uint64_t
dsl_dir_create_sync(dsl_pool_t *dp, dsl_dir_t *pds, const char *name,
dmu_tx_t *tx)
{
objset_t *mos = dp->dp_meta_objset;
uint64_t ddobj;
dsl_dir_phys_t *dsphys;
dmu_buf_t *dbuf;
ddobj = dmu_object_alloc(mos, DMU_OT_DSL_DIR, 0,
DMU_OT_DSL_DIR, sizeof (dsl_dir_phys_t), tx);
if (pds) {
VERIFY(0 == zap_add(mos, pds->dd_phys->dd_child_dir_zapobj,
name, sizeof (uint64_t), 1, &ddobj, tx));
} else {
/* it's the root dir */
VERIFY(0 == zap_add(mos, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_ROOT_DATASET, sizeof (uint64_t), 1, &ddobj, tx));
}
VERIFY(0 == dmu_bonus_hold(mos, ddobj, FTAG, &dbuf));
dmu_buf_will_dirty(dbuf, tx);
dsphys = dbuf->db_data;
dsphys->dd_creation_time = gethrestime_sec();
if (pds)
dsphys->dd_parent_obj = pds->dd_object;
dsphys->dd_props_zapobj = zap_create(mos,
DMU_OT_DSL_PROPS, DMU_OT_NONE, 0, tx);
dsphys->dd_child_dir_zapobj = zap_create(mos,
DMU_OT_DSL_DIR_CHILD_MAP, DMU_OT_NONE, 0, tx);
dmu_buf_rele(dbuf, FTAG);
return (ddobj);
}
int
zfs_set_userquota(zfs_sb_t *zsb, zfs_userquota_prop_t type,
const char *domain, uint64_t rid, uint64_t quota)
{
char buf[32];
int err;
dmu_tx_t *tx;
uint64_t *objp;
boolean_t fuid_dirtied;
if (type != ZFS_PROP_USERQUOTA && type != ZFS_PROP_GROUPQUOTA)
return (SET_ERROR(EINVAL));
if (zsb->z_version < ZPL_VERSION_USERSPACE)
return (SET_ERROR(ENOTSUP));
objp = (type == ZFS_PROP_USERQUOTA) ? &zsb->z_userquota_obj :
&zsb->z_groupquota_obj;
err = id_to_fuidstr(zsb, domain, rid, buf, B_TRUE);
if (err)
return (err);
fuid_dirtied = zsb->z_fuid_dirty;
tx = dmu_tx_create(zsb->z_os);
dmu_tx_hold_zap(tx, *objp ? *objp : DMU_NEW_OBJECT, B_TRUE, NULL);
if (*objp == 0) {
dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_TRUE,
zfs_userquota_prop_prefixes[type]);
}
if (fuid_dirtied)
zfs_fuid_txhold(zsb, tx);
err = dmu_tx_assign(tx, TXG_WAIT);
if (err) {
dmu_tx_abort(tx);
return (err);
}
mutex_enter(&zsb->z_lock);
if (*objp == 0) {
*objp = zap_create(zsb->z_os, DMU_OT_USERGROUP_QUOTA,
DMU_OT_NONE, 0, tx);
VERIFY(0 == zap_add(zsb->z_os, MASTER_NODE_OBJ,
zfs_userquota_prop_prefixes[type], 8, 1, objp, tx));
}
mutex_exit(&zsb->z_lock);
if (quota == 0) {
err = zap_remove(zsb->z_os, *objp, buf, tx);
if (err == ENOENT)
err = 0;
} else {
err = zap_update(zsb->z_os, *objp, buf, 8, 1, "a, tx);
}
ASSERT(err == 0);
if (fuid_dirtied)
zfs_fuid_sync(zsb, tx);
dmu_tx_commit(tx);
return (err);
}
static void
sa_attr_register_sync(sa_handle_t *hdl, dmu_tx_t *tx)
{
uint64_t attr_value = 0;
sa_os_t *sa = hdl->sa_os->os_sa;
sa_attr_table_t *tb = sa->sa_attr_table;
int i;
mutex_enter(&sa->sa_lock);
if (!sa->sa_need_attr_registration || sa->sa_master_obj == NULL) {
mutex_exit(&sa->sa_lock);
return;
}
if (sa->sa_reg_attr_obj == NULL) {
sa->sa_reg_attr_obj = zap_create(hdl->sa_os,
DMU_OT_SA_ATTR_REGISTRATION, DMU_OT_NONE, 0, tx);
VERIFY(zap_add(hdl->sa_os, sa->sa_master_obj,
SA_REGISTRY, 8, 1, &sa->sa_reg_attr_obj, tx) == 0);
}
for (i = 0; i != sa->sa_num_attrs; i++) {
if (sa->sa_attr_table[i].sa_registered)
