/
static void
spa_deactivate(spa_t *spa)
{
int t, q;
ASSERT(spa->spa_sync_on == B_FALSE);
ASSERT(spa->spa_dsl_pool == NULL);
ASSERT(spa->spa_root_vdev == NULL);
ASSERT(spa->spa_async_zio_root == NULL);
ASSERT(spa->spa_state != POOL_STATE_UNINITIALIZED);
txg_list_destroy(&spa->spa_vdev_txg_list);
list_destroy(&spa->spa_config_dirty_list);
list_destroy(&spa->spa_state_dirty_list);
for (t = 0; t < ZIO_TYPES; t++) {
for (q = 0; q < ZIO_TASKQ_TYPES; q++) {
if (spa->spa_zio_taskq[t][q] != NULL)
taskq_destroy(spa->spa_zio_taskq[t][q]);
spa->spa_zio_taskq[t][q] = NULL;
}
}
metaslab_class_destroy(spa->spa_normal_class);
spa->spa_normal_class = NULL;
metaslab_class_destroy(spa->spa_log_class);
spa->spa_log_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.
*/
spa_errlog_drain(spa);
avl_destroy(&spa->spa_errlist_scrub);
avl_destroy(&spa->spa_errlist_last);
spa->spa_state = POOL_STATE_UNINITIALIZED;
mutex_enter(&spa->spa_proc_lock);
if (spa->spa_proc_state != SPA_PROC_NONE) {
ASSERT(spa->spa_proc_state == SPA_PROC_ACTIVE);
spa->spa_proc_state = SPA_PROC_DEACTIVATE;
cv_broadcast(&spa->spa_proc_cv);
while (spa->spa_proc_state == SPA_PROC_DEACTIVATE) {
ASSERT(spa->spa_proc != &p0);
cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock);
}
ASSERT(spa->spa_proc_state == SPA_PROC_GONE);
spa->spa_proc_state = SPA_PROC_NONE;
}
ASSERT(spa->spa_proc == &p0);
mutex_exit(&spa->spa_proc_lock);
/*
* We want to make sure spa_thread() has actually exited the ZFS
* module, so that the module can't be unloaded out from underneath
* it.
*/
if (spa->spa_did != 0) {
thread_join(spa->spa_did);
spa->spa_did = 0;
}
}
/*
* Verify a pool configuration, and construct the vdev tree appropriately. This
* will create all the necessary vdevs in the appropriate layout, with each vdev
* in the CLOSED state. This will prep the pool before open/creation/import.
* All vdev validation is done by the vdev_alloc() routine.
*/
static int
spa_config_parse(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent,
uint_t id, int atype)
{
nvlist_t **child;
uint_t children;
int error;
int c;
if ((error = vdev_alloc(spa, vdp, nv, parent, id, atype)) != 0)
return (error);
if ((*vdp)->vdev_ops->vdev_op_leaf)
return (0);
error = nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
&child, &children);
if (error == ENOENT)
return (0);
if (error) {
vdev_free(*vdp);
*vdp = NULL;
return (EINVAL);
}
for (c = 0; c < children; c++) {
vdev_t *vd;
if ((error = spa_config_parse(spa, &vd, child[c], *vdp, c,
atype)) != 0) {
vdev_free(*vdp);
*vdp = NULL;
return (error);
}
}
ASSERT(*vdp != NULL);
return (0);
}
/*
* Opposite of spa_load().
*/
static void
spa_unload(spa_t *spa)
{
int i;
ASSERT(MUTEX_HELD(&spa_namespace_lock));
/*
* Stop async tasks.
*/
spa_async_suspend(spa);
/*
* Stop syncing.
*/
if (spa->spa_sync_on) {
txg_sync_stop(spa->spa_dsl_pool);
spa->spa_sync_on = B_FALSE;
}
/*
* Wait for any outstanding async I/O to complete.
*/
if (spa->spa_async_zio_root != NULL) {
(void) zio_wait(spa->spa_async_zio_root);
spa->spa_async_zio_root = NULL;
}
bpobj_close(&spa->spa_deferred_bpobj);
/*
* Close the dsl pool.
*/
if (spa->spa_dsl_pool) {
dsl_pool_close(spa->spa_dsl_pool);
spa->spa_dsl_pool = NULL;
spa->spa_meta_objset = NULL;
}
ddt_unload(spa);
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
/*
* Drop and purge level 2 cache
*/
spa_l2cache_drop(spa);
/*
* Close all vdevs.
*/
if (spa->spa_root_vdev)
vdev_free(spa->spa_root_vdev);
ASSERT(spa->spa_root_vdev == NULL);
for (i = 0; i < spa->spa_spares.sav_count; i++)
vdev_free(spa->spa_spares.sav_vdevs[i]);
if (spa->spa_spares.sav_vdevs) {
kmem_free(spa->spa_spares.sav_vdevs,
spa->spa_spares.sav_count * sizeof (void *));
spa->spa_spares.sav_vdevs = NULL;
}
if (spa->spa_spares.sav_config) {
nvlist_free(spa->spa_spares.sav_config);
spa->spa_spares.sav_config = NULL;
}
spa->spa_spares.sav_count = 0;
for (i = 0; i < spa->spa_l2cache.sav_count; i++)
vdev_free(spa->spa_l2cache.sav_vdevs[i]);
if (spa->spa_l2cache.sav_vdevs) {
kmem_free(spa->spa_l2cache.sav_vdevs,
spa->spa_l2cache.sav_count * sizeof (void *));
spa->spa_l2cache.sav_vdevs = NULL;
}
if (spa->spa_l2cache.sav_config) {
nvlist_free(spa->spa_l2cache.sav_config);
spa->spa_l2cache.sav_config = NULL;
}
spa->spa_l2cache.sav_count = 0;
spa->spa_async_suspended = 0;
spa_config_exit(spa, SCL_ALL, FTAG);
}
/*
* Load (or re-load) the current list of vdevs describing the active spares for
* this pool. When this is called, we have some form of basic information in
* 'spa_spares.sav_config'. We parse this into vdevs, try to open them, and
* then re-generate a more complete list including status information.
*/
static void
spa_load_spares(spa_t *spa)
{
nvlist_t **spares;
uint_t nspares;
int i;
vdev_t *vd, *tvd;
ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);
/*
* First, close and free any existing spare vdevs.
*/
for (i = 0; i < spa->spa_spares.sav_count; i++) {
vd = spa->spa_spares.sav_vdevs[i];
/* Undo the call to spa_activate() below */
if ((tvd = spa_lookup_by_guid(spa, vd->vdev_guid,
B_FALSE)) != NULL && tvd->vdev_isspare)
spa_spare_remove(tvd);
vdev_close(vd);
vdev_free(vd);
}
if (spa->spa_spares.sav_vdevs)
kmem_free(spa->spa_spares.sav_vdevs,
spa->spa_spares.sav_count * sizeof (void *));
if (spa->spa_spares.sav_config == NULL)
nspares = 0;
else
VERIFY(nvlist_lookup_nvlist_array(spa->spa_spares.sav_config,
ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0);
spa->spa_spares.sav_count = (int)nspares;
spa->spa_spares.sav_vdevs = NULL;
if (nspares == 0)
return;
/*
* Construct the array of vdevs, opening them to get status in the
* process. For each spare, there is potentially two different vdev_t
* structures associated with it: one in the list of spares (used only
* for basic validation purposes) and one in the active vdev
* configuration (if it's spared in). During this phase we open and
* validate each vdev on the spare list. If the vdev also exists in the
* active configuration, then we also mark this vdev as an active spare.
*/
spa->spa_spares.sav_vdevs = kmem_alloc(nspares * sizeof (void *),
KM_SLEEP);
for (i = 0; i < spa->spa_spares.sav_count; i++) {
VERIFY(spa_config_parse(spa, &vd, spares[i], NULL, 0,
VDEV_ALLOC_SPARE) == 0);
ASSERT(vd != NULL);
spa->spa_spares.sav_vdevs[i] = vd;
if ((tvd = spa_lookup_by_guid(spa, vd->vdev_guid,
B_FALSE)) != NULL) {
if (!tvd->vdev_isspare)
spa_spare_add(tvd);
/*
* We only mark the spare active if we were successfully
* able to load the vdev. Otherwise, importing a pool
* with a bad active spare would result in strange
* behavior, because multiple pool would think the spare
* is actively in use.
*
* There is a vulnerability here to an equally bizarre
* circumstance, where a dead active spare is later
* brought back to life (onlined or otherwise). Given
* the rarity of this scenario, and the extra complexity
* it adds, we ignore the possibility.
*/
if (!vdev_is_dead(tvd))
spa_spare_activate(tvd);
}
vd->vdev_top = vd;
vd->vdev_aux = &spa->spa_spares;
if (vdev_open(vd) != 0)
continue;
if (vdev_validate_aux(vd) == 0)
spa_spare_add(vd);
}
/*
* Recompute the stashed list of spares, with status information
* this time.
*/
VERIFY(nvlist_remove(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES,
DATA_TYPE_NVLIST_ARRAY) == 0);
spares = kmem_alloc(spa->spa_spares.sav_count * sizeof (void *),
KM_SLEEP);
for (i = 0; i < spa->spa_spares.sav_count; i++)
spares[i] = vdev_config_generate(spa,
spa->spa_spares.sav_vdevs[i], B_TRUE, VDEV_CONFIG_SPARE);
VERIFY(nvlist_add_nvlist_array(spa->spa_spares.sav_config,
ZPOOL_CONFIG_SPARES, spares, spa->spa_spares.sav_count) == 0);
for (i = 0; i < spa->spa_spares.sav_count; i++)
nvlist_free(spares[i]);
kmem_free(spares, spa->spa_spares.sav_count * sizeof (void *));
}
/*
* Load (or re-load) the current list of vdevs describing the active l2cache for
* this pool. When this is called, we have some form of basic information in
* 'spa_l2cache.sav_config'. We parse this into vdevs, try to open them, and
* then re-generate a more complete list including status information.
* Devices which are already active have their details maintained, and are
* not re-opened.
*/
static void
spa_load_l2cache(spa_t *spa)
{
nvlist_t **l2cache;
uint_t nl2cache;
int i, j, oldnvdevs;
uint64_t guid;
vdev_t *vd, **oldvdevs, **newvdevs = NULL;
spa_aux_vdev_t *sav = &spa->spa_l2cache;
ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);
if (sav->sav_config != NULL) {
VERIFY(nvlist_lookup_nvlist_array(sav->sav_config,
ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0);
newvdevs = kmem_alloc(nl2cache * sizeof (void *), KM_SLEEP);
} else {
nl2cache = 0;
}
oldvdevs = sav->sav_vdevs;
oldnvdevs = sav->sav_count;
sav->sav_vdevs = NULL;
sav->sav_count = 0;
/*
* Process new nvlist of vdevs.
*/
for (i = 0; i < nl2cache; i++) {
VERIFY(nvlist_lookup_uint64(l2cache[i], ZPOOL_CONFIG_GUID,
&guid) == 0);
newvdevs[i] = NULL;
for (j = 0; j < oldnvdevs; j++) {
vd = oldvdevs[j];
if (vd != NULL && guid == vd->vdev_guid) {
/*
* Retain previous vdev for add/remove ops.
*/
newvdevs[i] = vd;
oldvdevs[j] = NULL;
break;
}
}
if (newvdevs[i] == NULL) {
/*
* Create new vdev
*/
VERIFY(spa_config_parse(spa, &vd, l2cache[i], NULL, 0,
VDEV_ALLOC_L2CACHE) == 0);
ASSERT(vd != NULL);
newvdevs[i] = vd;
/*
* Commit this vdev as an l2cache device,
* even if it fails to open.
*/
spa_l2cache_add(vd);
vd->vdev_top = vd;
vd->vdev_aux = sav;
spa_l2cache_activate(vd);
if (vdev_open(vd) != 0)
continue;
(void) vdev_validate_aux(vd);
if (!vdev_is_dead(vd))
l2arc_add_vdev(spa, vd);
}
}
/*
* Purge vdevs that were dropped
*/
for (i = 0; i < oldnvdevs; i++) {
uint64_t pool;
vd = oldvdevs[i];
if (vd != NULL) {
if (spa_l2cache_exists(vd->vdev_guid, &pool) &&
pool != 0ULL && l2arc_vdev_present(vd))
l2arc_remove_vdev(vd);
(void) vdev_close(vd);
spa_l2cache_remove(vd);
}
}
if (oldvdevs)
kmem_free(oldvdevs, oldnvdevs * sizeof (void *));
if (sav->sav_config == NULL)
goto out;
sav->sav_vdevs = newvdevs;
sav->sav_count = (int)nl2cache;
/*
* Recompute the stashed list of l2cache devices, with status
* information this time.
*/
VERIFY(nvlist_remove(sav->sav_config, ZPOOL_CONFIG_L2CACHE,
DATA_TYPE_NVLIST_ARRAY) == 0);
l2cache = kmem_alloc(sav->sav_count * sizeof (void *), KM_SLEEP);
for (i = 0; i < sav->sav_count; i++)
l2cache[i] = vdev_config_generate(spa,
sav->sav_vdevs[i], B_TRUE, VDEV_CONFIG_L2CACHE);
VERIFY(nvlist_add_nvlist_array(sav->sav_config,
ZPOOL_CONFIG_L2CACHE, l2cache, sav->sav_count) == 0);
out:
for (i = 0; i < sav->sav_count; i++)
nvlist_free(l2cache[i]);
if (sav->sav_count)
kmem_free(l2cache, sav->sav_count * sizeof (void *));
}
static int
load_nvlist(spa_t *spa, uint64_t obj, nvlist_t **value)
{
dmu_buf_t *db;
char *packed = NULL;
size_t nvsize = 0;
int error;
*value = NULL;
VERIFY(0 == dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db));
nvsize = *(uint64_t *)db->db_data;
dmu_buf_rele(db, FTAG);
packed = kmem_alloc(nvsize, KM_SLEEP);
error = dmu_read(spa->spa_meta_objset, obj, 0, nvsize, packed,
DMU_READ_PREFETCH);
if (error == 0)
error = nvlist_unpack(packed, nvsize, value, 0);
kmem_free(packed, nvsize);
return (error);
}
/*
* 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)
{
int c;
for (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)) {
zfs_ereport_post(FM_EREPORT_RESOURCE_AUTOREPLACE,
vd->vdev_spa, vd, NULL, 0, 0);
spa_event_notify(vd->vdev_spa, vd, FM_EREPORT_ZFS_DEVICE_CHECK);
}
}
/*
* Load the slog device state from the config object since it's possible
* that the label does not contain the most up-to-date information.
*/
void
spa_load_log_state(spa_t *spa, nvlist_t *nv)
{
vdev_t *ovd, *rvd = spa->spa_root_vdev;
int c;
/*
* Load the original root vdev tree from the passed config.
*/
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
VERIFY(spa_config_parse(spa, &ovd, nv, NULL, 0, VDEV_ALLOC_LOAD) == 0);
for (c = 0; c < rvd->vdev_children; c++) {
vdev_t *cvd = rvd->vdev_child[c];
if (cvd->vdev_islog)
vdev_load_log_state(cvd, ovd->vdev_child[c]);
}
vdev_free(ovd);
spa_config_exit(spa, SCL_ALL, FTAG);
}
/*
* Check for missing log devices
*/
int
spa_check_logs(spa_t *spa)
{
switch (spa->spa_log_state) {
default:
break;
case SPA_LOG_MISSING:
/* need to recheck in case slog has been restored */
case SPA_LOG_UNKNOWN:
if (dmu_objset_find(spa->spa_name, zil_check_log_chain, NULL,
DS_FIND_CHILDREN)) {
spa_set_log_state(spa, SPA_LOG_MISSING);
return (1);
}
break;
}
return (0);
}
static boolean_t
spa_passivate_log(spa_t *spa)
{
vdev_t *rvd = spa->spa_root_vdev;
boolean_t slog_found = B_FALSE;
int c;
ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER));
if (!spa_has_slogs(spa))
return (B_FALSE);
for (c = 0; c < rvd->vdev_children; c++) {
vdev_t *tvd = rvd->vdev_child[c];
metaslab_group_t *mg = tvd->vdev_mg;
if (tvd->vdev_islog) {
metaslab_group_passivate(mg);
slog_found = B_TRUE;
}
}
return (slog_found);
}
static void
spa_activate_log(spa_t *spa)
{
vdev_t *rvd = spa->spa_root_vdev;
int c;
ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER));
for (c = 0; c < rvd->vdev_children; c++) {
vdev_t *tvd = rvd->vdev_child[c];
metaslab_group_t *mg = tvd->vdev_mg;
if (tvd->vdev_islog)
metaslab_group_activate(mg);
}
}
int
spa_offline_log(spa_t *spa)
{
int error = 0;
if ((error = dmu_objset_find(spa_name(spa), zil_vdev_offline,
NULL, DS_FIND_CHILDREN)) == 0) {
/*
* We successfully offlined the log device, sync out the
* current txg so that the "stubby" block can be removed
* by zil_sync().
