Print this page
OS-1566 filesystem limits for ZFS datasets
@@ -19,10 +19,11 @@
* CDDL HEADER END
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2012 by Delphix. All rights reserved.
+ * Copyright (c) 2012 Joyent, Inc. All rights reserved.
*/
#include <sys/dmu.h>
#include <sys/dmu_objset.h>
#include <sys/dmu_tx.h>
@@ -36,16 +37,89 @@
#include <sys/zap.h>
#include <sys/zio.h>
#include <sys/arc.h>
#include <sys/sunddi.h>
#include <sys/zfs_zone.h>
+#include <sys/zfeature.h>
#include "zfs_namecheck.h"
+#include "zfs_prop.h"
+/*
+ * Filesystem and Snapshot Limits
+ * ------------------------------
+ *
+ * These limits are used to restrict the number of filesystems and/or snapshots
+ * that can be created at a given level in the tree or below. The standard
+ * use-case is with a delegated dataset where the administrator wants to ensure
+ * that a user within the zone is not creating too many additional filesystems
+ * or snapshots, even though they're not exceeding their space quota.
+ *
+ * The count of filesystems and snapshots is stored in the dsl_dir_phys_t which
+ * impacts the on-disk format. As such, this capability is controlled by a
+ * feature flag and must be enabled to be used. Once enabled, the feature is
+ * not active until the first limit is set. At that point, future operations to
+ * create/destroy filesystems or snapshots will validate and update the counts.
+ *
+ * Because the on-disk counts will be uninitialized (0) before the feature is
+ * active, the counts are updated when a limit is first set on an uninitialized
+ * node (The filesystem/snapshot counts on a node includes all of the nested
+ * filesystems/snapshots, plus the node itself. Thus, a new leaf node has a
+ * filesystem count of 1 and a snapshot count of 0. A filesystem count of 0 on
+ * a node indicates uninitialized counts on that node.) When setting a limit on
+ * an uninitialized node, the code starts at the filesystem with the new limit
+ * and descends into all sub-filesystems and updates the counts to be accurate.
+ * In practice this is lightweight since a limit is typically set when the
+ * filesystem is created and thus has no children. Once valid, changing the
+ * limit value won't require a re-traversal since the counts are already valid.
+ * When recursively fixing the counts, if a node with a limit is encountered
+ * during the descent, the counts are known to be valid and there is no need to
+ * descend into that filesystem's children. The counts on filesystems above the
+ * one with the new limit will still be uninitialized (0), unless a limit is
+ * eventually set on one of those filesystems. It is possible for the counts
+ * to appear initialized, but be invalid, if the feature was previously active
+ * but then deactivated. For this reason, the counts are always recursively
+ * updated when a limit is set on a dataset, unless there is already a limit.
+ * When a new limit value is set on a filesystem with an existing limit, the
+ * new limit must be greater than the current count at that level or an error
+ * is returned and the limit is not changed.
+ *
+ * Once the feature is active, then whenever a filesystem or snapshot is
+ * created, the code recurses up the tree, validating the new count against the
+ * limit at each initialized level. In practice, most levels will not have a
+ * limit set. If there is a limit at any initialized level up the tree, the
+ * check must pass or the creation will fail. Likewise, when a filesystem or
+ * snapshot is destroyed, the counts are recursively adjusted all the way up
+ * the initizized nodes in the tree. Renaming a filesystem into different point
+ * in the tree will first validate, then update the counts on each branch up to
+ * the common ancestor. A receive will also validate the counts and then update
+ * them.
+ *
+ * An exception to the above behavior is that the limits are never enforced
+ * for the administrative user in the global zone. This is primarily so that
+ * recursive snapshots in the global zone always work. We want to prevent a
+ * denial-of-service in which a lower level delegated dataset could max out its
+ * limit and thus block recursive snapshots from being taken in the global zone.
+ * Because of this, it is possible for the snapshot count to be over the limit
+ * and snapshots taken in the global zone could cause a lower level dataset to
+ * hit or exceed its limit. The administrator taking the global zone recursive
+ * snapshot should be aware of this side-effect and behave accordingly.
+ * For consistency, the filesystem limit is also not enforced for the admin
+ * user in the global zone.
+ *
+ * The filesystem limit is validated by dsl_dir_fscount_check() and updated by
+ * dsl_dir_fscount_adjust(). The snapshot limit is validated by
+ * dsl_snapcount_check() and updated by dsl_snapcount_adjust().
+ * A new limit value is validated in dsl_dir_validate_fs_ss_limit() and the
+ * filesystem counts are adjusted, if necessary, by dsl_dir_set_fs_ss_count().