continue;
ATTR_ENCODE(attr_value, tb[i].sa_attr, tb[i].sa_length,
tb[i].sa_byteswap);
VERIFY(0 == zap_update(hdl->sa_os, sa->sa_reg_attr_obj,
tb[i].sa_name, 8, 1, &attr_value, tx));
tb[i].sa_registered = B_TRUE;
}
sa->sa_need_attr_registration = B_FALSE;
mutex_exit(&sa->sa_lock);
}
void
dsl_dataset_user_hold_sync_one(dsl_dataset_t *ds, const char *htag,
minor_t minor, uint64_t now, dmu_tx_t *tx)
{
dsl_pool_t *dp = ds->ds_dir->dd_pool;
objset_t *mos = dp->dp_meta_objset;
uint64_t zapobj;
mutex_enter(&ds->ds_lock);
if (ds->ds_phys->ds_userrefs_obj == 0) {
/*
* This is the first user hold for this dataset. Create
* the userrefs zap object.
*/
dmu_buf_will_dirty(ds->ds_dbuf, tx);
zapobj = ds->ds_phys->ds_userrefs_obj =
zap_create(mos, DMU_OT_USERREFS, DMU_OT_NONE, 0, tx);
} else {
zapobj = ds->ds_phys->ds_userrefs_obj;
}
ds->ds_userrefs++;
mutex_exit(&ds->ds_lock);
VERIFY0(zap_add(mos, zapobj, htag, 8, 1, &now, tx));
if (minor != 0) {
VERIFY0(dsl_pool_user_hold(dp, ds->ds_object,
htag, now, tx));
dsl_register_onexit_hold_cleanup(ds, htag, minor);
}
spa_history_log_internal_ds(ds, "hold", tx,
"tag=%s temp=%d refs=%llu",
htag, minor != 0, ds->ds_userrefs);
}
static sa_lot_t *
sa_add_layout_entry(objset_t *os, sa_attr_type_t *attrs, int attr_count,
uint64_t lot_num, uint64_t hash, boolean_t zapadd, dmu_tx_t *tx)
{
sa_os_t *sa = os->os_sa;
sa_lot_t *tb, *findtb;
int i;
avl_index_t loc;
ASSERT(MUTEX_HELD(&sa->sa_lock));
tb = kmem_zalloc(sizeof (sa_lot_t), KM_SLEEP);
tb->lot_attr_count = attr_count;
tb->lot_attrs = kmem_alloc(sizeof (sa_attr_type_t) * attr_count,
KM_SLEEP);
bcopy(attrs, tb->lot_attrs, sizeof (sa_attr_type_t) * attr_count);
tb->lot_num = lot_num;
tb->lot_hash = hash;
tb->lot_instance = 0;
if (zapadd) {
char attr_name[8];
if (sa->sa_layout_attr_obj == 0) {
sa->sa_layout_attr_obj = zap_create(os,
DMU_OT_SA_ATTR_LAYOUTS, DMU_OT_NONE, 0, tx);
VERIFY(zap_add(os, sa->sa_master_obj, SA_LAYOUTS, 8, 1,
&sa->sa_layout_attr_obj, tx) == 0);
}
(void) snprintf(attr_name, sizeof (attr_name),
"%d", (int)lot_num);
VERIFY(0 == zap_update(os, os->os_sa->sa_layout_attr_obj,
attr_name, 2, attr_count, attrs, tx));
}
list_create(&tb->lot_idx_tab, sizeof (sa_idx_tab_t),
offsetof(sa_idx_tab_t, sa_next));
for (i = 0; i != attr_count; i++) {
if (sa->sa_attr_table[tb->lot_attrs[i]].sa_length == 0)
tb->lot_var_sizes++;
}
avl_add(&sa->sa_layout_num_tree, tb);
/* verify we don't have a hash collision */
if ((findtb = avl_find(&sa->sa_layout_hash_tree, tb, &loc)) != NULL) {
for (; findtb && findtb->lot_hash == hash;
findtb = AVL_NEXT(&sa->sa_layout_hash_tree, findtb)) {
if (findtb->lot_instance != tb->lot_instance)
break;
tb->lot_instance++;
}
}
avl_add(&sa->sa_layout_hash_tree, tb);
return (tb);
}
static void
dsl_deleg_set_sync(void *arg1, void *arg2, cred_t *cr, dmu_tx_t *tx)
{
dsl_dir_t *dd = arg1;
nvlist_t *nvp = arg2;
objset_t *mos = dd->dd_pool->dp_meta_objset;
nvpair_t *whopair = NULL;
uint64_t zapobj = dd->dd_phys->dd_deleg_zapobj;