*/
txg_wait_synced(spa->spa_dsl_pool, 0);
}
return (error);
}
static void
spa_aux_check_removed(spa_aux_vdev_t *sav)
{
int i;
for (i = 0; i < sav->sav_count; i++)
spa_check_removed(sav->sav_vdevs[i]);
}
void
spa_claim_notify(zio_t *zio)
{
spa_t *spa = zio->io_spa;
if (zio->io_error)
return;
mutex_enter(&spa->spa_props_lock); /* any mutex will do */
if (spa->spa_claim_max_txg < zio->io_bp->blk_birth)
spa->spa_claim_max_txg = zio->io_bp->blk_birth;
mutex_exit(&spa->spa_props_lock);
}
typedef struct spa_load_error {
uint64_t sle_meta_count;
uint64_t sle_data_count;
} spa_load_error_t;
static void
spa_load_verify_done(zio_t *zio)
{
blkptr_t *bp = zio->io_bp;
spa_load_error_t *sle = zio->io_private;
dmu_object_type_t type = BP_GET_TYPE(bp);
int error = zio->io_error;
if (error) {
if ((BP_GET_LEVEL(bp) != 0 || dmu_ot[type].ot_metadata) &&
type != DMU_OT_INTENT_LOG)
atomic_add_64(&sle->sle_meta_count, 1);
else
atomic_add_64(&sle->sle_data_count, 1);
}
zio_data_buf_free(zio->io_data, zio->io_size);
}
/*ARGSUSED*/
static int
spa_load_verify_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
arc_buf_t *pbuf, const zbookmark_t *zb, const dnode_phys_t *dnp, void *arg)
{
if (bp != NULL) {
zio_t *rio = arg;
size_t size = BP_GET_PSIZE(bp);
void *data = zio_data_buf_alloc(size);
zio_nowait(zio_read(rio, spa, bp, data, size,
spa_load_verify_done, rio->io_private, ZIO_PRIORITY_SCRUB,
ZIO_FLAG_SPECULATIVE | ZIO_FLAG_CANFAIL |
ZIO_FLAG_SCRUB | ZIO_FLAG_RAW, zb));
}
return (0);
}
static int
spa_load_verify(spa_t *spa)
{
zio_t *rio;
spa_load_error_t sle = { 0 };
zpool_rewind_policy_t policy;
boolean_t verify_ok = B_FALSE;
int error;
zpool_get_rewind_policy(spa->spa_config, &policy);
if (policy.zrp_request & ZPOOL_NEVER_REWIND)
return (0);
rio = zio_root(spa, NULL, &sle,
ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE);
error = traverse_pool(spa, spa->spa_verify_min_txg,
TRAVERSE_PRE | TRAVERSE_PREFETCH, spa_load_verify_cb, rio);
(void) zio_wait(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) {
verify_ok = B_TRUE;
spa->spa_load_txg = spa->spa_uberblock.ub_txg;
spa->spa_load_txg_ts = spa->spa_uberblock.ub_timestamp;
} else {
spa->spa_load_max_txg = spa->spa_uberblock.ub_txg;
}
if (error) {
if (error != ENXIO && error != EIO)
error = EIO;
return (error);
}
return (verify_ok ? 0 : EIO);
}
/*
* Find a value in the pool props object.
*/
static void
spa_prop_find(spa_t *spa, zpool_prop_t prop, uint64_t *val)
{
(void) zap_lookup(spa->spa_meta_objset, spa->spa_pool_props_object,
zpool_prop_to_name(prop), sizeof (uint64_t), 1, val);
}
/*
* Find a value in the pool directory object.
*/
static int
spa_dir_prop(spa_t *spa, const char *name, uint64_t *val)
{
return (zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
name, sizeof (uint64_t), 1, val));
}
static int
spa_vdev_err(vdev_t *vdev, vdev_aux_t aux, int err)
{
vdev_set_state(vdev, B_TRUE, VDEV_STATE_CANT_OPEN, aux);
return (err);
}
/*
* 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
* pool have that entry in their config, but only the splitting one contains
* a list of all the guids of the vdevs that are being split off.
*
* This function determines what to do with that list: either rejoin
* all the disks to the pool, or complete the splitting process. To attempt
* the rejoin, each disk that is offlined is marked online again, and
* we do a reopen() call. If the vdev label for every disk that was
* marked online indicates it was successfully split off (VDEV_AUX_SPLIT_POOL)
* then we call vdev_split() on each disk, and complete the split.
*
* Otherwise we leave the config alone, with all the vdevs in place in
* the original pool.
*/
static void
spa_try_repair(spa_t *spa, nvlist_t *config)
{
uint_t extracted;
uint64_t *glist;
uint_t i, gcount;
nvlist_t *nvl;
vdev_t **vd;
boolean_t attempt_reopen;
if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT, &nvl) != 0)
return;
/* check that the config is complete */
if (nvlist_lookup_uint64_array(nvl, ZPOOL_CONFIG_SPLIT_LIST,
&glist, &gcount) != 0)
return;
vd = kmem_zalloc(gcount * sizeof (vdev_t *), KM_SLEEP);
/* attempt to online all the vdevs & validate */
attempt_reopen = B_TRUE;
for (i = 0; i < gcount; i++) {
if (glist[i] == 0) /* vdev is hole */
continue;
vd[i] = spa_lookup_by_guid(spa, glist[i], B_FALSE);
if (vd[i] == NULL) {
/*
* Don't bother attempting to reopen the disks;
* just do the split.
*/
attempt_reopen = B_FALSE;
} else {
/* attempt to re-online it */
vd[i]->vdev_offline = B_FALSE;
}
}
if (attempt_reopen) {
vdev_reopen(spa->spa_root_vdev);
/* check each device to see what state it's in */
for (extracted = 0, i = 0; i < gcount; i++) {
if (vd[i] != NULL &&
vd[i]->vdev_stat.vs_aux != VDEV_AUX_SPLIT_POOL)
break;
++extracted;
}
}
/*
* If every disk has been moved to the new pool, or if we never
* even attempted to look at them, then we split them off for
* good.
*/
if (!attempt_reopen || gcount == extracted) {
for (i = 0; i < gcount; i++)
if (vd[i] != NULL)
vdev_split(vd[i]);
vdev_reopen(spa->spa_root_vdev);
}
kmem_free(vd, gcount * sizeof (vdev_t *));
}
static int
spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type,
boolean_t mosconfig)
{
nvlist_t *config = spa->spa_config;
char *ereport = FM_EREPORT_ZFS_POOL;
int error;
uint64_t pool_guid;
nvlist_t *nvl;
if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &pool_guid))
return (EINVAL);
/*
* 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 = EEXIST;
} else {
spa->spa_load_guid = pool_guid;
if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT,
&nvl) == 0) {
VERIFY(nvlist_dup(nvl, &spa->spa_config_splitting,
KM_SLEEP) == 0);
}
error = spa_load_impl(spa, pool_guid, config, state, type,
mosconfig, &ereport);
}
spa->spa_minref = refcount_count(&spa->spa_refcount);
if (error && error != EBADF)
zfs_ereport_post(ereport, spa, NULL, NULL, 0, 0);
spa->spa_load_state = error ? SPA_LOAD_ERROR : SPA_LOAD_NONE;
spa->spa_ena = 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 mosconfig,
char **ereport)
{
int error = 0;
nvlist_t *nvroot = NULL;
vdev_t *rvd;
uberblock_t *ub = &spa->spa_uberblock;
uint64_t config_cache_txg = spa->spa_config_txg;
int orig_mode = spa->spa_mode;
int parse;
uint64_t obj;
/*
* If this is an untrusted config, access the pool in read-only mode.
* This prevents things like resilvering recently removed devices.
*/
if (!mosconfig)
spa->spa_mode = FREAD;
ASSERT(MUTEX_HELD(&spa_namespace_lock));
spa->spa_load_state = state;
if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot))
return (EINVAL);
parse = (type == SPA_IMPORT_EXISTING ?
VDEV_ALLOC_LOAD : VDEV_ALLOC_SPLIT);
/*
* Create "The Godfather" zio to hold all async IOs
*/
spa->spa_async_zio_root = zio_root(spa, NULL, NULL,
ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_GODFATHER);
/*
* Parse the configuration into a vdev tree. We explicitly set the
* value that will be returned by spa_version() since parsing the
* 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);
spa_config_exit(spa, SCL_ALL, FTAG);
if (error != 0)
return (error);
ASSERT(spa->spa_root_vdev == rvd);
if (type != SPA_IMPORT_ASSEMBLE) {
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);
/*
* 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.
*/
if (type != SPA_IMPORT_ASSEMBLE) {
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
error = vdev_validate(rvd);
spa_config_exit(spa, SCL_ALL, FTAG);
if (error != 0)
return (error);
if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN)
return (ENXIO);
}
/*
* Find the best uberblock.
*/
vdev_uberblock_load(NULL, rvd, ub);
/*
* If we weren't able to find a single valid uberblock, return failure.
*/
if (ub->ub_txg == 0)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, ENXIO));
/*
* If the pool is newer than the code, we can't open it.
*/
if (ub->ub_version > SPA_VERSION)
return (spa_vdev_err(rvd, VDEV_AUX_VERSION_NEWER, ENOTSUP));
/*
* If the vdev guid sum doesn't match the uberblock, we have an
* incomplete configuration.
*/
if (mosconfig && 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_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;
error = dsl_pool_open(spa, spa->spa_first_txg, &spa->spa_dsl_pool);
if (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 (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
if (!mosconfig) {
uint64_t hostid;
nvlist_t *policy = NULL, *nvconfig;
if (load_nvlist(spa, spa->spa_config_object, &nvconfig) != 0)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
if (!spa_is_root(spa) && nvlist_lookup_uint64(nvconfig,
ZPOOL_CONFIG_HOSTID, &hostid) == 0) {
char *hostname;
unsigned long myhostid = 0;
VERIFY(nvlist_lookup_string(nvconfig,
ZPOOL_CONFIG_HOSTNAME, &hostname) == 0);
#ifdef _KERNEL
myhostid = zone_get_hostid(NULL);
#else /* _KERNEL */
/*
* We're emulating the system's hostid in userland, so
* we can't use zone_get_hostid().
*/
(void) ddi_strtoul(hw_serial, NULL, 10, &myhostid);
#endif /* _KERNEL */
if (hostid != 0 && myhostid != 0 &&
hostid != myhostid) {
nvlist_free(nvconfig);
cmn_err(CE_WARN, "pool '%s' could not be "
"loaded as it was last accessed by "
"another system (host: %s hostid: 0x%lx). "
"See: http://www.sun.com/msg/ZFS-8000-EY",
spa_name(spa), hostname,
(unsigned long)hostid);
return (EBADF);
}
}
if (nvlist_lookup_nvlist(spa->spa_config,
ZPOOL_REWIND_POLICY, &policy) == 0)
VERIFY(nvlist_add_nvlist(nvconfig,
ZPOOL_REWIND_POLICY, policy) == 0);
spa_config_set(spa, nvconfig);
spa_unload(spa);
spa_deactivate(spa);
spa_activate(spa, orig_mode);
return (spa_load(spa, state, SPA_IMPORT_EXISTING, B_TRUE));
}
if (spa_dir_prop(spa, DMU_POOL_SYNC_BPOBJ, &obj) != 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)
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);
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);
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);
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);
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);
if (error != 0 && error != ENOENT)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
/*
* 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);
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));
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);
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));
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;
}
spa->spa_delegation = zpool_prop_default_numeric(ZPOOL_PROP_DELEGATION);
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));
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_DEDUPDITTO,
&spa->spa_dedup_ditto);
spa->spa_autoreplace = (autoreplace != 0);
}
/*
* 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.
*/
if (state != SPA_LOAD_IMPORT) {
spa_aux_check_removed(&spa->spa_spares);
spa_aux_check_removed(&spa->spa_l2cache);
}
}
/*
* Load the vdev state for all toplevel vdevs.
*/
vdev_load(rvd);
/*
* Propagate the leaf DTLs we just loaded all the way up the 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);
/*
* Check the state of the root vdev. If it can't be opened, it
* indicates one or more toplevel vdevs are faulted.
*/
if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN)
return (ENXIO);
/*
* Load the DDTs (dedup tables).
*/
error = ddt_load(spa);
if (error != 0)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
spa_update_dspace(spa);
if (state != SPA_LOAD_TRYIMPORT) {
error = spa_load_verify(spa);
if (error)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA,
error));
}
/*
* Load the intent log state and check log integrity. If we're
* assembling a pool from a split, the log is not transferred over.
*/
if (type != SPA_IMPORT_ASSEMBLE) {
nvlist_t *nvconfig;
if (load_nvlist(spa, spa->spa_config_object, &nvconfig) != 0)
return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
VERIFY(nvlist_lookup_nvlist(nvconfig, ZPOOL_CONFIG_VDEV_TREE,
&nvroot) == 0);
spa_load_log_state(spa, nvroot);
nvlist_free(nvconfig);
if (spa_check_logs(spa)) {
*ereport = FM_EREPORT_ZFS_LOG_REPLAY;
return (spa_vdev_err(rvd, VDEV_AUX_BAD_LOG, ENXIO));
}
}
if (spa_writeable(spa) && (state == SPA_LOAD_RECOVER ||
spa->spa_load_max_txg == UINT64_MAX)) {
dmu_tx_t *tx;
int need_update = B_FALSE;
int c;
ASSERT(state != SPA_LOAD_TRYIMPORT);
/*
* 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(spa_get_dsl(spa),
spa_first_txg(spa));
(void) dmu_objset_find(spa_name(spa),
zil_claim, tx, DS_FIND_CHILDREN);
dmu_tx_commit(tx);
spa->spa_claiming = B_FALSE;
spa_set_log_state(spa, SPA_LOG_GOOD);
spa->spa_sync_on = B_TRUE;
txg_sync_start(spa->spa_dsl_pool);
/*
* 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.
*/
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 spa_load_verbatim is true, trust the current
* in-core spa_config and update the disk labels.
*/
if (config_cache_txg != spa->spa_config_txg ||
state == SPA_LOAD_IMPORT || spa->spa_load_verbatim ||
state == SPA_LOAD_RECOVER)
need_update = B_TRUE;
for (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);
/*
* Check all DTLs to see if anything needs resilvering.
*/
if (!dsl_scan_resilvering(spa->spa_dsl_pool) &&
vdev_resilver_needed(rvd, NULL, NULL))
spa_async_request(spa, SPA_ASYNC_RESILVER);
/*
* Delete any inconsistent datasets.
*/
(void) dmu_objset_find(spa_name(spa),
dsl_destroy_inconsistent, NULL, DS_FIND_CHILDREN);
/*
* Clean up any stale temporary dataset userrefs.
*/
dsl_pool_clean_tmp_userrefs(spa->spa_dsl_pool);
}
return (0);
}
static int
spa_load_retry(spa_t *spa, spa_load_state_t state, int mosconfig)
{
spa_unload(spa);
spa_deactivate(spa);
spa->spa_load_max_txg--;
spa_activate(spa, spa_mode_global);
spa_async_suspend(spa);
return (spa_load(spa, state, SPA_IMPORT_EXISTING, mosconfig));
}
static int
spa_load_best(spa_t *spa, spa_load_state_t state, int mosconfig,
uint64_t max_request, int rewind_flags)
{
nvlist_t *config = NULL;
int load_error, rewind_error;
uint64_t safe_rewind_txg;
uint64_t min_txg;
if (spa->spa_load_txg && state == SPA_LOAD_RECOVER) {
spa->spa_load_max_txg = spa->spa_load_txg;
spa_set_log_state(spa, SPA_LOG_CLEAR);
} else {
spa->spa_load_max_txg = max_request;
}
load_error = rewind_error = spa_load(spa, state, SPA_IMPORT_EXISTING,
mosconfig);
if (load_error == 0)
return (0);
if (spa->spa_root_vdev != NULL)
config = spa_config_generate(spa, NULL, -1ULL, B_TRUE);
spa->spa_last_ubsync_txg = spa->spa_uberblock.ub_txg;
spa->spa_last_ubsync_txg_ts = spa->spa_uberblock.ub_timestamp;
if (rewind_flags & ZPOOL_NEVER_REWIND) {
nvlist_free(config);
return (load_error);
}
/* Price of rolling back is discarding txgs, including log */
if (state == SPA_LOAD_RECOVER)
spa_set_log_state(spa, SPA_LOG_CLEAR);
spa->spa_load_max_txg = spa->spa_last_ubsync_txg;
safe_rewind_txg = spa->spa_last_ubsync_txg - TXG_DEFER_SIZE;
min_txg = (rewind_flags & ZPOOL_EXTREME_REWIND) ?
TXG_INITIAL : safe_rewind_txg;
/*
* Continue as long as we're finding errors, we're still within
* the acceptable rewind range, and we're still finding uberblocks
*/
while (rewind_error && spa->spa_uberblock.ub_txg >= min_txg &&
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);
}
if (config)
spa_rewind_data_to_nvlist(spa, config);
spa->spa_extreme_rewind = B_FALSE;
spa->spa_load_max_txg = UINT64_MAX;
if (config && (rewind_error || state != SPA_LOAD_RECOVER))
spa_config_set(spa, config);
return (state == SPA_LOAD_RECOVER ? rewind_error : load_error);
}
/*
* Pool Open/Import
*
* The import case is identical to an open except that the configuration is sent
* down from userland, instead of grabbed from the configuration cache. For the
* case of an open, the pool configuration will exist in the
* POOL_STATE_UNINITIALIZED state.
*
* The stats information (gen/count/ustats) is used to gather vdev statistics at
* the same time open the pool, without having to keep around the spa_t in some
* ambiguous state.
*/
static int
spa_open_common(const char *pool, spa_t **spapp, void *tag, nvlist_t *nvpolicy,
nvlist_t **config)
{
spa_t *spa;
int error;
int locked = B_FALSE;
*spapp = NULL;
/*
* As disgusting as this is, we need to support recursive calls to this
* function because dsl_dir_open() is called during spa_load(), and ends
* up calling spa_open() again. The real fix is to figure out how to
* avoid dsl_dir_open() calling this in the first place.