if (zapobj == 0) {
dmu_buf_will_dirty(dd->dd_dbuf, tx);
zapobj = dd->dd_phys->dd_deleg_zapobj = zap_create(mos,
DMU_OT_DSL_PERMS, DMU_OT_NONE, 0, tx);
}
while (whopair = nvlist_next_nvpair(nvp, whopair)) {
const char *whokey = nvpair_name(whopair);
nvlist_t *perms;
nvpair_t *permpair = NULL;
uint64_t jumpobj;
VERIFY(nvpair_value_nvlist(whopair, &perms) == 0);
if (zap_lookup(mos, zapobj, whokey, 8, 1, &jumpobj) != 0) {
jumpobj = zap_create(mos, DMU_OT_DSL_PERMS,
DMU_OT_NONE, 0, tx);
VERIFY(zap_update(mos, zapobj,
whokey, 8, 1, &jumpobj, tx) == 0);
}
while (permpair = nvlist_next_nvpair(perms, permpair)) {
const char *perm = nvpair_name(permpair);
uint64_t n = 0;
VERIFY(zap_update(mos, jumpobj,
perm, 8, 1, &n, tx) == 0);
spa_history_internal_log(LOG_DS_PERM_UPDATE,
dd->dd_pool->dp_spa, tx, cr,
"%s %s dataset = %llu", whokey, perm,
dd->dd_phys->dd_head_dataset_obj);
}
}
}
static void
copy_create_perms(dsl_dir_t *dd, uint64_t pzapobj,
boolean_t dosets, uint64_t uid, dmu_tx_t *tx)
{
objset_t *mos = dd->dd_pool->dp_meta_objset;
uint64_t jumpobj, pjumpobj;
uint64_t zapobj = dd->dd_phys->dd_deleg_zapobj;
zap_cursor_t zc;
zap_attribute_t za;
char whokey[ZFS_MAX_DELEG_NAME];
zfs_deleg_whokey(whokey,
dosets ? ZFS_DELEG_CREATE_SETS : ZFS_DELEG_CREATE,
ZFS_DELEG_LOCAL, NULL);
if (zap_lookup(mos, pzapobj, whokey, 8, 1, &pjumpobj) != 0)
return;
if (zapobj == 0) {
dmu_buf_will_dirty(dd->dd_dbuf, tx);
zapobj = dd->dd_phys->dd_deleg_zapobj = zap_create(mos,
DMU_OT_DSL_PERMS, DMU_OT_NONE, 0, tx);
}
zfs_deleg_whokey(whokey,
dosets ? ZFS_DELEG_USER_SETS : ZFS_DELEG_USER,
ZFS_DELEG_LOCAL, &uid);
if (zap_lookup(mos, zapobj, whokey, 8, 1, &jumpobj) == ENOENT) {
jumpobj = zap_create(mos, DMU_OT_DSL_PERMS, DMU_OT_NONE, 0, tx);
VERIFY(zap_add(mos, zapobj, whokey, 8, 1, &jumpobj, tx) == 0);
}
for (zap_cursor_init(&zc, mos, pjumpobj);
zap_cursor_retrieve(&zc, &za) == 0;
zap_cursor_advance(&zc)) {
uint64_t zero = 0;
ASSERT(za.za_integer_length == 8 && za.za_num_integers == 1);
VERIFY(zap_update(mos, jumpobj, za.za_name,
8, 1, &zero, tx) == 0);
}
zap_cursor_fini(&zc);
}
int
zfs_set_version(zfs_sb_t *zsb, uint64_t newvers)
{
int error;
objset_t *os = zsb->z_os;
dmu_tx_t *tx;
if (newvers < ZPL_VERSION_INITIAL || newvers > ZPL_VERSION)
return (SET_ERROR(EINVAL));
if (newvers < zsb->z_version)
return (SET_ERROR(EINVAL));
if (zfs_spa_version_map(newvers) >
spa_version(dmu_objset_spa(zsb->z_os)))
return (SET_ERROR(ENOTSUP));
tx = dmu_tx_create(os);
dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_FALSE, ZPL_VERSION_STR);
if (newvers >= ZPL_VERSION_SA && !zsb->z_use_sa) {
dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_TRUE,
ZFS_SA_ATTRS);
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
}
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
return (error);
}
error = zap_update(os, MASTER_NODE_OBJ, ZPL_VERSION_STR,
8, 1, &newvers, tx);
if (error) {
dmu_tx_commit(tx);
return (error);
}
if (newvers >= ZPL_VERSION_SA && !zsb->z_use_sa) {