*/
if (mutex_owner(&spa_namespace_lock) != curthread) {
mutex_enter(&spa_namespace_lock);
locked = B_TRUE;
}
if ((spa = spa_lookup(pool)) == NULL) {
if (locked)
mutex_exit(&spa_namespace_lock);
return (ENOENT);
}
if (spa->spa_state == POOL_STATE_UNINITIALIZED) {
spa_load_state_t state = SPA_LOAD_OPEN;
zpool_rewind_policy_t policy;
zpool_get_rewind_policy(nvpolicy ? nvpolicy : spa->spa_config,
&policy);
if (policy.zrp_request & 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;
error = spa_load_best(spa, state, B_FALSE, policy.zrp_txg,
policy.zrp_request);
if (error == EBADF) {
/*
* If vdev_validate() returns failure (indicated by
* EBADF), it indicates that one of the vdevs indicates
* that the pool has been exported or destroyed. If
* this is the case, the config cache is out of sync and
* we should remove the pool from the namespace.
*/
spa_unload(spa);
spa_deactivate(spa);
spa_config_sync(spa, B_TRUE, B_TRUE);
spa_remove(spa);
if (locked)
mutex_exit(&spa_namespace_lock);
return (ENOENT);
}
if (error) {
/*
* We can't open the pool, but we still have useful
* information: the state of each vdev after the
* attempted vdev_open(). Return this to the user.
*/
if (config != NULL && spa->spa_config)
VERIFY(nvlist_dup(spa->spa_config, config,
KM_SLEEP) == 0);
spa_unload(spa);
spa_deactivate(spa);
spa->spa_last_open_failed = error;
if (locked)
mutex_exit(&spa_namespace_lock);
*spapp = NULL;
return (error);
}
}
spa_open_ref(spa, tag);
if (config != NULL)
*config = spa_config_generate(spa, NULL, -1ULL, B_TRUE);
if (locked) {
spa->spa_last_open_failed = 0;
spa->spa_last_ubsync_txg = 0;
spa->spa_load_txg = 0;
mutex_exit(&spa_namespace_lock);
}
*spapp = spa;
return (0);
}
int
spa_open_rewind(const char *name, spa_t **spapp, void *tag, nvlist_t *policy,
nvlist_t **config)
{
return (spa_open_common(name, spapp, tag, policy, config));
}
int
spa_open(const char *name, spa_t **spapp, void *tag)
{
return (spa_open_common(name, spapp, tag, NULL, NULL));
}
/*
* Lookup the given spa_t, incrementing the inject count in the process,
* preventing it from being exported or destroyed.
*/
spa_t *
spa_inject_addref(char *name)
{
spa_t *spa;
mutex_enter(&spa_namespace_lock);
if ((spa = spa_lookup(name)) == NULL) {
mutex_exit(&spa_namespace_lock);
return (NULL);
}
spa->spa_inject_ref++;
mutex_exit(&spa_namespace_lock);
return (spa);
}
void
spa_inject_delref(spa_t *spa)
{
mutex_enter(&spa_namespace_lock);
spa->spa_inject_ref--;
mutex_exit(&spa_namespace_lock);
}
/*
* Add spares device information to the nvlist.
*/
static void
spa_add_spares(spa_t *spa, nvlist_t *config)
{
nvlist_t **spares;
uint_t i, nspares;
nvlist_t *nvroot;
uint64_t guid;
vdev_stat_t *vs;
uint_t vsc;
uint64_t pool;
ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER));
if (spa->spa_spares.sav_count == 0)
return;
VERIFY(nvlist_lookup_nvlist(config,
ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0);
VERIFY(nvlist_lookup_nvlist_array(spa->spa_spares.sav_config,
ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0);
if (nspares != 0) {
VERIFY(nvlist_add_nvlist_array(nvroot,
ZPOOL_CONFIG_SPARES, spares, nspares) == 0);
VERIFY(nvlist_lookup_nvlist_array(nvroot,
ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0);
/*
* Go through and find any spares which have since been
* repurposed as an active spare. If this is the case, update
* their status appropriately.
*/
for (i = 0; i < nspares; i++) {
VERIFY(nvlist_lookup_uint64(spares[i],
ZPOOL_CONFIG_GUID, &guid) == 0);
if (spa_spare_exists(guid, &pool, NULL) &&
pool != 0ULL) {
VERIFY(nvlist_lookup_uint64_array(
spares[i], ZPOOL_CONFIG_VDEV_STATS,
(uint64_t **)&vs, &vsc) == 0);
vs->vs_state = VDEV_STATE_CANT_OPEN;
vs->vs_aux = VDEV_AUX_SPARED;
}
}
}
}
/*
* Add l2cache device information to the nvlist, including vdev stats.
*/
static void
spa_add_l2cache(spa_t *spa, nvlist_t *config)
{
nvlist_t **l2cache;
uint_t i, j, nl2cache;
nvlist_t *nvroot;
uint64_t guid;
vdev_t *vd;
vdev_stat_t *vs;
uint_t vsc;
ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER));
if (spa->spa_l2cache.sav_count == 0)
return;
VERIFY(nvlist_lookup_nvlist(config,
ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0);
VERIFY(nvlist_lookup_nvlist_array(spa->spa_l2cache.sav_config,
ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0);
if (nl2cache != 0) {
VERIFY(nvlist_add_nvlist_array(nvroot,
ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache) == 0);
VERIFY(nvlist_lookup_nvlist_array(nvroot,
ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0);
/*
* Update level 2 cache device stats.
*/
for (i = 0; i < nl2cache; i++) {
VERIFY(nvlist_lookup_uint64(l2cache[i],
ZPOOL_CONFIG_GUID, &guid) == 0);
vd = NULL;
for (j = 0; j < spa->spa_l2cache.sav_count; j++) {
if (guid ==
spa->spa_l2cache.sav_vdevs[j]->vdev_guid) {
vd = spa->spa_l2cache.sav_vdevs[j];
break;
}
}
ASSERT(vd != NULL);
VERIFY(nvlist_lookup_uint64_array(l2cache[i],
ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&vs, &vsc)
== 0);
vdev_get_stats(vd, vs);
}
}
}
int
spa_get_stats(const char *name, nvlist_t **config, char *altroot, size_t buflen)
{
int error;
spa_t *spa;
*config = NULL;
error = spa_open_common(name, &spa, FTAG, NULL, config);
if (spa != NULL) {
/*
* This still leaves a window of inconsistency where the spares
* or l2cache devices could change and the config would be
* self-inconsistent.
*/
spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
if (*config != NULL) {
VERIFY(nvlist_add_uint64(*config,
ZPOOL_CONFIG_ERRCOUNT,
spa_get_errlog_size(spa)) == 0);
if (spa_suspended(spa))
VERIFY(nvlist_add_uint64(*config,
ZPOOL_CONFIG_SUSPENDED,
spa->spa_failmode) == 0);
spa_add_spares(spa, *config);
spa_add_l2cache(spa, *config);
}
}
/*
* We want to get the alternate root even for faulted pools, so we cheat
* and call spa_lookup() directly.
*/
if (altroot) {
if (spa == NULL) {
mutex_enter(&spa_namespace_lock);
spa = spa_lookup(name);
if (spa)
spa_altroot(spa, altroot, buflen);
else
altroot[0] = '\0';
spa = NULL;
mutex_exit(&spa_namespace_lock);
} else {
spa_altroot(spa, altroot, buflen);
}
}
if (spa != NULL) {
spa_config_exit(spa, SCL_CONFIG, FTAG);
spa_close(spa, FTAG);
}
return (error);
}
/*
* Validate that the auxiliary device array is well formed. We must have an
* array of nvlists, each which describes a valid leaf vdev. If this is an
* import (mode is VDEV_ALLOC_SPARE), then we allow corrupted spares to be
* specified, as long as they are well-formed.
*/
static int
spa_validate_aux_devs(spa_t *spa, nvlist_t *nvroot, uint64_t crtxg, int mode,
spa_aux_vdev_t *sav, const char *config, uint64_t version,
vdev_labeltype_t label)
{
nvlist_t **dev;
uint_t i, ndev;
vdev_t *vd;
int error;
ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);
/*
* It's acceptable to have no devs specified.
*/
if (nvlist_lookup_nvlist_array(nvroot, config, &dev, &ndev) != 0)
return (0);
if (ndev == 0)
return (EINVAL);
/*
* Make sure the pool is formatted with a version that supports this
* device type.
*/
if (spa_version(spa) < version)
return (ENOTSUP);
/*
* Set the pending device list so we correctly handle device in-use
* checking.
*/
sav->sav_pending = dev;
sav->sav_npending = ndev;
for (i = 0; i < ndev; i++) {
if ((error = spa_config_parse(spa, &vd, dev[i], NULL, 0,
mode)) != 0)
goto out;
if (!vd->vdev_ops->vdev_op_leaf) {
vdev_free(vd);
error = EINVAL;
goto out;
}
/*
* The L2ARC currently only supports disk devices in
* kernel context. For user-level testing, we allow it.
*/
#ifdef _KERNEL
if ((strcmp(config, ZPOOL_CONFIG_L2CACHE) == 0) &&
strcmp(vd->vdev_ops->vdev_op_type, VDEV_TYPE_DISK) != 0) {
error = ENOTBLK;
goto out;
}
#endif
vd->vdev_top = vd;
if ((error = vdev_open(vd)) == 0 &&
(error = vdev_label_init(vd, crtxg, label)) == 0) {
VERIFY(nvlist_add_uint64(dev[i], ZPOOL_CONFIG_GUID,
vd->vdev_guid) == 0);
}
vdev_free(vd);
if (error &&
(mode != VDEV_ALLOC_SPARE && mode != VDEV_ALLOC_L2CACHE))
goto out;
else
error = 0;
}
out:
sav->sav_pending = NULL;
sav->sav_npending = 0;
return (error);
}
static int
spa_validate_aux(spa_t *spa, nvlist_t *nvroot, uint64_t crtxg, int mode)
{
int error;
ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);
if ((error = spa_validate_aux_devs(spa, nvroot, crtxg, mode,
&spa->spa_spares, ZPOOL_CONFIG_SPARES, SPA_VERSION_SPARES,
VDEV_LABEL_SPARE)) != 0) {
return (error);
}
return (spa_validate_aux_devs(spa, nvroot, crtxg, mode,
&spa->spa_l2cache, ZPOOL_CONFIG_L2CACHE, SPA_VERSION_L2CACHE,
VDEV_LABEL_L2CACHE));
}
static void
spa_set_aux_vdevs(spa_aux_vdev_t *sav, nvlist_t **devs, int ndevs,
const char *config)
{
int i;
if (sav->sav_config != NULL) {
nvlist_t **olddevs;
uint_t oldndevs;
nvlist_t **newdevs;
/*
* Generate new dev list by concatentating with the
* current dev list.
*/
VERIFY(nvlist_lookup_nvlist_array(sav->sav_config, config,
&olddevs, &oldndevs) == 0);
newdevs = kmem_alloc(sizeof (void *) *
(ndevs + oldndevs), KM_SLEEP);
for (i = 0; i < oldndevs; i++)
VERIFY(nvlist_dup(olddevs[i], &newdevs[i],
KM_SLEEP) == 0);
for (i = 0; i < ndevs; i++)
VERIFY(nvlist_dup(devs[i], &newdevs[i + oldndevs],
KM_SLEEP) == 0);
VERIFY(nvlist_remove(sav->sav_config, config,
DATA_TYPE_NVLIST_ARRAY) == 0);
VERIFY(nvlist_add_nvlist_array(sav->sav_config,
config, newdevs, ndevs + oldndevs) == 0);
for (i = 0; i < oldndevs + ndevs; i++)
nvlist_free(newdevs[i]);
kmem_free(newdevs, (oldndevs + ndevs) * sizeof (void *));
} else {
/*
* Generate a new dev list.
*/
VERIFY(nvlist_alloc(&sav->sav_config, NV_UNIQUE_NAME,
KM_SLEEP) == 0);
VERIFY(nvlist_add_nvlist_array(sav->sav_config, config,
devs, ndevs) == 0);
}
}
/*
* Stop and drop level 2 ARC devices
*/
void
spa_l2cache_drop(spa_t *spa)
{
vdev_t *vd;
int i;
spa_aux_vdev_t *sav = &spa->spa_l2cache;
for (i = 0; i < sav->sav_count; i++) {
uint64_t pool;
vd = sav->sav_vdevs[i];
ASSERT(vd != NULL);
if (spa_l2cache_exists(vd->vdev_guid, &pool) &&
pool != 0ULL && l2arc_vdev_present(vd))
l2arc_remove_vdev(vd);
if (vd->vdev_isl2cache)
spa_l2cache_remove(vd);
vdev_clear_stats(vd);
(void) vdev_close(vd);
}
}
/*
* Pool Creation
*/
int
spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
const char *history_str, nvlist_t *zplprops)
{
spa_t *spa;
char *altroot = NULL;
vdev_t *rvd;
dsl_pool_t *dp;
dmu_tx_t *tx;
int error = 0;
uint64_t txg = TXG_INITIAL;
nvlist_t **spares, **l2cache;
uint_t nspares, nl2cache;
uint64_t version, obj;
int c;
/*
* If this pool already exists, return failure.
*/
mutex_enter(&spa_namespace_lock);
if (spa_lookup(pool) != NULL) {
mutex_exit(&spa_namespace_lock);
return (EEXIST);
}
/*
* Allocate a new spa_t structure.
*/
(void) nvlist_lookup_string(props,
zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot);
spa = spa_add(pool, NULL, altroot);
spa_activate(spa, spa_mode_global);
if (props && (error = spa_prop_validate(spa, props))) {
spa_deactivate(spa);
spa_remove(spa);
mutex_exit(&spa_namespace_lock);
return (error);
}
if (nvlist_lookup_uint64(props, zpool_prop_to_name(ZPOOL_PROP_VERSION),
&version) != 0)
version = SPA_VERSION;
ASSERT(version <= SPA_VERSION);
spa->spa_first_txg = txg;
spa->spa_uberblock.ub_txg = txg - 1;
spa->spa_uberblock.ub_version = version;
spa->spa_ubsync = spa->spa_uberblock;
/*
* Create "The Godfather" zio to hold all async IOs
*/
spa->spa_async_zio_root = zio_root(spa, NULL, NULL,
ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_GODFATHER);
/*
* Create the root vdev.
*/
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
error = spa_config_parse(spa, &rvd, nvroot, NULL, 0, VDEV_ALLOC_ADD);
ASSERT(error != 0 || rvd != NULL);
ASSERT(error != 0 || spa->spa_root_vdev == rvd);
if (error == 0 && !zfs_allocatable_devs(nvroot))
error = EINVAL;
if (error == 0 &&
(error = vdev_create(rvd, txg, B_FALSE)) == 0 &&
(error = spa_validate_aux(spa, nvroot, txg,
VDEV_ALLOC_ADD)) == 0) {
for (c = 0; c < rvd->vdev_children; c++) {
vdev_metaslab_set_size(rvd->vdev_child[c]);
vdev_expand(rvd->vdev_child[c], txg);
}
}
spa_config_exit(spa, SCL_ALL, FTAG);
if (error != 0) {
spa_unload(spa);
spa_deactivate(spa);
spa_remove(spa);
mutex_exit(&spa_namespace_lock);
return (error);
}
/*
* Get the list of spares, if specified.
*/
if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
&spares, &nspares) == 0) {
VERIFY(nvlist_alloc(&spa->spa_spares.sav_config, NV_UNIQUE_NAME,
KM_SLEEP) == 0);
VERIFY(nvlist_add_nvlist_array(spa->spa_spares.sav_config,
ZPOOL_CONFIG_SPARES, spares, nspares) == 0);
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
spa_load_spares(spa);
spa_config_exit(spa, SCL_ALL, FTAG);
spa->spa_spares.sav_sync = B_TRUE;
}
/*
* Get the list of level 2 cache devices, if specified.
*/
if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
&l2cache, &nl2cache) == 0) {
VERIFY(nvlist_alloc(&spa->spa_l2cache.sav_config,
NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_nvlist_array(spa->spa_l2cache.sav_config,
ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache) == 0);
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
spa_load_l2cache(spa);
spa_config_exit(spa, SCL_ALL, FTAG);
spa->spa_l2cache.sav_sync = B_TRUE;
}
spa->spa_dsl_pool = dp = dsl_pool_create(spa, zplprops, txg);
spa->spa_meta_objset = dp->dp_meta_objset;
/*
* Create DDTs (dedup tables).
*/
ddt_create(spa);
spa_update_dspace(spa);
tx = dmu_tx_create_assigned(dp, txg);
/*
* Create the pool config object.
*/
spa->spa_config_object = dmu_object_alloc(spa->spa_meta_objset,
DMU_OT_PACKED_NVLIST, SPA_CONFIG_BLOCKSIZE,
DMU_OT_PACKED_NVLIST_SIZE, sizeof (uint64_t), tx);
if (zap_add(spa->spa_meta_objset,
DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CONFIG,
sizeof (uint64_t), 1, &spa->spa_config_object, tx) != 0) {
cmn_err(CE_PANIC, "failed to add pool config");
}
if (zap_add(spa->spa_meta_objset,
DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CREATION_VERSION,
sizeof (uint64_t), 1, &version, tx) != 0) {
cmn_err(CE_PANIC, "failed to add pool version");
}
/* Newly created pools with the right version are always deflated. */
if (version >= SPA_VERSION_RAIDZ_DEFLATE) {
spa->spa_deflate = TRUE;
if (zap_add(spa->spa_meta_objset,
DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DEFLATE,
sizeof (uint64_t), 1, &spa->spa_deflate, tx) != 0) {
cmn_err(CE_PANIC, "failed to add deflate");
}
}
/*
* Create the deferred-free bpobj. Turn off compression
* because sync-to-convergence takes longer if the blocksize
* keeps changing.
*/
obj = bpobj_alloc(spa->spa_meta_objset, 1 << 14, tx);
dmu_object_set_compress(spa->spa_meta_objset, obj,
ZIO_COMPRESS_OFF, tx);
if (zap_add(spa->spa_meta_objset,
DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SYNC_BPOBJ,
sizeof (uint64_t), 1, &obj, tx) != 0) {
cmn_err(CE_PANIC, "failed to add bpobj");
}
VERIFY3U(0, ==, bpobj_open(&spa->spa_deferred_bpobj,
spa->spa_meta_objset, obj));
/*
* Create the pool's history object.
*/
if (version >= SPA_VERSION_ZPOOL_HISTORY)
spa_history_create_obj(spa, tx);
/*
* Set pool properties.
*/
spa->spa_bootfs = zpool_prop_default_numeric(ZPOOL_PROP_BOOTFS);
spa->spa_delegation = zpool_prop_default_numeric(ZPOOL_PROP_DELEGATION);
spa->spa_failmode = zpool_prop_default_numeric(ZPOOL_PROP_FAILUREMODE);
spa->spa_autoexpand = zpool_prop_default_numeric(ZPOOL_PROP_AUTOEXPAND);
if (props != NULL) {
spa_configfile_set(spa, props, B_FALSE);
spa_sync_props(spa, props, tx);
}
dmu_tx_commit(tx);
spa->spa_sync_on = B_TRUE;
txg_sync_start(spa->spa_dsl_pool);
/*
* We explicitly wait for the first transaction to complete so that our
* bean counters are appropriately updated.
*/
txg_wait_synced(spa->spa_dsl_pool, txg);
spa_config_sync(spa, B_FALSE, B_TRUE);
if (version >= SPA_VERSION_ZPOOL_HISTORY && history_str != NULL)
(void) spa_history_log(spa, history_str, LOG_CMD_POOL_CREATE);
spa_history_log_version(spa, LOG_POOL_CREATE);
spa->spa_minref = refcount_count(&spa->spa_refcount);
mutex_exit(&spa_namespace_lock);
return (0);
}
#ifdef _KERNEL
/*
* Get the root pool information from the root disk, then import the root pool
* during the system boot up time.
*/
extern int vdev_disk_read_rootlabel(char *, char *, nvlist_t **);
static nvlist_t *
spa_generate_rootconf(char *devpath, char *devid, uint64_t *guid)
{
nvlist_t *config;
nvlist_t *nvtop, *nvroot;
uint64_t pgid;
if (vdev_disk_read_rootlabel(devpath, devid, &config) != 0)
return (NULL);
/*
* Add this top-level vdev to the child array.
*/
VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
&nvtop) == 0);
VERIFY(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID,
&pgid) == 0);
VERIFY(nvlist_lookup_uint64(config, ZPOOL_CONFIG_GUID, guid) == 0);
/*
* Put this pool's top-level vdevs into a root vdev.
*/
VERIFY(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_string(nvroot, ZPOOL_CONFIG_TYPE,
VDEV_TYPE_ROOT) == 0);
VERIFY(nvlist_add_uint64(nvroot, ZPOOL_CONFIG_ID, 0ULL) == 0);
VERIFY(nvlist_add_uint64(nvroot, ZPOOL_CONFIG_GUID, pgid) == 0);
VERIFY(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
&nvtop, 1) == 0);
/*
* Replace the existing vdev_tree with the new root vdev in
* this pool's configuration (remove the old, add the new).
*/
VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, nvroot) == 0);
nvlist_free(nvroot);
return (config);
}
/*
* Walk the vdev tree and see if we can find a device with "better"
* configuration. A configuration is "better" if the label on that
* device has a more recent txg.
*/
static void
spa_alt_rootvdev(vdev_t *vd, vdev_t **avd, uint64_t *txg)
{
int c;
for (c = 0; c < vd->vdev_children; c++)
spa_alt_rootvdev(vd->vdev_child[c], avd, txg);
if (vd->vdev_ops->vdev_op_leaf) {
nvlist_t *label;
uint64_t label_txg;
if (vdev_disk_read_rootlabel(vd->vdev_physpath, vd->vdev_devid,
&label) != 0)
return;
VERIFY(nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_TXG,
&label_txg) == 0);
/*
* Do we have a better boot device?
*/
if (label_txg > *txg) {
*txg = label_txg;
*avd = vd;
}
nvlist_free(label);
}
}
/*
* Import a root pool.
*
* For x86. devpath_list will consist of devid and/or physpath name of
* the vdev (e.g. "id1,sd@SSEAGATE..." or "/pci@1f,0/ide@d/disk@0,0:a").
* The GRUB "findroot" command will return the vdev we should boot.
*
* For Sparc, devpath_list consists the physpath name of the booting device
* no matter the rootpool is a single device pool or a mirrored pool.
* e.g.
* "/pci@1f,0/ide@d/disk@0,0:a"
*/
int
spa_import_rootpool(char *devpath, char *devid)
{
spa_t *spa;
vdev_t *rvd, *bvd, *avd = NULL;
nvlist_t *config, *nvtop;
uint64_t guid, txg;
char *pname;
int error;
/*
* Read the label from the boot device and generate a configuration.
*/
config = spa_generate_rootconf(devpath, devid, &guid);
#if defined(_OBP) && defined(_KERNEL)
if (config == NULL) {
if (strstr(devpath, "/iscsi/ssd") != NULL) {
/* iscsi boot */
get_iscsi_bootpath_phy(devpath);
config = spa_generate_rootconf(devpath, devid, &guid);
}
}
#endif
if (config == NULL) {
cmn_err(CE_NOTE, "Can not read the pool label from '%s'",
devpath);
return (EIO);
}
VERIFY(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME,
&pname) == 0);
VERIFY(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG, &txg) == 0);
mutex_enter(&spa_namespace_lock);
if ((spa = spa_lookup(pname)) != NULL) {
/*
* Remove the existing root pool from the namespace so that we
* can replace it with the correct config we just read in.
*/
spa_remove(spa);
}
spa = spa_add(pname, config, NULL);
spa->spa_is_root = B_TRUE;
spa->spa_load_verbatim = B_TRUE;
/*
* Build up a vdev tree based on the boot device's label config.
*/
VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
&nvtop) == 0);
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
error = spa_config_parse(spa, &rvd, nvtop, NULL, 0,
VDEV_ALLOC_ROOTPOOL);
spa_config_exit(spa, SCL_ALL, FTAG);
if (error) {
mutex_exit(&spa_namespace_lock);
nvlist_free(config);
cmn_err(CE_NOTE, "Can not parse the config for pool '%s'",
pname);
return (error);
}
/*
* Get the boot vdev.
*/
if ((bvd = vdev_lookup_by_guid(rvd, guid)) == NULL) {
cmn_err(CE_NOTE, "Can not find the boot vdev for guid %llu",
(u_longlong_t)guid);
error = ENOENT;
goto out;
}
/*
* Determine if there is a better boot device.
*/
avd = bvd;
spa_alt_rootvdev(rvd, &avd, &txg);
if (avd != bvd) {
cmn_err(CE_NOTE, "The boot device is 'degraded'. Please "
"try booting from '%s'", avd->vdev_path);
error = EINVAL;
goto out;
}
/*
* If the boot device is part of a spare vdev then ensure that
* we're booting off the active spare.
*/
if (bvd->vdev_parent->vdev_ops == &vdev_spare_ops &&
!bvd->vdev_isspare) {
cmn_err(CE_NOTE, "The boot device is currently spared. Please "
"try booting from '%s'",
bvd->vdev_parent->vdev_child[1]->vdev_path);
error = EINVAL;
goto out;
}
error = 0;
spa_history_log_version(spa, LOG_POOL_IMPORT);
out:
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
vdev_free(rvd);
spa_config_exit(spa, SCL_ALL, FTAG);
mutex_exit(&spa_namespace_lock);
nvlist_free(config);
return (error);
}
#endif
/*
* Take a pool and insert it into the namespace as if it had been loaded at
* boot.
*/
int
spa_import_verbatim(const char *pool, nvlist_t *config, nvlist_t *props)
{
spa_t *spa;
char *altroot = NULL;
mutex_enter(&spa_namespace_lock);
if (spa_lookup(pool) != NULL) {
mutex_exit(&spa_namespace_lock);
return (EEXIST);
}
(void) nvlist_lookup_string(props,
zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot);
spa = spa_add(pool, config, altroot);
spa->spa_load_verbatim = B_TRUE;
if (props != NULL)
spa_configfile_set(spa, props, B_FALSE);
spa_config_sync(spa, B_FALSE, B_TRUE);
mutex_exit(&spa_namespace_lock);
spa_history_log_version(spa, LOG_POOL_IMPORT);
return (0);
}
/*
* Import a non-root pool into the system.
*/
int
spa_import(const char *pool, nvlist_t *config, nvlist_t *props)
{
spa_t *spa;
char *altroot = NULL;
spa_load_state_t state = SPA_LOAD_IMPORT;
zpool_rewind_policy_t policy;
int error;
nvlist_t *nvroot;
nvlist_t **spares, **l2cache;
uint_t nspares, nl2cache;
/*
* If a pool with this name exists, return failure.
*/
mutex_enter(&spa_namespace_lock);
if (spa_lookup(pool) != NULL) {
mutex_exit(&spa_namespace_lock);
return (EEXIST);
}
zpool_get_rewind_policy(config, &policy);
if (policy.zrp_request & ZPOOL_DO_REWIND)
state = SPA_LOAD_RECOVER;
/*
* Create and initialize the spa structure.
*/
(void) nvlist_lookup_string(props,
zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot);
spa = spa_add(pool, config, altroot);
spa_activate(spa, spa_mode_global);
/*
* Don't start async tasks until we know everything is healthy.
*/
spa_async_suspend(spa);
/*
* 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;
error = spa_load_best(spa, state, B_TRUE, policy.zrp_txg,
policy.zrp_request);
/*
* Propagate anything learned about failing or best txgs
* back to caller
*/
spa_rewind_data_to_nvlist(spa, config);
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
/*
* Toss any existing sparelist, as it doesn't have any validity
* anymore, and conflicts with spa_has_spare().
*/
if (spa->spa_spares.sav_config) {
nvlist_free(spa->spa_spares.sav_config);
spa->spa_spares.sav_config = NULL;
spa_load_spares(spa);
}
if (spa->spa_l2cache.sav_config) {
nvlist_free(spa->spa_l2cache.sav_config);
spa->spa_l2cache.sav_config = NULL;
spa_load_l2cache(spa);
}
VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
&nvroot) == 0);
if (error == 0)
error = spa_validate_aux(spa, nvroot, -1ULL,
VDEV_ALLOC_SPARE);
if (error == 0)
error = spa_validate_aux(spa, nvroot, -1ULL,
VDEV_ALLOC_L2CACHE);
spa_config_exit(spa, SCL_ALL, FTAG);
if (props != NULL)
spa_configfile_set(spa, props, B_FALSE);
if (error != 0 || (props && spa_writeable(spa) &&
(error = spa_prop_set(spa, props)))) {
spa_unload(spa);
spa_deactivate(spa);
spa_remove(spa);
mutex_exit(&spa_namespace_lock);
return (error);
}
/*
* Override any spares and level 2 cache devices as specified by
* the user, as these may have correct device names/devids, etc.
*/
if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
&spares, &nspares) == 0) {
if (spa->spa_spares.sav_config)
VERIFY(nvlist_remove(spa->spa_spares.sav_config,
ZPOOL_CONFIG_SPARES, DATA_TYPE_NVLIST_ARRAY) == 0);
else
VERIFY(nvlist_alloc(&spa->spa_spares.sav_config,
NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_nvlist_array(spa->spa_spares.sav_config,
ZPOOL_CONFIG_SPARES, spares, nspares) == 0);
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
spa_load_spares(spa);
spa_config_exit(spa, SCL_ALL, FTAG);
spa->spa_spares.sav_sync = B_TRUE;
}
if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
&l2cache, &nl2cache) == 0) {
if (spa->spa_l2cache.sav_config)
VERIFY(nvlist_remove(spa->spa_l2cache.sav_config,
ZPOOL_CONFIG_L2CACHE, DATA_TYPE_NVLIST_ARRAY) == 0);
else
VERIFY(nvlist_alloc(&spa->spa_l2cache.sav_config,
NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_nvlist_array(spa->spa_l2cache.sav_config,
ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache) == 0);
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
spa_load_l2cache(spa);
spa_config_exit(spa, SCL_ALL, FTAG);
spa->spa_l2cache.sav_sync = B_TRUE;
}
/*
* Check for any removed devices.
*/
if (spa->spa_autoreplace) {
spa_aux_check_removed(&spa->spa_spares);
spa_aux_check_removed(&spa->spa_l2cache);
}
if (spa_writeable(spa)) {
/*
* Update the config cache to include the newly-imported pool.
*/
spa_config_update(spa, SPA_CONFIG_UPDATE_POOL);
}
spa_async_resume(spa);
/*
* It's possible that the pool was expanded while it was exported.
* We kick off an async task to handle this for us.
*/
spa_async_request(spa, SPA_ASYNC_AUTOEXPAND);
mutex_exit(&spa_namespace_lock);
spa_history_log_version(spa, LOG_POOL_IMPORT);
return (0);
}
nvlist_t *
spa_tryimport(nvlist_t *tryconfig)
{
nvlist_t *config = NULL;
char *poolname;
spa_t *spa;
uint64_t state;
int error;
if (nvlist_lookup_string(tryconfig, ZPOOL_CONFIG_POOL_NAME, &poolname))
return (NULL);
if (nvlist_lookup_uint64(tryconfig, ZPOOL_CONFIG_POOL_STATE, &state))
return (NULL);
/*
* Create and initialize the spa structure.
*/
mutex_enter(&spa_namespace_lock);
spa = spa_add(TRYIMPORT_NAME, tryconfig, 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.
*/
error = spa_load(spa, SPA_LOAD_TRYIMPORT, SPA_IMPORT_EXISTING, B_TRUE);
/*
* If 'tryconfig' was at least parsable, return the current config.
*/
if (spa->spa_root_vdev != NULL) {
config = spa_config_generate(spa, NULL, -1ULL, B_TRUE);
VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME,
poolname) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE,
state) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_TIMESTAMP,
spa->spa_uberblock.ub_timestamp) == 0);
/*
* If the bootfs property exists on this pool then we
* copy it out so that external consumers can tell which
* pools are bootable.
*/
if ((!error || error == EEXIST) && spa->spa_bootfs) {
char *tmpname = kmem_alloc(MAXPATHLEN, KM_SLEEP);
/*
* We have to play games with the name since the
* pool was opened as TRYIMPORT_NAME.
*/
if (dsl_dsobj_to_dsname(spa_name(spa),
spa->spa_bootfs, tmpname) == 0) {
char *cp;
char *dsname = kmem_alloc(MAXPATHLEN, KM_SLEEP);
cp = strchr(tmpname, '/');
if (cp == NULL) {
(void) strlcpy(dsname, tmpname,
MAXPATHLEN);
} else {
(void) snprintf(dsname, MAXPATHLEN,
"%s/%s", poolname, ++cp);
}
VERIFY(nvlist_add_string(config,
ZPOOL_CONFIG_BOOTFS, dsname) == 0);
kmem_free(dsname, MAXPATHLEN);
}
kmem_free(tmpname, MAXPATHLEN);
}
/*
* Add the list of hot spares and level 2 cache devices.
*/
spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
spa_add_spares(spa, config);
spa_add_l2cache(spa, config);
spa_config_exit(spa, SCL_CONFIG, FTAG);
}
spa_unload(spa);
spa_deactivate(spa);
spa_remove(spa);
mutex_exit(&spa_namespace_lock);
return (config);
}
/*
* Pool export/destroy
*
* The act of destroying or exporting a pool is very simple. We make sure there
* is no more pending I/O and any references to the pool are gone. Then, we
* update the pool state and sync all the labels to disk, removing the
* configuration from the cache afterwards. If the 'hardforce' flag is set, then
* we don't sync the labels or remove the configuration cache.
*/
static int
spa_export_common(char *pool, int new_state, nvlist_t **oldconfig,
boolean_t force, boolean_t hardforce)
{
spa_t *spa;
if (oldconfig)
*oldconfig = NULL;
if (!(spa_mode_global & FWRITE))
return (EROFS);
mutex_enter(&spa_namespace_lock);
if ((spa = spa_lookup(pool)) == NULL) {
mutex_exit(&spa_namespace_lock);
return (ENOENT);
}
/*
* Put a hold on the pool, drop the namespace lock, stop async tasks,
* reacquire the namespace lock, and see if we can export.
*/
spa_open_ref(spa, FTAG);
mutex_exit(&spa_namespace_lock);
spa_async_suspend(spa);
mutex_enter(&spa_namespace_lock);
spa_close(spa, FTAG);
/*
* The pool will be in core if it's openable,
* in which case we can modify its state.
*/
if (spa->spa_state != POOL_STATE_UNINITIALIZED && spa->spa_sync_on) {
/*
* Objsets may be open only because they're dirty, so we
* have to force it to sync before checking spa_refcnt.
*/
txg_wait_synced(spa->spa_dsl_pool, 0);
/*
* A pool cannot be exported or destroyed if there are active
* references. If we are resetting a pool, allow references by
* fault injection handlers.