uint64_t sa_obj;
ASSERT3U(spa_version(dmu_objset_spa(zsb->z_os)), >=,
SPA_VERSION_SA);
sa_obj = zap_create(os, DMU_OT_SA_MASTER_NODE,
DMU_OT_NONE, 0, tx);
error = zap_add(os, MASTER_NODE_OBJ,
ZFS_SA_ATTRS, 8, 1, &sa_obj, tx);
ASSERT0(error);
VERIFY(0 == sa_set_sa_object(os, sa_obj));
sa_register_update_callback(os, zfs_sa_upgrade);
}
static void
dsl_dataset_user_hold_sync_one_impl(nvlist_t *tmpholds, dsl_dataset_t *ds,
const char *htag, minor_t minor, uint64_t now, dmu_tx_t *tx)
{
dsl_pool_t *dp = ds->ds_dir->dd_pool;
objset_t *mos = dp->dp_meta_objset;
uint64_t zapobj;
ASSERT(RRW_WRITE_HELD(&dp->dp_config_rwlock));
if (ds->ds_phys->ds_userrefs_obj == 0) {
/*
* This is the first user hold for this dataset. Create
* the userrefs zap object.
*/
dmu_buf_will_dirty(ds->ds_dbuf, tx);
zapobj = ds->ds_phys->ds_userrefs_obj =
zap_create(mos, DMU_OT_USERREFS, DMU_OT_NONE, 0, tx);
} else {
zapobj = ds->ds_phys->ds_userrefs_obj;
}
ds->ds_userrefs++;
VERIFY0(zap_add(mos, zapobj, htag, 8, 1, &now, tx));
if (minor != 0) {
char name[MAXNAMELEN];
nvlist_t *tags;
VERIFY0(dsl_pool_user_hold(dp, ds->ds_object,
htag, now, tx));
(void) snprintf(name, sizeof (name), "%llx",
(u_longlong_t)ds->ds_object);
if (nvlist_lookup_nvlist(tmpholds, name, &tags) != 0) {
VERIFY0(nvlist_alloc(&tags, NV_UNIQUE_NAME,
KM_PUSHPAGE));
fnvlist_add_boolean(tags, htag);
fnvlist_add_nvlist(tmpholds, name, tags);
fnvlist_free(tags);
} else {
fnvlist_add_boolean(tags, htag);
}
}
spa_history_log_internal_ds(ds, "hold", tx,
"tag=%s temp=%d refs=%llu",
htag, minor != 0, ds->ds_userrefs);
}
static void
dsl_deleg_set_sync(void *arg, dmu_tx_t *tx)
{
dsl_deleg_arg_t *dda = arg;
dsl_dir_t *dd;
dsl_pool_t *dp = dmu_tx_pool(tx);
objset_t *mos = dp->dp_meta_objset;
nvpair_t *whopair = NULL;
uint64_t zapobj;
VERIFY0(dsl_dir_hold(dp, dda->dda_name, FTAG, &dd, NULL));
zapobj = dd->dd_phys->dd_deleg_zapobj;
if (zapobj == 0) {
dmu_buf_will_dirty(dd->dd_dbuf, tx);
zapobj = dd->dd_phys->dd_deleg_zapobj = zap_create(mos,
DMU_OT_DSL_PERMS, DMU_OT_NONE, 0, tx);
}
while ((whopair = nvlist_next_nvpair(dda->dda_nvlist, whopair))) {
const char *whokey = nvpair_name(whopair);
nvlist_t *perms;
nvpair_t *permpair = NULL;
uint64_t jumpobj;
perms = fnvpair_value_nvlist(whopair);
if (zap_lookup(mos, zapobj, whokey, 8, 1, &jumpobj) != 0) {
jumpobj = zap_create_link(mos, DMU_OT_DSL_PERMS,
zapobj, whokey, tx);
}
while ((permpair = nvlist_next_nvpair(perms, permpair))) {
const char *perm = nvpair_name(permpair);
uint64_t n = 0;
VERIFY(zap_update(mos, jumpobj,
perm, 8, 1, &n, tx) == 0);
spa_history_log_internal_dd(dd, "permission update", tx,
"%s %s", whokey, perm);
}
}
dsl_dir_rele(dd, FTAG);
}
static void
dsl_scan_setup_sync(void *arg, dmu_tx_t *tx)
{
dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan;
pool_scan_func_t *funcp = arg;
dmu_object_type_t ot = 0;
dsl_pool_t *dp = scn->scn_dp;
spa_t *spa = dp->dp_spa;
ASSERT(scn->scn_phys.scn_state != DSS_SCANNING);
ASSERT(*funcp > POOL_SCAN_NONE && *funcp < POOL_SCAN_FUNCS);