*/
if (!spa_refcount_zero(spa) ||
(spa->spa_inject_ref != 0 &&
new_state != POOL_STATE_UNINITIALIZED)) {
spa_async_resume(spa);
mutex_exit(&spa_namespace_lock);
return (EBUSY);
}
/*
* A pool cannot be exported if it has an active shared spare.
* This is to prevent other pools stealing the active spare
* from an exported pool. At user's own will, such pool can
* be forcedly exported.
*/
if (!force && new_state == POOL_STATE_EXPORTED &&
spa_has_active_shared_spare(spa)) {
spa_async_resume(spa);
mutex_exit(&spa_namespace_lock);
return (EXDEV);
}
/*
* 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.
*/
if (new_state != POOL_STATE_UNINITIALIZED && !hardforce) {
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
spa->spa_state = new_state;
spa->spa_final_txg = spa_last_synced_txg(spa) +
TXG_DEFER_SIZE + 1;
vdev_config_dirty(spa->spa_root_vdev);
spa_config_exit(spa, SCL_ALL, FTAG);
}
}
spa_event_notify(spa, NULL, FM_EREPORT_ZFS_POOL_DESTROY);
if (spa->spa_state != POOL_STATE_UNINITIALIZED) {
spa_unload(spa);
spa_deactivate(spa);
}
if (oldconfig && spa->spa_config)
VERIFY(nvlist_dup(spa->spa_config, oldconfig, 0) == 0);
if (new_state != POOL_STATE_UNINITIALIZED) {
if (!hardforce)
spa_config_sync(spa, B_TRUE, B_TRUE);
spa_remove(spa);
}
mutex_exit(&spa_namespace_lock);
return (0);
}
/*
* Destroy a storage pool.
*/
int
spa_destroy(char *pool)
{
return (spa_export_common(pool, POOL_STATE_DESTROYED, NULL,
B_FALSE, B_FALSE));
}
/*
* Export a storage pool.
*/
int
spa_export(char *pool, nvlist_t **oldconfig, boolean_t force,
boolean_t hardforce)
{
return (spa_export_common(pool, POOL_STATE_EXPORTED, oldconfig,
force, hardforce));
}
/*
* Similar to spa_export(), this unloads the spa_t without actually removing it
* from the namespace in any way.
*/
int
spa_reset(char *pool)
{
return (spa_export_common(pool, POOL_STATE_UNINITIALIZED, NULL,
B_FALSE, B_FALSE));
}
/*
* ==========================================================================
* Device manipulation
* ==========================================================================
*/
/*
* Add a device to a storage pool.
*/
int
spa_vdev_add(spa_t *spa, nvlist_t *nvroot)
{
uint64_t txg, id;
int error;
vdev_t *rvd = spa->spa_root_vdev;
vdev_t *vd, *tvd;
nvlist_t **spares, **l2cache;
uint_t nspares, nl2cache;
int c;
txg = spa_vdev_enter(spa);
if ((error = spa_config_parse(spa, &vd, nvroot, NULL, 0,
VDEV_ALLOC_ADD)) != 0)
return (spa_vdev_exit(spa, NULL, txg, error));
spa->spa_pending_vdev = vd; /* spa_vdev_exit() will clear this */
if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares,
&nspares) != 0)
nspares = 0;
if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache,
&nl2cache) != 0)
nl2cache = 0;
if (vd->vdev_children == 0 && nspares == 0 && nl2cache == 0)
return (spa_vdev_exit(spa, vd, txg, EINVAL));
if (vd->vdev_children != 0 &&
(error = vdev_create(vd, txg, B_FALSE)) != 0)
return (spa_vdev_exit(spa, vd, txg, error));
/*
* We must validate the spares and l2cache devices after checking the
* children. Otherwise, vdev_inuse() will blindly overwrite the spare.
*/
if ((error = spa_validate_aux(spa, nvroot, txg, VDEV_ALLOC_ADD)) != 0)
return (spa_vdev_exit(spa, vd, txg, error));
/*
* Transfer each new top-level vdev from vd to rvd.
*/
for (c = 0; c < vd->vdev_children; c++) {
/*
* Set the vdev id to the first hole, if one exists.
*/
for (id = 0; id < rvd->vdev_children; id++) {
if (rvd->vdev_child[id]->vdev_ishole) {
vdev_free(rvd->vdev_child[id]);
break;
}
}
tvd = vd->vdev_child[c];
vdev_remove_child(vd, tvd);
tvd->vdev_id = id;
vdev_add_child(rvd, tvd);
vdev_config_dirty(tvd);
}
if (nspares != 0) {
spa_set_aux_vdevs(&spa->spa_spares, spares, nspares,
ZPOOL_CONFIG_SPARES);
spa_load_spares(spa);
spa->spa_spares.sav_sync = B_TRUE;
}
if (nl2cache != 0) {
spa_set_aux_vdevs(&spa->spa_l2cache, l2cache, nl2cache,
ZPOOL_CONFIG_L2CACHE);
spa_load_l2cache(spa);
spa->spa_l2cache.sav_sync = B_TRUE;
}
/*
* We have to be careful when adding new vdevs to an existing pool.
* If other threads start allocating from these vdevs before we
* sync the config cache, and we lose power, then upon reboot we may
* fail to open the pool because there are DVAs that the config cache
* can't translate. Therefore, we first add the vdevs without
* initializing metaslabs; sync the config cache (via spa_vdev_exit());
* and then let spa_config_update() initialize the new metaslabs.
*
* spa_load() checks for added-but-not-initialized vdevs, so that
* if we lose power at any point in this sequence, the remaining
* steps will be completed the next time we load the pool.
*/
(void) spa_vdev_exit(spa, vd, txg, 0);
mutex_enter(&spa_namespace_lock);
spa_config_update(spa, SPA_CONFIG_UPDATE_POOL);
mutex_exit(&spa_namespace_lock);
return (0);
}
/*
* Attach a device to a mirror. The arguments are the path to any device
* in the mirror, and the nvroot for the new device. If the path specifies
* a device that is not mirrored, we automatically insert the mirror vdev.
*
* If 'replacing' is specified, the new device is intended to replace the
* existing device; in this case the two devices are made into their own
* mirror using the 'replacing' vdev, which is functionally identical to
* the mirror vdev (it actually reuses all the same ops) but has a few
* extra rules: you can't attach to it after it's been created, and upon
* completion of resilvering, the first disk (the one being replaced)
* is automatically detached.
*/
int
spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing)
{
uint64_t txg, dtl_max_txg;
ASSERTV(vdev_t *rvd = spa->spa_root_vdev;)
vdev_t *oldvd, *newvd, *newrootvd, *pvd, *tvd;
vdev_ops_t *pvops;
char *oldvdpath, *newvdpath;
int newvd_isspare;
int error;
txg = spa_vdev_enter(spa);
oldvd = spa_lookup_by_guid(spa, guid, B_FALSE);
if (oldvd == NULL)
return (spa_vdev_exit(spa, NULL, txg, ENODEV));
if (!oldvd->vdev_ops->vdev_op_leaf)
return (spa_vdev_exit(spa, NULL, txg, ENOTSUP));
pvd = oldvd->vdev_parent;
if ((error = spa_config_parse(spa, &newrootvd, nvroot, NULL, 0,
VDEV_ALLOC_ADD)) != 0)
return (spa_vdev_exit(spa, NULL, txg, EINVAL));
if (newrootvd->vdev_children != 1)
return (spa_vdev_exit(spa, newrootvd, txg, EINVAL));
newvd = newrootvd->vdev_child[0];
if (!newvd->vdev_ops->vdev_op_leaf)
return (spa_vdev_exit(spa, newrootvd, txg, EINVAL));
if ((error = vdev_create(newrootvd, txg, replacing)) != 0)
return (spa_vdev_exit(spa, newrootvd, txg, error));
/*
* Spares can't replace logs
*/
if (oldvd->vdev_top->vdev_islog && newvd->vdev_isspare)
return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));
if (!replacing) {
/*
* For attach, the only allowable parent is a mirror or the root
* vdev.
*/
if (pvd->vdev_ops != &vdev_mirror_ops &&
pvd->vdev_ops != &vdev_root_ops)
return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));
pvops = &vdev_mirror_ops;
} else {
/*
* Active hot spares can only be replaced by inactive hot
* spares.
*/
if (pvd->vdev_ops == &vdev_spare_ops &&
pvd->vdev_child[1] == oldvd &&
!spa_has_spare(spa, newvd->vdev_guid))
return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));
/*
* If the source is a hot spare, and the parent isn't already a
* spare, then we want to create a new hot spare. Otherwise, we
* want to create a replacing vdev. The user is not allowed to
* attach to a spared vdev child unless the 'isspare' state is
* the same (spare replaces spare, non-spare replaces
* non-spare).
*/
if (pvd->vdev_ops == &vdev_replacing_ops)
return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));
else if (pvd->vdev_ops == &vdev_spare_ops &&
newvd->vdev_isspare != oldvd->vdev_isspare)
return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));
else if (pvd->vdev_ops != &vdev_spare_ops &&
newvd->vdev_isspare)
pvops = &vdev_spare_ops;
else
pvops = &vdev_replacing_ops;
}
/*
* Make sure the new device is big enough.
*/
if (newvd->vdev_asize < vdev_get_min_asize(oldvd))
return (spa_vdev_exit(spa, newrootvd, txg, EOVERFLOW));
/*
* The new device cannot have a higher alignment requirement
* than the top-level vdev.
*/
if (newvd->vdev_ashift > oldvd->vdev_top->vdev_ashift)
return (spa_vdev_exit(spa, newrootvd, txg, EDOM));
/*
* If this is an in-place replacement, update oldvd's path and devid
* to make it distinguishable from newvd, and unopenable from now on.
*/
if (strcmp(oldvd->vdev_path, newvd->vdev_path) == 0) {
spa_strfree(oldvd->vdev_path);
oldvd->vdev_path = kmem_alloc(strlen(newvd->vdev_path) + 5,
KM_SLEEP);
(void) sprintf(oldvd->vdev_path, "%s/%s",
newvd->vdev_path, "old");
if (oldvd->vdev_devid != NULL) {
spa_strfree(oldvd->vdev_devid);
oldvd->vdev_devid = NULL;
}
}
/*
* If the parent is not a mirror, or if we're replacing, insert the new
* mirror/replacing/spare vdev above oldvd.
*/
if (pvd->vdev_ops != pvops)
pvd = vdev_add_parent(oldvd, pvops);
ASSERT(pvd->vdev_top->vdev_parent == rvd);
ASSERT(pvd->vdev_ops == pvops);
ASSERT(oldvd->vdev_parent == pvd);
/*
* Extract the new device from its root and add it to pvd.
*/
vdev_remove_child(newrootvd, newvd);
newvd->vdev_id = pvd->vdev_children;
newvd->vdev_crtxg = oldvd->vdev_crtxg;
vdev_add_child(pvd, newvd);
tvd = newvd->vdev_top;
ASSERT(pvd->vdev_top == tvd);
ASSERT(tvd->vdev_parent == rvd);
vdev_config_dirty(tvd);
/*
* Set newvd's DTL to [TXG_INITIAL, dtl_max_txg) so that we account
* for any dmu_sync-ed blocks. It will propagate upward when
* spa_vdev_exit() calls vdev_dtl_reassess().
*/
dtl_max_txg = txg + TXG_CONCURRENT_STATES;
vdev_dtl_dirty(newvd, DTL_MISSING, TXG_INITIAL,
dtl_max_txg - TXG_INITIAL);
if (newvd->vdev_isspare) {
spa_spare_activate(newvd);
spa_event_notify(spa, newvd, FM_EREPORT_ZFS_DEVICE_SPARE);
}
oldvdpath = spa_strdup(oldvd->vdev_path);
newvdpath = spa_strdup(newvd->vdev_path);
newvd_isspare = newvd->vdev_isspare;
/*
* Mark newvd's DTL dirty in this txg.
*/
vdev_dirty(tvd, VDD_DTL, newvd, txg);
/*
* Restart the resilver
*/
dsl_resilver_restart(spa->spa_dsl_pool, dtl_max_txg);
/*
* Commit the config
*/
(void) spa_vdev_exit(spa, newrootvd, dtl_max_txg, 0);
spa_history_log_internal(LOG_POOL_VDEV_ATTACH, spa, NULL,
"%s vdev=%s %s vdev=%s",
replacing && newvd_isspare ? "spare in" :
replacing ? "replace" : "attach", newvdpath,
replacing ? "for" : "to", oldvdpath);
spa_strfree(oldvdpath);
spa_strfree(newvdpath);
return (0);
}
/*
* Detach a device from a mirror or replacing vdev.
* If 'replace_done' is specified, only detach if the parent
* is a replacing vdev.
*/
int
spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done)
{
uint64_t txg;
int error;
ASSERTV(vdev_t *rvd = spa->spa_root_vdev;)
vdev_t *vd, *pvd, *cvd, *tvd;
boolean_t unspare = B_FALSE;
uint64_t unspare_guid = 0;
size_t len;
char *vdpath;
int t;
txg = spa_vdev_enter(spa);
vd = spa_lookup_by_guid(spa, guid, B_FALSE);
if (vd == NULL)
return (spa_vdev_exit(spa, NULL, txg, ENODEV));
if (!vd->vdev_ops->vdev_op_leaf)
return (spa_vdev_exit(spa, NULL, txg, ENOTSUP));
pvd = vd->vdev_parent;
/*
* If the parent/child relationship is not as expected, don't do it.
* Consider M(A,R(B,C)) -- that is, a mirror of A with a replacing
* vdev that's replacing B with C. The user's intent in replacing
* is to go from M(A,B) to M(A,C). If the user decides to cancel
* the replace by detaching C, the expected behavior is to end up
* M(A,B). But suppose that right after deciding to detach C,
* the replacement of B completes. We would have M(A,C), and then
* ask to detach C, which would leave us with just A -- not what
* the user wanted. To prevent this, we make sure that the
* parent/child relationship hasn't changed -- in this example,
* that C's parent is still the replacing vdev R.
*/
if (pvd->vdev_guid != pguid && pguid != 0)
return (spa_vdev_exit(spa, NULL, txg, EBUSY));
/*
* If replace_done is specified, only remove this device if it's
* the first child of a replacing vdev. For the 'spare' vdev, either
* disk can be removed.
*/
if (replace_done) {
if (pvd->vdev_ops == &vdev_replacing_ops) {
if (vd->vdev_id != 0)
return (spa_vdev_exit(spa, NULL, txg, ENOTSUP));
} else if (pvd->vdev_ops != &vdev_spare_ops) {
return (spa_vdev_exit(spa, NULL, txg, ENOTSUP));
}
}
ASSERT(pvd->vdev_ops != &vdev_spare_ops ||
spa_version(spa) >= SPA_VERSION_SPARES);
/*
* Only mirror, replacing, and spare vdevs support detach.
*/
if (pvd->vdev_ops != &vdev_replacing_ops &&
pvd->vdev_ops != &vdev_mirror_ops &&
pvd->vdev_ops != &vdev_spare_ops)
return (spa_vdev_exit(spa, NULL, txg, ENOTSUP));
/*
* If this device has the only valid copy of some data,
* we cannot safely detach it.
*/
if (vdev_dtl_required(vd))
return (spa_vdev_exit(spa, NULL, txg, EBUSY));
ASSERT(pvd->vdev_children >= 2);
/*
* If we are detaching the second disk from a replacing vdev, then
* check to see if we changed the original vdev's path to have "/old"
* at the end in spa_vdev_attach(). If so, undo that change now.
*/
if (pvd->vdev_ops == &vdev_replacing_ops && vd->vdev_id == 1 &&
pvd->vdev_child[0]->vdev_path != NULL &&
pvd->vdev_child[1]->vdev_path != NULL) {
ASSERT(pvd->vdev_child[1] == vd);
cvd = pvd->vdev_child[0];
len = strlen(vd->vdev_path);
if (strncmp(cvd->vdev_path, vd->vdev_path, len) == 0 &&
strcmp(cvd->vdev_path + len, "/old") == 0) {
spa_strfree(cvd->vdev_path);
cvd->vdev_path = spa_strdup(vd->vdev_path);
}
}
/*
* If we are detaching the original disk from a spare, then it implies
* that the spare should become a real disk, and be removed from the
* active spare list for the pool.
*/
if (pvd->vdev_ops == &vdev_spare_ops &&
vd->vdev_id == 0 && pvd->vdev_child[1]->vdev_isspare)
unspare = B_TRUE;
/*
* Erase the disk labels so the disk can be used for other things.
* This must be done after all other error cases are handled,
* but before we disembowel vd (so we can still do I/O to it).
* But if we can't do it, don't treat the error as fatal --
* it may be that the unwritability of the disk is the reason
* it's being detached!
*/
error = vdev_label_init(vd, 0, VDEV_LABEL_REMOVE);
/*
* Remove vd from its parent and compact the parent's children.
*/
vdev_remove_child(pvd, vd);
vdev_compact_children(pvd);
/*
* Remember one of the remaining children so we can get tvd below.
*/
cvd = pvd->vdev_child[0];
/*
* If we need to remove the remaining child from the list of hot spares,
* do it now, marking the vdev as no longer a spare in the process.
* We must do this before vdev_remove_parent(), because that can
* change the GUID if it creates a new toplevel GUID. For a similar
* reason, we must remove the spare now, in the same txg as the detach;
* otherwise someone could attach a new sibling, change the GUID, and
* the subsequent attempt to spa_vdev_remove(unspare_guid) would fail.
*/
if (unspare) {
ASSERT(cvd->vdev_isspare);
spa_spare_remove(cvd);
unspare_guid = cvd->vdev_guid;
(void) spa_vdev_remove(spa, unspare_guid, B_TRUE);
}
/*
* If the parent mirror/replacing vdev only has one child,
* the parent is no longer needed. Remove it from the tree.
*/
if (pvd->vdev_children == 1)
vdev_remove_parent(cvd);
/*
* We don't set tvd until now because the parent we just removed
* may have been the previous top-level vdev.
*/
tvd = cvd->vdev_top;
ASSERT(tvd->vdev_parent == rvd);
/*
* Reevaluate the parent vdev state.
*/
vdev_propagate_state(cvd);
/*
* If the 'autoexpand' property is set on the pool then automatically
* try to expand the size of the pool. For example if the device we
* just detached was smaller than the others, it may be possible to
* add metaslabs (i.e. grow the pool). We need to reopen the vdev
* first so that we can obtain the updated sizes of the leaf vdevs.
*/
if (spa->spa_autoexpand) {
vdev_reopen(tvd);
vdev_expand(tvd, txg);
}
vdev_config_dirty(tvd);
/*
* Mark vd's DTL as dirty in this txg. vdev_dtl_sync() will see that
* vd->vdev_detached is set and free vd's DTL object in syncing context.
* But first make sure we're not on any *other* txg's DTL list, to
* prevent vd from being accessed after it's freed.
*/
vdpath = spa_strdup(vd->vdev_path);
for (t = 0; t < TXG_SIZE; t++)
(void) txg_list_remove_this(&tvd->vdev_dtl_list, vd, t);
vd->vdev_detached = B_TRUE;
vdev_dirty(tvd, VDD_DTL, vd, txg);
spa_event_notify(spa, vd, FM_EREPORT_ZFS_DEVICE_REMOVE);
error = spa_vdev_exit(spa, vd, txg, 0);
spa_history_log_internal(LOG_POOL_VDEV_DETACH, spa, NULL,
"vdev=%s", vdpath);
spa_strfree(vdpath);
/*
* If this was the removal of the original device in a hot spare vdev,
* then we want to go through and remove the device from the hot spare
* list of every other pool.
*/
if (unspare) {
spa_t *myspa = spa;
spa = NULL;
mutex_enter(&spa_namespace_lock);
while ((spa = spa_next(spa)) != NULL) {
if (spa->spa_state != POOL_STATE_ACTIVE)
continue;
if (spa == myspa)
continue;
spa_open_ref(spa, FTAG);
mutex_exit(&spa_namespace_lock);
(void) spa_vdev_remove(spa, unspare_guid,
B_TRUE);
mutex_enter(&spa_namespace_lock);
spa_close(spa, FTAG);
}
mutex_exit(&spa_namespace_lock);
}
return (error);
}
/*
* Split a set of devices from their mirrors, and create a new pool from them.
*/
int
spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config,
nvlist_t *props, boolean_t exp)
{
int error = 0;
uint64_t txg, *glist;
spa_t *newspa;
uint_t c, children, lastlog;
nvlist_t **child, *nvl, *tmp;
dmu_tx_t *tx;
char *altroot = NULL;
vdev_t *rvd, **vml = NULL; /* vdev modify list */
boolean_t activate_slog;
if (!spa_writeable(spa))
return (EROFS);
txg = spa_vdev_enter(spa);
/* 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);
error = spa_offline_log(spa);
txg = spa_vdev_config_enter(spa);
if (activate_slog)
spa_activate_log(spa);
if (error != 0)
return (spa_vdev_exit(spa, NULL, txg, error));
/* check new spa name before going any further */
if (spa_lookup(newname) != NULL)
return (spa_vdev_exit(spa, NULL, txg, EEXIST));
/*
* scan through all the children to ensure they're all mirrors
*/
if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvl) != 0 ||
nvlist_lookup_nvlist_array(nvl, ZPOOL_CONFIG_CHILDREN, &child,
&children) != 0)
return (spa_vdev_exit(spa, NULL, txg, EINVAL));
/* first, check to ensure we've got the right child count */
rvd = spa->spa_root_vdev;
lastlog = 0;
for (c = 0; c < rvd->vdev_children; c++) {
vdev_t *vd = rvd->vdev_child[c];
/* don't count the holes & logs as children */
if (vd->vdev_islog || vd->vdev_ishole) {
if (lastlog == 0)
lastlog = c;
continue;
}
lastlog = 0;
}
if (children != (lastlog != 0 ? lastlog : rvd->vdev_children))
return (spa_vdev_exit(spa, NULL, txg, EINVAL));
/* next, ensure no spare or cache devices are part of the split */
if (nvlist_lookup_nvlist(nvl, ZPOOL_CONFIG_SPARES, &tmp) == 0 ||
nvlist_lookup_nvlist(nvl, ZPOOL_CONFIG_L2CACHE, &tmp) == 0)
return (spa_vdev_exit(spa, NULL, txg, EINVAL));
vml = kmem_zalloc(children * sizeof (vdev_t *), KM_SLEEP);
glist = kmem_zalloc(children * sizeof (uint64_t), KM_SLEEP);
/* then, loop over each vdev and validate it */
for (c = 0; c < children; c++) {
uint64_t is_hole = 0;
(void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_HOLE,
&is_hole);
if (is_hole != 0) {
if (spa->spa_root_vdev->vdev_child[c]->vdev_ishole ||
spa->spa_root_vdev->vdev_child[c]->vdev_islog) {
continue;
} else {
error = EINVAL;
break;
}
}
/* which disk is going to be split? */
if (nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_GUID,
&glist[c]) != 0) {
error = EINVAL;
break;
}
/* look it up in the spa */
vml[c] = spa_lookup_by_guid(spa, glist[c], B_FALSE);
if (vml[c] == NULL) {
error = ENODEV;
break;
}
/* make sure there's nothing stopping the split */
if (vml[c]->vdev_parent->vdev_ops != &vdev_mirror_ops ||
vml[c]->vdev_islog ||
vml[c]->vdev_ishole ||
vml[c]->vdev_isspare ||
vml[c]->vdev_isl2cache ||
!vdev_writeable(vml[c]) ||
vml[c]->vdev_children != 0 ||
vml[c]->vdev_state != VDEV_STATE_HEALTHY ||
c != spa->spa_root_vdev->vdev_child[c]->vdev_id) {
error = EINVAL;
break;
}
if (vdev_dtl_required(vml[c])) {
error = EBUSY;
break;
}
/* we need certain info from the top level */
VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_METASLAB_ARRAY,
vml[c]->vdev_top->vdev_ms_array) == 0);
VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_METASLAB_SHIFT,
vml[c]->vdev_top->vdev_ms_shift) == 0);
VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_ASIZE,
vml[c]->vdev_top->vdev_asize) == 0);
VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_ASHIFT,
vml[c]->vdev_top->vdev_ashift) == 0);
}
if (error != 0) {
kmem_free(vml, children * sizeof (vdev_t *));
kmem_free(glist, children * sizeof (uint64_t));
return (spa_vdev_exit(spa, NULL, txg, error));
}
/* stop writers from using the disks */
for (c = 0; c < children; c++) {
if (vml[c] != NULL)
vml[c]->vdev_offline = B_TRUE;
}
vdev_reopen(spa->spa_root_vdev);
/*
* Temporarily record the splitting vdevs in the spa config. This
* will disappear once the config is regenerated.
*/
VERIFY(nvlist_alloc(&nvl, NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_uint64_array(nvl, ZPOOL_CONFIG_SPLIT_LIST,
glist, children) == 0);
kmem_free(glist, children * sizeof (uint64_t));
mutex_enter(&spa->spa_props_lock);
VERIFY(nvlist_add_nvlist(spa->spa_config, ZPOOL_CONFIG_SPLIT,
nvl) == 0);
mutex_exit(&spa->spa_props_lock);
spa->spa_config_splitting = nvl;
vdev_config_dirty(spa->spa_root_vdev);
/* configure and create the new pool */
VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME, newname) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE,
exp ? POOL_STATE_EXPORTED : POOL_STATE_ACTIVE) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_VERSION,
spa_version(spa)) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG,
spa->spa_config_txg) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID,
spa_generate_guid(NULL)) == 0);
(void) nvlist_lookup_string(props,
zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot);
/* add the new pool to the namespace */
newspa = spa_add(newname, config, altroot);
newspa->spa_config_txg = spa->spa_config_txg;
spa_set_log_state(newspa, SPA_LOG_CLEAR);
/* release the spa config lock, retaining the namespace lock */
spa_vdev_config_exit(spa, NULL, txg, 0, FTAG);
if (zio_injection_enabled)
zio_handle_panic_injection(spa, FTAG, 1);
spa_activate(newspa, spa_mode_global);
spa_async_suspend(newspa);
/* create the new pool from the disks of the original pool */
error = spa_load(newspa, SPA_LOAD_IMPORT, SPA_IMPORT_ASSEMBLE, B_TRUE);
if (error)
goto out;
/* if that worked, generate a real config for the new pool */
if (newspa->spa_root_vdev != NULL) {
VERIFY(nvlist_alloc(&newspa->spa_config_splitting,
NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_uint64(newspa->spa_config_splitting,
ZPOOL_CONFIG_SPLIT_GUID, spa_guid(spa)) == 0);
spa_config_set(newspa, spa_config_generate(newspa, NULL, -1ULL,
B_TRUE));
}
/* set the props */
if (props != NULL) {
spa_configfile_set(newspa, props, B_FALSE);
error = spa_prop_set(newspa, props);
if (error)
goto out;
}
/* flush everything */
txg = spa_vdev_config_enter(newspa);
vdev_config_dirty(newspa->spa_root_vdev);
(void) spa_vdev_config_exit(newspa, NULL, txg, 0, FTAG);
if (zio_injection_enabled)
zio_handle_panic_injection(spa, FTAG, 2);
spa_async_resume(newspa);
/* finally, update the original pool's config */
txg = spa_vdev_config_enter(spa);
tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error != 0)
dmu_tx_abort(tx);
for (c = 0; c < children; c++) {
if (vml[c] != NULL) {
vdev_split(vml[c]);
if (error == 0)
spa_history_log_internal(LOG_POOL_VDEV_DETACH,
spa, tx, "vdev=%s",
vml[c]->vdev_path);
vdev_free(vml[c]);
}
}
vdev_config_dirty(spa->spa_root_vdev);
spa->spa_config_splitting = NULL;
nvlist_free(nvl);
if (error == 0)
dmu_tx_commit(tx);
(void) spa_vdev_exit(spa, NULL, txg, 0);
if (zio_injection_enabled)
zio_handle_panic_injection(spa, FTAG, 3);
/* split is complete; log a history record */
spa_history_log_internal(LOG_POOL_SPLIT, newspa, NULL,
"split new pool %s from pool %s", newname, spa_name(spa));
kmem_free(vml, children * sizeof (vdev_t *));
/* if we're not going to mount the filesystems in userland, export */
if (exp)
error = spa_export_common(newname, POOL_STATE_EXPORTED, NULL,
B_FALSE, B_FALSE);
return (error);
out:
spa_unload(newspa);
spa_deactivate(newspa);
spa_remove(newspa);
txg = spa_vdev_config_enter(spa);
/* re-online all offlined disks */
for (c = 0; c < children; c++) {
if (vml[c] != NULL)
vml[c]->vdev_offline = B_FALSE;
}
vdev_reopen(spa->spa_root_vdev);
nvlist_free(spa->spa_config_splitting);
spa->spa_config_splitting = NULL;
(void) spa_vdev_exit(spa, NULL, txg, error);
kmem_free(vml, children * sizeof (vdev_t *));
return (error);
}
static nvlist_t *
spa_nvlist_lookup_by_guid(nvlist_t **nvpp, int count, uint64_t target_guid)
{
int i;
for (i = 0; i < count; i++) {
uint64_t guid;
VERIFY(nvlist_lookup_uint64(nvpp[i], ZPOOL_CONFIG_GUID,
&guid) == 0);
if (guid == target_guid)
return (nvpp[i]);
}
return (NULL);
}
static void
spa_vdev_remove_aux(nvlist_t *config, char *name, nvlist_t **dev, int count,
nvlist_t *dev_to_remove)
{
nvlist_t **newdev = NULL;
int i, j;
if (count > 1)
newdev = kmem_alloc((count - 1) * sizeof (void *), KM_SLEEP);
for (i = 0, j = 0; i < count; i++) {
if (dev[i] == dev_to_remove)
continue;
VERIFY(nvlist_dup(dev[i], &newdev[j++], KM_SLEEP) == 0);
}
VERIFY(nvlist_remove(config, name, DATA_TYPE_NVLIST_ARRAY) == 0);
VERIFY(nvlist_add_nvlist_array(config, name, newdev, count - 1) == 0);
for (i = 0; i < count - 1; i++)
nvlist_free(newdev[i]);
if (count > 1)
kmem_free(newdev, (count - 1) * sizeof (void *));
}
/*
* Evacuate the device.
*/
static int
spa_vdev_remove_evacuate(spa_t *spa, vdev_t *vd)
{
uint64_t txg;
int error = 0;
ASSERT(MUTEX_HELD(&spa_namespace_lock));
ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0);
ASSERT(vd == vd->vdev_top);
/*
* Evacuate the device. We don't hold the config lock as writer
* since we need to do I/O but we do keep the
* spa_namespace_lock held. Once this completes the device
* should no longer have any blocks allocated on it.
*/
if (vd->vdev_islog) {
if (vd->vdev_stat.vs_alloc != 0)
error = spa_offline_log(spa);
} else {
error = ENOTSUP;
}
if (error)
return (error);
/*
* The evacuation succeeded. Remove any remaining MOS metadata
* associated with this vdev, and wait for these changes to sync.
*/
ASSERT3U(vd->vdev_stat.vs_alloc, ==, 0);
txg = spa_vdev_config_enter(spa);
vd->vdev_removing = B_TRUE;
vdev_dirty(vd, 0, NULL, txg);
vdev_config_dirty(vd);
spa_vdev_config_exit(spa, NULL, txg, 0, FTAG);
return (0);
}
/*
* Complete the removal by cleaning up the namespace.
*/
static void
spa_vdev_remove_from_namespace(spa_t *spa, vdev_t *vd)
{
vdev_t *rvd = spa->spa_root_vdev;
uint64_t id = vd->vdev_id;
boolean_t last_vdev = (id == (rvd->vdev_children - 1));
ASSERT(MUTEX_HELD(&spa_namespace_lock));
ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);
ASSERT(vd == vd->vdev_top);
/*
* Only remove any devices which are empty.
*/
if (vd->vdev_stat.vs_alloc != 0)
return;
(void) vdev_label_init(vd, 0, VDEV_LABEL_REMOVE);
if (list_link_active(&vd->vdev_state_dirty_node))
vdev_state_clean(vd);
if (list_link_active(&vd->vdev_config_dirty_node))
vdev_config_clean(vd);
vdev_free(vd);
if (last_vdev) {
vdev_compact_children(rvd);
} else {
vd = vdev_alloc_common(spa, id, 0, &vdev_hole_ops);
vdev_add_child(rvd, vd);
}
vdev_config_dirty(rvd);
/*
* Reassess the health of our root vdev.
*/
vdev_reopen(rvd);
}
/*
* Remove a device from the pool -
*
* Removing a device from the vdev namespace requires several steps
* and can take a significant amount of time. As a result we use
* the spa_vdev_config_[enter/exit] functions which allow us to
* grab and release the spa_config_lock while still holding the namespace
* lock. During each step the configuration is synced out.
*/
/*
* Remove a device from the pool. Currently, this supports removing only hot
* spares, slogs, and level 2 ARC devices.
*/
int
spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare)
{
vdev_t *vd;
metaslab_group_t *mg;
nvlist_t **spares, **l2cache, *nv;
uint64_t txg = 0;
uint_t nspares, nl2cache;
int error = 0;
boolean_t locked = MUTEX_HELD(&spa_namespace_lock);
if (!locked)
txg = spa_vdev_enter(spa);
vd = spa_lookup_by_guid(spa, guid, B_FALSE);
if (spa->spa_spares.sav_vdevs != NULL &&
nvlist_lookup_nvlist_array(spa->spa_spares.sav_config,
ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0 &&
(nv = spa_nvlist_lookup_by_guid(spares, nspares, guid)) != NULL) {
/*
* Only remove the hot spare if it's not currently in use
* in this pool.
*/
if (vd == NULL || unspare) {
spa_vdev_remove_aux(spa->spa_spares.sav_config,
ZPOOL_CONFIG_SPARES, spares, nspares, nv);
spa_load_spares(spa);
spa->spa_spares.sav_sync = B_TRUE;
} else {
error = EBUSY;
}
} else if (spa->spa_l2cache.sav_vdevs != NULL &&
nvlist_lookup_nvlist_array(spa->spa_l2cache.sav_config,
ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0 &&
(nv = spa_nvlist_lookup_by_guid(l2cache, nl2cache, guid)) != NULL) {
/*
* Cache devices can always be removed.
*/
spa_vdev_remove_aux(spa->spa_l2cache.sav_config,
ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache, nv);
spa_load_l2cache(spa);
spa->spa_l2cache.sav_sync = B_TRUE;
} else if (vd != NULL && vd->vdev_islog) {
ASSERT(!locked);
ASSERT(vd == vd->vdev_top);
/*
* XXX - Once we have bp-rewrite this should
* become the common case.
*/
mg = vd->vdev_mg;
/*
* Stop allocating from this vdev.
*/
metaslab_group_passivate(mg);
/*
* Wait for the youngest allocations and frees to sync,
* and then wait for the deferral of those frees to finish.