bzero(&scn->scn_phys, sizeof (scn->scn_phys));
scn->scn_phys.scn_func = *funcp;
scn->scn_phys.scn_state = DSS_SCANNING;
scn->scn_phys.scn_min_txg = 0;
scn->scn_phys.scn_max_txg = tx->tx_txg;
scn->scn_phys.scn_ddt_class_max = DDT_CLASSES - 1; /* the entire DDT */
scn->scn_phys.scn_start_time = gethrestime_sec();
scn->scn_phys.scn_errors = 0;
scn->scn_phys.scn_to_examine = spa->spa_root_vdev->vdev_stat.vs_alloc;
scn->scn_restart_txg = 0;
spa_scan_stat_init(spa);
if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) {
scn->scn_phys.scn_ddt_class_max = zfs_scrub_ddt_class_max;
/* rewrite all disk labels */
vdev_config_dirty(spa->spa_root_vdev);
if (vdev_resilver_needed(spa->spa_root_vdev,
&scn->scn_phys.scn_min_txg, &scn->scn_phys.scn_max_txg)) {
spa_event_notify(spa, NULL, ESC_ZFS_RESILVER_START);
} else {
spa_event_notify(spa, NULL, ESC_ZFS_SCRUB_START);
}
spa->spa_scrub_started = B_TRUE;
/*
* If this is an incremental scrub, limit the DDT scrub phase
* to just the auto-ditto class (for correctness); the rest
* of the scrub should go faster using top-down pruning.
*/
if (scn->scn_phys.scn_min_txg > TXG_INITIAL)
scn->scn_phys.scn_ddt_class_max = DDT_CLASS_DITTO;
}
/* back to the generic stuff */
if (dp->dp_blkstats == NULL) {
dp->dp_blkstats =
kmem_alloc(sizeof (zfs_all_blkstats_t), KM_SLEEP);
}
bzero(dp->dp_blkstats, sizeof (zfs_all_blkstats_t));
if (spa_version(spa) < SPA_VERSION_DSL_SCRUB)
ot = DMU_OT_ZAP_OTHER;
scn->scn_phys.scn_queue_obj = zap_create(dp->dp_meta_objset,
ot ? ot : DMU_OT_SCAN_QUEUE, DMU_OT_NONE, 0, tx);
dsl_scan_sync_state(scn, tx);
spa_history_log_internal(spa, "scan setup", tx,
"func=%u mintxg=%llu maxtxg=%llu",
*funcp, scn->scn_phys.scn_min_txg, scn->scn_phys.scn_max_txg);
}
/* ARGSUSED */
static void
dsl_pool_scrub_setup_sync(void *arg1, void *arg2, cred_t *cr, dmu_tx_t *tx)
{
dsl_pool_t *dp = arg1;
enum scrub_func *funcp = arg2;
dmu_object_type_t ot = 0;
boolean_t complete = B_FALSE;
dsl_pool_scrub_cancel_sync(dp, &complete, cr, tx);
ASSERT(dp->dp_scrub_func == SCRUB_FUNC_NONE);
ASSERT(*funcp > SCRUB_FUNC_NONE);
ASSERT(*funcp < SCRUB_FUNC_NUMFUNCS);
dp->dp_scrub_min_txg = 0;
dp->dp_scrub_max_txg = tx->tx_txg;
dp->dp_scrub_ddt_class_max = zfs_scrub_ddt_class_max;
if (*funcp == SCRUB_FUNC_CLEAN) {
vdev_t *rvd = dp->dp_spa->spa_root_vdev;
/* rewrite all disk labels */
vdev_config_dirty(rvd);
if (vdev_resilver_needed(rvd,
&dp->dp_scrub_min_txg, &dp->dp_scrub_max_txg)) {
spa_event_notify(dp->dp_spa, NULL,
ESC_ZFS_RESILVER_START);
dp->dp_scrub_max_txg = MIN(dp->dp_scrub_max_txg,
tx->tx_txg);
} else {
spa_event_notify(dp->dp_spa, NULL,
ESC_ZFS_SCRUB_START);
}
/* zero out the scrub stats in all vdev_stat_t's */
vdev_scrub_stat_update(rvd,
dp->dp_scrub_min_txg ? POOL_SCRUB_RESILVER :
POOL_SCRUB_EVERYTHING, B_FALSE);
/*
* If this is an incremental scrub, limit the DDT scrub phase
* to just the auto-ditto class (for correctness); the rest
* of the scrub should go faster using top-down pruning.