*/
spa_vdev_config_exit(spa, NULL,
txg + TXG_CONCURRENT_STATES + TXG_DEFER_SIZE, 0, FTAG);
/*
* Attempt to evacuate the vdev.
*/
error = spa_vdev_remove_evacuate(spa, vd);
txg = spa_vdev_config_enter(spa);
/*
* If we couldn't evacuate the vdev, unwind.
*/
if (error) {
metaslab_group_activate(mg);
return (spa_vdev_exit(spa, NULL, txg, error));
}
/*
* Clean up the vdev namespace.
*/
spa_vdev_remove_from_namespace(spa, vd);
} else if (vd != NULL) {
/*
* Normal vdevs cannot be removed (yet).
*/
error = ENOTSUP;
} else {
/*
* There is no vdev of any kind with the specified guid.
*/
error = ENOENT;
}
if (!locked)
return (spa_vdev_exit(spa, NULL, txg, error));
return (error);
}
/*
* Find any device that's done replacing, or a vdev marked 'unspare' that's
* current spared, so we can detach it.
*/
static vdev_t *
spa_vdev_resilver_done_hunt(vdev_t *vd)
{
vdev_t *newvd, *oldvd;
int c;
for (c = 0; c < vd->vdev_children; c++) {
oldvd = spa_vdev_resilver_done_hunt(vd->vdev_child[c]);
if (oldvd != NULL)
return (oldvd);
}
/*
* Check for a completed replacement.
*/
if (vd->vdev_ops == &vdev_replacing_ops && vd->vdev_children == 2) {
oldvd = vd->vdev_child[0];
newvd = vd->vdev_child[1];
if (vdev_dtl_empty(newvd, DTL_MISSING) &&
vdev_dtl_empty(newvd, DTL_OUTAGE) &&
!vdev_dtl_required(oldvd))
return (oldvd);
}
/*
* Check for a completed resilver with the 'unspare' flag set.
*/
if (vd->vdev_ops == &vdev_spare_ops && vd->vdev_children == 2) {
newvd = vd->vdev_child[0];
oldvd = vd->vdev_child[1];
if (newvd->vdev_unspare &&
vdev_dtl_empty(newvd, DTL_MISSING) &&
vdev_dtl_empty(newvd, DTL_OUTAGE) &&
!vdev_dtl_required(oldvd)) {
newvd->vdev_unspare = 0;
return (oldvd);
}
}
return (NULL);
}
static void
spa_vdev_resilver_done(spa_t *spa)
{
vdev_t *vd, *pvd, *ppvd;
uint64_t guid, sguid, pguid, ppguid;
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
while ((vd = spa_vdev_resilver_done_hunt(spa->spa_root_vdev)) != NULL) {
pvd = vd->vdev_parent;
ppvd = pvd->vdev_parent;
guid = vd->vdev_guid;
pguid = pvd->vdev_guid;
ppguid = ppvd->vdev_guid;
sguid = 0;
/*
* If we have just finished replacing a hot spared device, then
* we need to detach the parent's first child (the original hot
* spare) as well.
*/
if (ppvd->vdev_ops == &vdev_spare_ops && pvd->vdev_id == 0) {
ASSERT(pvd->vdev_ops == &vdev_replacing_ops);
ASSERT(ppvd->vdev_children == 2);
sguid = ppvd->vdev_child[1]->vdev_guid;
}
spa_config_exit(spa, SCL_ALL, FTAG);
if (spa_vdev_detach(spa, guid, pguid, B_TRUE) != 0)
return;
if (sguid && spa_vdev_detach(spa, sguid, ppguid, B_TRUE) != 0)
return;
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
}
spa_config_exit(spa, SCL_ALL, FTAG);
}
/*
* Update the stored path or FRU for this vdev.
*/
int
spa_vdev_set_common(spa_t *spa, uint64_t guid, const char *value,
boolean_t ispath)
{
vdev_t *vd;
boolean_t sync = B_FALSE;
spa_vdev_state_enter(spa, SCL_ALL);
if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL)
return (spa_vdev_state_exit(spa, NULL, ENOENT));
if (!vd->vdev_ops->vdev_op_leaf)
return (spa_vdev_state_exit(spa, NULL, ENOTSUP));
if (ispath) {
if (strcmp(value, vd->vdev_path) != 0) {
spa_strfree(vd->vdev_path);
vd->vdev_path = spa_strdup(value);
sync = B_TRUE;
}
} else {
if (vd->vdev_fru == NULL) {
vd->vdev_fru = spa_strdup(value);
sync = B_TRUE;
} else if (strcmp(value, vd->vdev_fru) != 0) {
spa_strfree(vd->vdev_fru);
vd->vdev_fru = spa_strdup(value);
sync = B_TRUE;
}
}
return (spa_vdev_state_exit(spa, sync ? vd : NULL, 0));
}
int
spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath)
{
return (spa_vdev_set_common(spa, guid, newpath, B_TRUE));
}
int
spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru)
{
return (spa_vdev_set_common(spa, guid, newfru, B_FALSE));
}
/*
* ==========================================================================
* SPA Scanning
* ==========================================================================
*/
int
spa_scan_stop(spa_t *spa)
{
ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0);
if (dsl_scan_resilvering(spa->spa_dsl_pool))
return (EBUSY);
return (dsl_scan_cancel(spa->spa_dsl_pool));
}
int
spa_scan(spa_t *spa, pool_scan_func_t func)
{
ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0);
if (func >= POOL_SCAN_FUNCS || func == POOL_SCAN_NONE)
return (ENOTSUP);
/*
* If a resilver was requested, but there is no DTL on a
* writeable leaf device, we have nothing to do.
*/
if (func == POOL_SCAN_RESILVER &&
!vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL)) {
spa_async_request(spa, SPA_ASYNC_RESILVER_DONE);
return (0);
}
return (dsl_scan(spa->spa_dsl_pool, func));
}
/*
* ==========================================================================
* SPA async task processing
* ==========================================================================
*/
static void
spa_async_remove(spa_t *spa, vdev_t *vd)
{
int c;
if (vd->vdev_remove_wanted) {
vd->vdev_remove_wanted = B_FALSE;
vd->vdev_delayed_close = B_FALSE;
vdev_set_state(vd, B_FALSE, VDEV_STATE_REMOVED, VDEV_AUX_NONE);
/*
* We want to clear the stats, but we don't want to do a full
* vdev_clear() as that will cause us to throw away
* degraded/faulted state as well as attempt to reopen the
* device, all of which is a waste.
*/
vd->vdev_stat.vs_read_errors = 0;
vd->vdev_stat.vs_write_errors = 0;
vd->vdev_stat.vs_checksum_errors = 0;
vdev_state_dirty(vd->vdev_top);
}
for (c = 0; c < vd->vdev_children; c++)
spa_async_remove(spa, vd->vdev_child[c]);
}
static void
spa_async_probe(spa_t *spa, vdev_t *vd)
{
int c;
if (vd->vdev_probe_wanted) {
vd->vdev_probe_wanted = B_FALSE;
vdev_reopen(vd); /* vdev_open() does the actual probe */
}
for (c = 0; c < vd->vdev_children; c++)
spa_async_probe(spa, vd->vdev_child[c]);
}
static void
spa_async_autoexpand(spa_t *spa, vdev_t *vd)
{
int c;
if (!spa->spa_autoexpand)
return;
for (c = 0; c < vd->vdev_children; c++) {
vdev_t *cvd = vd->vdev_child[c];
spa_async_autoexpand(spa, cvd);
}
if (!vd->vdev_ops->vdev_op_leaf || vd->vdev_physpath == NULL)
return;
spa_event_notify(vd->vdev_spa, vd, FM_EREPORT_ZFS_DEVICE_AUTOEXPAND);
}
static void
spa_async_thread(spa_t *spa)
{
int tasks, i;
ASSERT(spa->spa_sync_on);
mutex_enter(&spa->spa_async_lock);
tasks = spa->spa_async_tasks;
spa->spa_async_tasks = 0;
mutex_exit(&spa->spa_async_lock);
/*
* See if the config needs to be updated.
*/
if (tasks & SPA_ASYNC_CONFIG_UPDATE) {
uint64_t old_space, new_space;
mutex_enter(&spa_namespace_lock);
old_space = metaslab_class_get_space(spa_normal_class(spa));
spa_config_update(spa, SPA_CONFIG_UPDATE_POOL);
new_space = metaslab_class_get_space(spa_normal_class(spa));
mutex_exit(&spa_namespace_lock);
/*
* If the pool grew as a result of the config update,
* then log an internal history event.
*/
if (new_space != old_space) {
spa_history_log_internal(LOG_POOL_VDEV_ONLINE,
spa, NULL,
"pool '%s' size: %llu(+%llu)",
spa_name(spa), new_space, new_space - old_space);
}
}
/*
* See if any devices need to be marked REMOVED.
*/
if (tasks & SPA_ASYNC_REMOVE) {
spa_vdev_state_enter(spa, SCL_NONE);
spa_async_remove(spa, spa->spa_root_vdev);
for (i = 0; i < spa->spa_l2cache.sav_count; i++)
spa_async_remove(spa, spa->spa_l2cache.sav_vdevs[i]);
for (i = 0; i < spa->spa_spares.sav_count; i++)
spa_async_remove(spa, spa->spa_spares.sav_vdevs[i]);
(void) spa_vdev_state_exit(spa, NULL, 0);
}
if ((tasks & SPA_ASYNC_AUTOEXPAND) && !spa_suspended(spa)) {
spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
spa_async_autoexpand(spa, spa->spa_root_vdev);
spa_config_exit(spa, SCL_CONFIG, FTAG);
}
/*
* See if any devices need to be probed.
*/
if (tasks & SPA_ASYNC_PROBE) {
spa_vdev_state_enter(spa, SCL_NONE);
spa_async_probe(spa, spa->spa_root_vdev);
(void) spa_vdev_state_exit(spa, NULL, 0);
}
/*
* If any devices are done replacing, detach them.
*/
if (tasks & SPA_ASYNC_RESILVER_DONE)
spa_vdev_resilver_done(spa);
/*
* Kick off a resilver.
*/
if (tasks & SPA_ASYNC_RESILVER)
dsl_resilver_restart(spa->spa_dsl_pool, 0);
/*
* Let the world know that we're done.
*/
mutex_enter(&spa->spa_async_lock);
spa->spa_async_thread = NULL;
cv_broadcast(&spa->spa_async_cv);
mutex_exit(&spa->spa_async_lock);
thread_exit();
}
void
spa_async_suspend(spa_t *spa)
{
mutex_enter(&spa->spa_async_lock);
spa->spa_async_suspended++;
while (spa->spa_async_thread != NULL)
cv_wait(&spa->spa_async_cv, &spa->spa_async_lock);
mutex_exit(&spa->spa_async_lock);
}
void
spa_async_resume(spa_t *spa)
{
mutex_enter(&spa->spa_async_lock);
ASSERT(spa->spa_async_suspended != 0);
spa->spa_async_suspended--;
mutex_exit(&spa->spa_async_lock);
}
static void
spa_async_dispatch(spa_t *spa)
{
mutex_enter(&spa->spa_async_lock);
if (spa->spa_async_tasks && !spa->spa_async_suspended &&
spa->spa_async_thread == NULL &&
rootdir != NULL && !vn_is_readonly(rootdir))
spa->spa_async_thread = thread_create(NULL, 0,
spa_async_thread, spa, 0, &p0, TS_RUN, maxclsyspri);
mutex_exit(&spa->spa_async_lock);
}
void
spa_async_request(spa_t *spa, int task)
{
zfs_dbgmsg("spa=%s async request task=%u", spa->spa_name, task);
mutex_enter(&spa->spa_async_lock);
spa->spa_async_tasks |= task;
mutex_exit(&spa->spa_async_lock);
}
/*
* ==========================================================================
* SPA syncing routines
* ==========================================================================
*/
static int
bpobj_enqueue_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx)
{
bpobj_t *bpo = arg;
bpobj_enqueue(bpo, bp, tx);
return (0);
}
static int
spa_free_sync_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx)
{
zio_t *zio = arg;
zio_nowait(zio_free_sync(zio, zio->io_spa, dmu_tx_get_txg(tx), bp,
zio->io_flags));
return (0);
}
static void
spa_sync_nvlist(spa_t *spa, uint64_t obj, nvlist_t *nv, dmu_tx_t *tx)
{
char *packed = NULL;
size_t bufsize;
size_t nvsize = 0;
dmu_buf_t *db;
VERIFY(nvlist_size(nv, &nvsize, NV_ENCODE_XDR) == 0);
/*
* Write full (SPA_CONFIG_BLOCKSIZE) blocks of configuration
* information. This avoids the dbuf_will_dirty() path and
* saves us a pre-read to get data we don't actually care about.
*/
bufsize = P2ROUNDUP(nvsize, SPA_CONFIG_BLOCKSIZE);
packed = vmem_alloc(bufsize, KM_SLEEP);
VERIFY(nvlist_pack(nv, &packed, &nvsize, NV_ENCODE_XDR,
KM_SLEEP) == 0);
bzero(packed + nvsize, bufsize - nvsize);
dmu_write(spa->spa_meta_objset, obj, 0, bufsize, packed, tx);
vmem_free(packed, bufsize);
VERIFY(0 == dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db));
dmu_buf_will_dirty(db, tx);
*(uint64_t *)db->db_data = nvsize;
dmu_buf_rele(db, FTAG);
}
static void
spa_sync_aux_dev(spa_t *spa, spa_aux_vdev_t *sav, dmu_tx_t *tx,
const char *config, const char *entry)
{
nvlist_t *nvroot;
nvlist_t **list;
int i;
if (!sav->sav_sync)
return;
/*
* Update the MOS nvlist describing the list of available devices.
* spa_validate_aux() will have already made sure this nvlist is
* valid and the vdevs are labeled appropriately.
*/
if (sav->sav_object == 0) {
sav->sav_object = dmu_object_alloc(spa->spa_meta_objset,
DMU_OT_PACKED_NVLIST, 1 << 14, DMU_OT_PACKED_NVLIST_SIZE,
sizeof (uint64_t), tx);
VERIFY(zap_update(spa->spa_meta_objset,
DMU_POOL_DIRECTORY_OBJECT, entry, sizeof (uint64_t), 1,
&sav->sav_object, tx) == 0);
}
VERIFY(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, KM_SLEEP) == 0);
if (sav->sav_count == 0) {
VERIFY(nvlist_add_nvlist_array(nvroot, config, NULL, 0) == 0);
} else {
list = kmem_alloc(sav->sav_count * sizeof (void *), KM_SLEEP);
for (i = 0; i < sav->sav_count; i++)
list[i] = vdev_config_generate(spa, sav->sav_vdevs[i],
B_FALSE, VDEV_CONFIG_L2CACHE);
VERIFY(nvlist_add_nvlist_array(nvroot, config, list,
sav->sav_count) == 0);
for (i = 0; i < sav->sav_count; i++)
nvlist_free(list[i]);
kmem_free(list, sav->sav_count * sizeof (void *));
}
spa_sync_nvlist(spa, sav->sav_object, nvroot, tx);
nvlist_free(nvroot);
sav->sav_sync = B_FALSE;
}
static void
spa_sync_config_object(spa_t *spa, dmu_tx_t *tx)
{
nvlist_t *config;
if (list_is_empty(&spa->spa_config_dirty_list))
return;
spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
config = spa_config_generate(spa, spa->spa_root_vdev,
dmu_tx_get_txg(tx), B_FALSE);
spa_config_exit(spa, SCL_STATE, FTAG);
if (spa->spa_config_syncing)
nvlist_free(spa->spa_config_syncing);
spa->spa_config_syncing = config;
spa_sync_nvlist(spa, spa->spa_config_object, config, tx);
}
/*
* Set zpool properties.
*/
static void
spa_sync_props(void *arg1, void *arg2, dmu_tx_t *tx)
{
spa_t *spa = arg1;
objset_t *mos = spa->spa_meta_objset;
nvlist_t *nvp = arg2;
nvpair_t *elem;
uint64_t intval;
char *strval;
zpool_prop_t prop;
const char *propname;
zprop_type_t proptype;
mutex_enter(&spa->spa_props_lock);
elem = NULL;
while ((elem = nvlist_next_nvpair(nvp, elem))) {
switch (prop = zpool_name_to_prop(nvpair_name(elem))) {
case ZPOOL_PROP_VERSION:
/*
* Only set version for non-zpool-creation cases
* (set/import). spa_create() needs special care
* for version setting.
*/
if (tx->tx_txg != TXG_INITIAL) {
VERIFY(nvpair_value_uint64(elem,
&intval) == 0);
ASSERT(intval <= SPA_VERSION);
ASSERT(intval >= spa_version(spa));
spa->spa_uberblock.ub_version = intval;
vdev_config_dirty(spa->spa_root_vdev);
}
break;
case ZPOOL_PROP_ALTROOT:
/*
* 'altroot' is a non-persistent property. It should
* have been set temporarily at creation or import time.
*/
ASSERT(spa->spa_root != NULL);
break;
case ZPOOL_PROP_CACHEFILE:
/*
* 'cachefile' is also a non-persisitent property.
*/
break;
default:
/*
* Set pool property values in the poolprops mos object.