*/
if (dp->dp_scrub_min_txg > TXG_INITIAL)
dp->dp_scrub_ddt_class_max = DDT_CLASS_DITTO;
dp->dp_spa->spa_scrub_started = B_TRUE;
}
/* back to the generic stuff */
if (dp->dp_blkstats == NULL) {
dp->dp_blkstats =
kmem_alloc(sizeof (zfs_all_blkstats_t), KM_SLEEP);
}
bzero(dp->dp_blkstats, sizeof (zfs_all_blkstats_t));
if (spa_version(dp->dp_spa) < SPA_VERSION_DSL_SCRUB)
ot = DMU_OT_ZAP_OTHER;
dp->dp_scrub_func = *funcp;
dp->dp_scrub_queue_obj = zap_create(dp->dp_meta_objset,
ot ? ot : DMU_OT_SCRUB_QUEUE, DMU_OT_NONE, 0, tx);
bzero(&dp->dp_scrub_bookmark, sizeof (zbookmark_t));
bzero(&dp->dp_scrub_ddt_bookmark, sizeof (ddt_bookmark_t));
dp->dp_scrub_restart = B_FALSE;
dp->dp_spa->spa_scrub_errors = 0;
VERIFY(0 == zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_FUNC, sizeof (uint32_t), 1,
&dp->dp_scrub_func, tx));
VERIFY(0 == zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_QUEUE, sizeof (uint64_t), 1,
&dp->dp_scrub_queue_obj, tx));
VERIFY(0 == zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_MIN_TXG, sizeof (uint64_t), 1,
&dp->dp_scrub_min_txg, tx));
VERIFY(0 == zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_MAX_TXG, sizeof (uint64_t), 1,
&dp->dp_scrub_max_txg, tx));
VERIFY(0 == zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_BOOKMARK, sizeof (uint64_t),
sizeof (dp->dp_scrub_bookmark) / sizeof (uint64_t),
&dp->dp_scrub_bookmark, tx));
VERIFY(0 == zap_update(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_DDT_BOOKMARK, sizeof (uint64_t),
sizeof (dp->dp_scrub_ddt_bookmark) / sizeof (uint64_t),
&dp->dp_scrub_ddt_bookmark, tx));
VERIFY(0 == zap_update(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_DDT_CLASS_MAX, sizeof (uint64_t), 1,
&dp->dp_scrub_ddt_class_max, tx));
VERIFY(0 == zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_ERRORS, sizeof (uint64_t), 1,
&dp->dp_spa->spa_scrub_errors, tx));
spa_history_internal_log(LOG_POOL_SCRUB, dp->dp_spa, tx, cr,
"func=%u mintxg=%llu maxtxg=%llu",
*funcp, dp->dp_scrub_min_txg, dp->dp_scrub_max_txg);
}
/* ARGSUSED */
static void
dsl_pool_scrub_setup_sync(void *arg1, void *arg2, cred_t *cr, dmu_tx_t *tx)
{
dsl_pool_t *dp = arg1;
enum scrub_func *funcp = arg2;
dmu_object_type_t ot = 0;
boolean_t complete = B_FALSE;
dsl_pool_scrub_cancel_sync(dp, &complete, cr, tx);
ASSERT(dp->dp_scrub_func == SCRUB_FUNC_NONE);
ASSERT(*funcp > SCRUB_FUNC_NONE);
ASSERT(*funcp < SCRUB_FUNC_NUMFUNCS);
dp->dp_scrub_min_txg = 0;
dp->dp_scrub_max_txg = tx->tx_txg;