*/
if (spa->spa_pool_props_object == 0) {
VERIFY((spa->spa_pool_props_object =
zap_create(mos, DMU_OT_POOL_PROPS,
DMU_OT_NONE, 0, tx)) > 0);
VERIFY(zap_update(mos,
DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_PROPS,
8, 1, &spa->spa_pool_props_object, tx)
== 0);
}
/* normalize the property name */
propname = zpool_prop_to_name(prop);
proptype = zpool_prop_get_type(prop);
if (nvpair_type(elem) == DATA_TYPE_STRING) {
ASSERT(proptype == PROP_TYPE_STRING);
VERIFY(nvpair_value_string(elem, &strval) == 0);
VERIFY(zap_update(mos,
spa->spa_pool_props_object, propname,
1, strlen(strval) + 1, strval, tx) == 0);
} else if (nvpair_type(elem) == DATA_TYPE_UINT64) {
VERIFY(nvpair_value_uint64(elem, &intval) == 0);
if (proptype == PROP_TYPE_INDEX) {
const char *unused;
VERIFY(zpool_prop_index_to_string(
prop, intval, &unused) == 0);
}
VERIFY(zap_update(mos,
spa->spa_pool_props_object, propname,
8, 1, &intval, tx) == 0);
} else {
ASSERT(0); /* not allowed */
}
switch (prop) {
case ZPOOL_PROP_DELEGATION:
spa->spa_delegation = intval;
break;
case ZPOOL_PROP_BOOTFS:
spa->spa_bootfs = intval;
break;
case ZPOOL_PROP_FAILUREMODE:
spa->spa_failmode = intval;
break;
case ZPOOL_PROP_AUTOEXPAND:
spa->spa_autoexpand = intval;
if (tx->tx_txg != TXG_INITIAL)
spa_async_request(spa,
SPA_ASYNC_AUTOEXPAND);
break;
case ZPOOL_PROP_DEDUPDITTO:
spa->spa_dedup_ditto = intval;
break;
default:
break;
}
}
/* log internal history if this is not a zpool create */
if (spa_version(spa) >= SPA_VERSION_ZPOOL_HISTORY &&
tx->tx_txg != TXG_INITIAL) {
spa_history_log_internal(LOG_POOL_PROPSET,
spa, tx, "%s %lld %s",
nvpair_name(elem), intval, spa_name(spa));
}
}
mutex_exit(&spa->spa_props_lock);
}
/*
* Perform one-time upgrade on-disk changes. spa_version() does not
* reflect the new version this txg, so there must be no changes this
* txg to anything that the upgrade code depends on after it executes.
* Therefore this must be called after dsl_pool_sync() does the sync
* tasks.
*/
static void
spa_sync_upgrades(spa_t *spa, dmu_tx_t *tx)
{
dsl_pool_t *dp = spa->spa_dsl_pool;
ASSERT(spa->spa_sync_pass == 1);
if (spa->spa_ubsync.ub_version < SPA_VERSION_ORIGIN &&
spa->spa_uberblock.ub_version >= SPA_VERSION_ORIGIN) {
dsl_pool_create_origin(dp, tx);
/* Keeping the origin open increases spa_minref */
spa->spa_minref += 3;
}
if (spa->spa_ubsync.ub_version < SPA_VERSION_NEXT_CLONES &&
spa->spa_uberblock.ub_version >= SPA_VERSION_NEXT_CLONES) {
dsl_pool_upgrade_clones(dp, tx);
}
if (spa->spa_ubsync.ub_version < SPA_VERSION_DIR_CLONES &&
spa->spa_uberblock.ub_version >= SPA_VERSION_DIR_CLONES) {
dsl_pool_upgrade_dir_clones(dp, tx);
/* Keeping the freedir open increases spa_minref */
spa->spa_minref += 3;
}
}
/*
* Sync the specified transaction group. New blocks may be dirtied as
* part of the process, so we iterate until it converges.
*/
void
spa_sync(spa_t *spa, uint64_t txg)
{
dsl_pool_t *dp = spa->spa_dsl_pool;
objset_t *mos = spa->spa_meta_objset;
bpobj_t *defer_bpo = &spa->spa_deferred_bpobj;
bplist_t *free_bpl = &spa->spa_free_bplist[txg & TXG_MASK];
vdev_t *rvd = spa->spa_root_vdev;
vdev_t *vd;
dmu_tx_t *tx;
int error;
int c;
/*
* Lock out configuration changes.
*/
spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
spa->spa_syncing_txg = txg;
spa->spa_sync_pass = 0;
/*
* If there are any pending vdev state changes, convert them
* into config changes that go out with this transaction group.
*/
spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
while (list_head(&spa->spa_state_dirty_list) != NULL) {
/*
* We need the write lock here because, for aux vdevs,
* calling vdev_config_dirty() modifies sav_config.
* This is ugly and will become unnecessary when we
* eliminate the aux vdev wart by integrating all vdevs
* into the root vdev tree.
*/
spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG);
spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_WRITER);
while ((vd = list_head(&spa->spa_state_dirty_list)) != NULL) {
vdev_state_clean(vd);
vdev_config_dirty(vd);
}
spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG);
spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER);
}
spa_config_exit(spa, SCL_STATE, FTAG);
tx = dmu_tx_create_assigned(dp, txg);
/*
* If we are upgrading to SPA_VERSION_RAIDZ_DEFLATE this txg,
* set spa_deflate if we have no raid-z vdevs.
*/
if (spa->spa_ubsync.ub_version < SPA_VERSION_RAIDZ_DEFLATE &&
spa->spa_uberblock.ub_version >= SPA_VERSION_RAIDZ_DEFLATE) {
int i;
for (i = 0; i < rvd->vdev_children; i++) {
vd = rvd->vdev_child[i];
if (vd->vdev_deflate_ratio != SPA_MINBLOCKSIZE)
break;
}
if (i == rvd->vdev_children) {
spa->spa_deflate = TRUE;
VERIFY(0 == zap_add(spa->spa_meta_objset,
DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DEFLATE,
sizeof (uint64_t), 1, &spa->spa_deflate, tx));
}
}
/*
* If anything has changed in this txg, or if someone is waiting
* for this txg to sync (eg, spa_vdev_remove()), push the
* deferred frees from the previous txg. If not, leave them
* alone so that we don't generate work on an otherwise idle
* system.
*/
if (!txg_list_empty(&dp->dp_dirty_datasets, txg) ||
!txg_list_empty(&dp->dp_dirty_dirs, txg) ||
!txg_list_empty(&dp->dp_sync_tasks, txg) ||
((dsl_scan_active(dp->dp_scan) ||
txg_sync_waiting(dp)) && !spa_shutting_down(spa))) {
zio_t *zio = zio_root(spa, NULL, NULL, 0);
VERIFY3U(bpobj_iterate(defer_bpo,
spa_free_sync_cb, zio, tx), ==, 0);
VERIFY3U(zio_wait(zio), ==, 0);
}
/*
* Iterate to convergence.
*/
do {
int pass = ++spa->spa_sync_pass;
spa_sync_config_object(spa, tx);
spa_sync_aux_dev(spa, &spa->spa_spares, tx,
ZPOOL_CONFIG_SPARES, DMU_POOL_SPARES);
spa_sync_aux_dev(spa, &spa->spa_l2cache, tx,
ZPOOL_CONFIG_L2CACHE, DMU_POOL_L2CACHE);
spa_errlog_sync(spa, txg);
dsl_pool_sync(dp, txg);
if (pass <= SYNC_PASS_DEFERRED_FREE) {
zio_t *zio = zio_root(spa, NULL, NULL, 0);
bplist_iterate(free_bpl, spa_free_sync_cb,
zio, tx);
VERIFY(zio_wait(zio) == 0);
} else {
bplist_iterate(free_bpl, bpobj_enqueue_cb,
defer_bpo, tx);
}
ddt_sync(spa, txg);
dsl_scan_sync(dp, tx);
while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, txg)))
vdev_sync(vd, txg);
if (pass == 1)
spa_sync_upgrades(spa, tx);
} while (dmu_objset_is_dirty(mos, txg));
/*
* Rewrite the vdev configuration (which includes the uberblock)
* to commit the transaction group.
*
* If there are no dirty vdevs, we sync the uberblock to a few
* random top-level vdevs that are known to be visible in the
* config cache (see spa_vdev_add() for a complete description).
* If there *are* dirty vdevs, sync the uberblock to all vdevs.
*/
for (;;) {
/*
* We hold SCL_STATE to prevent vdev open/close/etc.
* while we're attempting to write the vdev labels.
*/
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];
int svdcount = 0;
int children = rvd->vdev_children;
int c0 = spa_get_random(children);
for (c = 0; c < children; c++) {
vd = rvd->vdev_child[(c0 + c) % children];
if (vd->vdev_ms_array == 0 || vd->vdev_islog)
continue;
svd[svdcount++] = vd;
if (svdcount == SPA_DVAS_PER_BP)
break;
}
error = vdev_config_sync(svd, svdcount, txg, B_FALSE);
if (error != 0)
error = vdev_config_sync(svd, svdcount, txg,
B_TRUE);
} else {
error = vdev_config_sync(rvd->vdev_child,
rvd->vdev_children, txg, B_FALSE);
if (error != 0)
error = vdev_config_sync(rvd->vdev_child,
rvd->vdev_children, txg, B_TRUE);
}
spa_config_exit(spa, SCL_STATE, FTAG);
if (error == 0)
break;
zio_suspend(spa, NULL);
zio_resume_wait(spa);
}
dmu_tx_commit(tx);
/*
* Clear the dirty config list.
*/
while ((vd = list_head(&spa->spa_config_dirty_list)) != NULL)
vdev_config_clean(vd);
/*
* Now that the new config has synced transactionally,
* let it become visible to the config cache.
*/
if (spa->spa_config_syncing != NULL) {
spa_config_set(spa, spa->spa_config_syncing);
spa->spa_config_txg = txg;
spa->spa_config_syncing = NULL;
}
spa->spa_ubsync = spa->spa_uberblock;
dsl_pool_sync_done(dp, txg);
/*
* Update usable space statistics.
*/
while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, TXG_CLEAN(txg))))
vdev_sync_done(vd, txg);
spa_update_dspace(spa);
/*
* It had better be the case that we didn't dirty anything
* since vdev_config_sync().
*/
ASSERT(txg_list_empty(&dp->dp_dirty_datasets, txg));
ASSERT(txg_list_empty(&dp->dp_dirty_dirs, txg));
ASSERT(txg_list_empty(&spa->spa_vdev_txg_list, txg));
spa->spa_sync_pass = 0;
spa_config_exit(spa, SCL_CONFIG, FTAG);
spa_handle_ignored_writes(spa);
/*
* If any async tasks have been requested, kick them off.
*/
spa_async_dispatch(spa);
}
/*
* Sync all pools. We don't want to hold the namespace lock across these
* operations, so we take a reference on the spa_t and drop the lock during the
* sync.
*/
void
spa_sync_allpools(void)
{
spa_t *spa = NULL;
mutex_enter(&spa_namespace_lock);
while ((spa = spa_next(spa)) != NULL) {
if (spa_state(spa) != POOL_STATE_ACTIVE || spa_suspended(spa))
continue;
spa_open_ref(spa, FTAG);
mutex_exit(&spa_namespace_lock);
txg_wait_synced(spa_get_dsl(spa), 0);
mutex_enter(&spa_namespace_lock);
spa_close(spa, FTAG);
}
mutex_exit(&spa_namespace_lock);
}
/*
* ==========================================================================
* Miscellaneous routines
* ==========================================================================
*/
/*
* Remove all pools in the system.
*/
void
spa_evict_all(void)
{
spa_t *spa;
/*
* Remove all cached state. All pools should be closed now,
* so every spa in the AVL tree should be unreferenced.
*/
mutex_enter(&spa_namespace_lock);
while ((spa = spa_next(NULL)) != NULL) {
/*
* Stop async tasks. The async thread may need to detach
* a device that's been replaced, which requires grabbing
* spa_namespace_lock, so we must drop it here.
*/
spa_open_ref(spa, FTAG);
mutex_exit(&spa_namespace_lock);
spa_async_suspend(spa);
mutex_enter(&spa_namespace_lock);
spa_close(spa, FTAG);
if (spa->spa_state != POOL_STATE_UNINITIALIZED) {
spa_unload(spa);
spa_deactivate(spa);
}
spa_remove(spa);
}
mutex_exit(&spa_namespace_lock);
}
vdev_t *
spa_lookup_by_guid(spa_t *spa, uint64_t guid, boolean_t aux)
{
vdev_t *vd;
int i;
if ((vd = vdev_lookup_by_guid(spa->spa_root_vdev, guid)) != NULL)
return (vd);
if (aux) {
for (i = 0; i < spa->spa_l2cache.sav_count; i++) {
vd = spa->spa_l2cache.sav_vdevs[i];
if (vd->vdev_guid == guid)
return (vd);
}
for (i = 0; i < spa->spa_spares.sav_count; i++) {
vd = spa->spa_spares.sav_vdevs[i];
if (vd->vdev_guid == guid)
return (vd);
}
}
return (NULL);
}
void
spa_upgrade(spa_t *spa, uint64_t version)
{
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
/*
* This should only be called for a non-faulted pool, and since a
* future version would result in an unopenable pool, this shouldn't be
* possible.
*/
ASSERT(spa->spa_uberblock.ub_version <= SPA_VERSION);
ASSERT(version >= spa->spa_uberblock.ub_version);
spa->spa_uberblock.ub_version = version;
vdev_config_dirty(spa->spa_root_vdev);
spa_config_exit(spa, SCL_ALL, FTAG);
txg_wait_synced(spa_get_dsl(spa), 0);
}
boolean_t
spa_has_spare(spa_t *spa, uint64_t guid)
{
int i;
uint64_t spareguid;
spa_aux_vdev_t *sav = &spa->spa_spares;
for (i = 0; i < sav->sav_count; i++)
if (sav->sav_vdevs[i]->vdev_guid == guid)
return (B_TRUE);
for (i = 0; i < sav->sav_npending; i++) {
if (nvlist_lookup_uint64(sav->sav_pending[i], ZPOOL_CONFIG_GUID,
&spareguid) == 0 && spareguid == guid)
return (B_TRUE);
}
return (B_FALSE);
}
/*
* Check if a pool has an active shared spare device.
* Note: reference count of an active spare is 2, as a spare and as a replace
*/
static boolean_t
spa_has_active_shared_spare(spa_t *spa)
{
int i, refcnt;
uint64_t pool;
spa_aux_vdev_t *sav = &spa->spa_spares;
for (i = 0; i < sav->sav_count; i++) {
if (spa_spare_exists(sav->sav_vdevs[i]->vdev_guid, &pool,
&refcnt) && pool != 0ULL && pool == spa_guid(spa) &&
refcnt > 2)
return (B_TRUE);
}
return (B_FALSE);
}
/*
* Post a FM_EREPORT_ZFS_* event from sys/fm/fs/zfs.h. The payload will be
* filled in from the spa and (optionally) the vdev. This doesn't do anything
* in the userland libzpool, as we don't want consumers to misinterpret ztest
* or zdb as real changes.
*/
void
spa_event_notify(spa_t *spa, vdev_t *vd, const char *name)
{
#ifdef _KERNEL
zfs_ereport_post(name, spa, vd, NULL, 0, 0);
#endif
}
#if defined(_KERNEL) && defined(HAVE_SPL)
/* state manipulation functions */
EXPORT_SYMBOL(spa_open);
EXPORT_SYMBOL(spa_open_rewind);
EXPORT_SYMBOL(spa_get_stats);
EXPORT_SYMBOL(spa_create);
EXPORT_SYMBOL(spa_import_rootpool);
EXPORT_SYMBOL(spa_import);
EXPORT_SYMBOL(spa_import_verbatim);
EXPORT_SYMBOL(spa_tryimport);
EXPORT_SYMBOL(spa_destroy);
EXPORT_SYMBOL(spa_export);
EXPORT_SYMBOL(spa_reset);
EXPORT_SYMBOL(spa_async_request);
EXPORT_SYMBOL(spa_async_suspend);
EXPORT_SYMBOL(spa_async_resume);
EXPORT_SYMBOL(spa_inject_addref);
EXPORT_SYMBOL(spa_inject_delref);
EXPORT_SYMBOL(spa_scan_stat_init);
EXPORT_SYMBOL(spa_scan_get_stats);
/* device maniion */
EXPORT_SYMBOL(spa_vdev_add);
EXPORT_SYMBOL(spa_vdev_attach);
EXPORT_SYMBOL(spa_vdev_detach);
EXPORT_SYMBOL(spa_vdev_remove);
EXPORT_SYMBOL(spa_vdev_setpath);
EXPORT_SYMBOL(spa_vdev_setfru);
EXPORT_SYMBOL(spa_vdev_split_mirror);
/* spare statech is global across all pools) */
EXPORT_SYMBOL(spa_spare_add);
EXPORT_SYMBOL(spa_spare_remove);
EXPORT_SYMBOL(spa_spare_exists);
EXPORT_SYMBOL(spa_spare_activate);
/* L2ARC statech is global across all pools) */
EXPORT_SYMBOL(spa_l2cache_add);
EXPORT_SYMBOL(spa_l2cache_remove);
EXPORT_SYMBOL(spa_l2cache_exists);
EXPORT_SYMBOL(spa_l2cache_activate);
EXPORT_SYMBOL(spa_l2cache_drop);
/* scanning */
EXPORT_SYMBOL(spa_scan);
EXPORT_SYMBOL(spa_scan_stop);
/* spa syncing */
EXPORT_SYMBOL(spa_sync); /* only for DMU use */
EXPORT_SYMBOL(spa_sync_allpools);
/* properties */
EXPORT_SYMBOL(spa_prop_set);
EXPORT_SYMBOL(spa_prop_get);
EXPORT_SYMBOL(spa_prop_clear_bootfs);
/* asynchronous event notification */
EXPORT_SYMBOL(spa_event_notify);
#endif