if (*funcp == SCRUB_FUNC_CLEAN) {
vdev_t *rvd = dp->dp_spa->spa_root_vdev;
/* rewrite all disk labels */
vdev_config_dirty(rvd);
if (vdev_resilver_needed(rvd,
&dp->dp_scrub_min_txg, &dp->dp_scrub_max_txg)) {
spa_event_notify(dp->dp_spa, NULL,
ESC_ZFS_RESILVER_START);
dp->dp_scrub_max_txg = MIN(dp->dp_scrub_max_txg,
tx->tx_txg);
}
/* zero out the scrub stats in all vdev_stat_t's */
vdev_scrub_stat_update(rvd,
dp->dp_scrub_min_txg ? POOL_SCRUB_RESILVER :
POOL_SCRUB_EVERYTHING, B_FALSE);
dp->dp_spa->spa_scrub_started = B_TRUE;
}
/* back to the generic stuff */
if (spa_version(dp->dp_spa) < SPA_VERSION_DSL_SCRUB)
ot = DMU_OT_ZAP_OTHER;
dp->dp_scrub_func = *funcp;
dp->dp_scrub_queue_obj = zap_create(dp->dp_meta_objset,
ot ? ot : DMU_OT_SCRUB_QUEUE, DMU_OT_NONE, 0, tx);
bzero(&dp->dp_scrub_bookmark, sizeof (zbookmark_t));
dp->dp_scrub_restart = B_FALSE;
dp->dp_spa->spa_scrub_errors = 0;
VERIFY(0 == zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_FUNC, sizeof (uint32_t), 1,
&dp->dp_scrub_func, tx));
VERIFY(0 == zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_QUEUE, sizeof (uint64_t), 1,
&dp->dp_scrub_queue_obj, tx));
VERIFY(0 == zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_MIN_TXG, sizeof (uint64_t), 1,
&dp->dp_scrub_min_txg, tx));
VERIFY(0 == zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_MAX_TXG, sizeof (uint64_t), 1,
&dp->dp_scrub_max_txg, tx));
VERIFY(0 == zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_BOOKMARK, sizeof (uint64_t), 4,
&dp->dp_scrub_bookmark, tx));
VERIFY(0 == zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_SCRUB_ERRORS, sizeof (uint64_t), 1,
&dp->dp_spa->spa_scrub_errors, tx));
spa_history_internal_log(LOG_POOL_SCRUB, dp->dp_spa, tx, cr,
"func=%u mintxg=%llu maxtxg=%llu",
*funcp, dp->dp_scrub_min_txg, dp->dp_scrub_max_txg);
}
void
zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *zplprops, dmu_tx_t *tx)
{
struct super_block *sb;
zfs_sb_t *zsb;
uint64_t moid, obj, sa_obj, version;
uint64_t sense = ZFS_CASE_SENSITIVE;
uint64_t norm = 0;
nvpair_t *elem;
int error;
int i;
znode_t *rootzp = NULL;
vattr_t vattr;
znode_t *zp;
zfs_acl_ids_t acl_ids;
/*
* First attempt to create master node.
*/
/*
* In an empty objset, there are no blocks to read and thus
* there can be no i/o errors (which we assert below).
*/
moid = MASTER_NODE_OBJ;
error = zap_create_claim(os, moid, DMU_OT_MASTER_NODE,
DMU_OT_NONE, 0, tx);
ASSERT(error == 0);
/*
* Set starting attributes.
*/
version = zfs_zpl_version_map(spa_version(dmu_objset_spa(os)));
elem = NULL;
while ((elem = nvlist_next_nvpair(zplprops, elem)) != NULL) {
/* For the moment we expect all zpl props to be uint64_ts */
uint64_t val;
char *name;
ASSERT(nvpair_type(elem) == DATA_TYPE_UINT64);
VERIFY(nvpair_value_uint64(elem, &val) == 0);
name = nvpair_name(elem);
if (strcmp(name, zfs_prop_to_name(ZFS_PROP_VERSION)) == 0) {
if (val < version)
version = val;
} else {
error = zap_update(os, moid, name, 8, 1, &val, tx);
}
ASSERT(error == 0);
if (strcmp(name, zfs_prop_to_name(ZFS_PROP_NORMALIZE)) == 0)
norm = val;
else if (strcmp(name, zfs_prop_to_name(ZFS_PROP_CASE)) == 0)
sense = val;
}
ASSERT(version != 0);
error = zap_update(os, moid, ZPL_VERSION_STR, 8, 1, &version, tx);
/*
* Create zap object used for SA attribute registration
*/
if (version >= ZPL_VERSION_SA) {
sa_obj = zap_create(os, DMU_OT_SA_MASTER_NODE,
DMU_OT_NONE, 0, tx);
error = zap_add(os, moid, ZFS_SA_ATTRS, 8, 1, &sa_obj, tx);
ASSERT(error == 0);
} else {
sa_obj = 0;
}
/*
* Create a delete queue.
*/
obj = zap_create(os, DMU_OT_UNLINKED_SET, DMU_OT_NONE, 0, tx);
error = zap_add(os, moid, ZFS_UNLINKED_SET, 8, 1, &obj, tx);
ASSERT(error == 0);
/*
* Create root znode. Create minimal znode/inode/zsb/sb
* to allow zfs_mknode to work.
*/
vattr.va_mask = ATTR_MODE|ATTR_UID|ATTR_GID;
vattr.va_mode = S_IFDIR|0755;
vattr.va_uid = crgetuid(cr);
vattr.va_gid = crgetgid(cr);
rootzp = kmem_cache_alloc(znode_cache, KM_PUSHPAGE);
rootzp->z_moved = 0;
rootzp->z_unlinked = 0;
rootzp->z_atime_dirty = 0;
rootzp->z_is_sa = USE_SA(version, os);
zsb = kmem_zalloc(sizeof (zfs_sb_t), KM_PUSHPAGE | KM_NODEBUG);
zsb->z_os = os;
zsb->z_parent = zsb;
zsb->z_version = version;
zsb->z_use_fuids = USE_FUIDS(version, os);
zsb->z_use_sa = USE_SA(version, os);
zsb->z_norm = norm;
sb = kmem_zalloc(sizeof (struct super_block), KM_PUSHPAGE);
sb->s_fs_info = zsb;
ZTOI(rootzp)->i_sb = sb;
error = sa_setup(os, sa_obj, zfs_attr_table, ZPL_END,
&zsb->z_attr_table);
ASSERT(error == 0);
/*
* Fold case on file systems that are always or sometimes case
* insensitive.
*/
if (sense == ZFS_CASE_INSENSITIVE || sense == ZFS_CASE_MIXED)
zsb->z_norm |= U8_TEXTPREP_TOUPPER;
mutex_init(&zsb->z_znodes_lock, NULL, MUTEX_DEFAULT, NULL);
list_create(&zsb->z_all_znodes, sizeof (znode_t),
offsetof(znode_t, z_link_node));
for (i = 0; i != ZFS_OBJ_MTX_SZ; i++)
mutex_init(&zsb->z_hold_mtx[i], NULL, MUTEX_DEFAULT, NULL);
VERIFY(0 == zfs_acl_ids_create(rootzp, IS_ROOT_NODE, &vattr,
cr, NULL, &acl_ids));
zfs_mknode(rootzp, &vattr, tx, cr, IS_ROOT_NODE, &zp, &acl_ids);
ASSERT3P(zp, ==, rootzp);
error = zap_add(os, moid, ZFS_ROOT_OBJ, 8, 1, &rootzp->z_id, tx);
ASSERT(error == 0);
zfs_acl_ids_free(&acl_ids);
atomic_set(&ZTOI(rootzp)->i_count, 0);
sa_handle_destroy(rootzp->z_sa_hdl);
kmem_cache_free(znode_cache, rootzp);
/*
* Create shares directory
*/
error = zfs_create_share_dir(zsb, tx);
ASSERT(error == 0);
for (i = 0; i != ZFS_OBJ_MTX_SZ; i++)
mutex_destroy(&zsb->z_hold_mtx[i]);
kmem_free(sb, sizeof (struct super_block));
kmem_free(zsb, sizeof (zfs_sb_t));
}