+ */
+
static uint64_t dsl_dir_space_towrite(dsl_dir_t *dd);
static void dsl_dir_set_reservation_sync_impl(dsl_dir_t *dd,
uint64_t value, dmu_tx_t *tx);
+extern dsl_syncfunc_t dsl_prop_set_sync;
+
/* ARGSUSED */
static void
dsl_dir_evict(dmu_buf_t *db, void *arg)
{
dsl_dir_t *dd = arg;
@@ -405,19 +479,381 @@
dsl_dir_open(const char *name, void *tag, dsl_dir_t **ddp, const char **tailp)
{
return (dsl_dir_open_spa(NULL, name, tag, ddp, tailp));
}
+/*
+ * Check if the counts are already valid for this filesystem and its
+ * descendants. The counts on this filesystem, and those below, may be
+ * uninitialized due to either the use of a pre-existing pool which did not
+ * support the filesystem/snapshot limit feature, or one in which the feature
+ * had not yet been enabled. The counts can also be invalid if the feature was
+ * previously active but then deactivated.
+ *
+ * Recursively descend the filesystem tree and update the filesystem/snapshot
+ * counts on each filesystem below, then update the cumulative count on the
+ * current filesystem. If the filesystem already has a limit set on it,
+ * then we know that its counts, and the counts on the filesystems below it,
+ * have been updated to be correct, so we can skip this filesystem.
+ */
+static void
+dsl_dir_set_fs_ss_count(const char *nm, dsl_dir_t *dd, dmu_tx_t *tx,
+ uint64_t *fscnt, uint64_t *sscnt)
+{
+ uint64_t my_fs_cnt = 0;
+ uint64_t my_ss_cnt = 0;
+ objset_t *os = dd->dd_pool->dp_meta_objset;
+ zap_cursor_t *zc;
+ zap_attribute_t *za;
+ char *namebuf;
+ int err;
+ boolean_t limit_set = B_FALSE;
+ uint64_t fslimit, sslimit;
+ dsl_dataset_t *ds;
+
+ err = dsl_prop_get_dd(dd, zfs_prop_to_name(ZFS_PROP_FILESYSTEM_LIMIT),
+ 8, 1, &fslimit, NULL, B_FALSE);
+ if (err == 0 && fslimit != MAXLIMIT)
+ limit_set = B_TRUE;
+
+ if (!limit_set) {
+ err = dsl_prop_get_dd(dd,
+ zfs_prop_to_name(ZFS_PROP_SNAPSHOT_LIMIT), 8, 1, &sslimit,
+ NULL, B_FALSE);
+ if (err == 0 && sslimit != MAXLIMIT)
+ limit_set = B_TRUE;
+ }
+
+ /*
+ * If the dd has a limit, we know its count is already good and we
+ * don't need to recurse down any further.
+ *
+ * We can't check for an initialized (non-0) count since the feature
+ * might have been previously active, then deactivated and is now
+ * being activated again.
+ */
+ if (limit_set) {
+ *fscnt = dd->dd_phys->dd_filesystem_count;
+ *sscnt = dd->dd_phys->dd_snapshot_count;
+ return;
+ }
+
+ zc = kmem_alloc(sizeof (zap_cursor_t), KM_SLEEP);
+ za = kmem_alloc(sizeof (zap_attribute_t), KM_SLEEP);
+ namebuf = kmem_alloc(MAXPATHLEN, KM_SLEEP);
+
+ mutex_enter(&dd->dd_lock);
+
+ /* Iterate datasets */
+ for (zap_cursor_init(zc, os, dd->dd_phys->dd_child_dir_zapobj);
+ zap_cursor_retrieve(zc, za) == 0;
+ zap_cursor_advance(zc)) {
+ dsl_dir_t *chld_dd;
+ uint64_t chld_fs_cnt = 0;
+ uint64_t chld_ss_cnt = 0;
+
+ (void) snprintf(namebuf, MAXPATHLEN, "%s/%s", nm, za->za_name);
+
+ if (dsl_dir_open(namebuf, FTAG, &chld_dd, NULL))
+ continue;
+
+ dsl_dir_set_fs_ss_count(namebuf, chld_dd, tx, &chld_fs_cnt,
+ &chld_ss_cnt);
+
+ dsl_dir_close(chld_dd, FTAG);
+
+ my_fs_cnt += chld_fs_cnt;
+ my_ss_cnt += chld_ss_cnt;
+ }
+ zap_cursor_fini(zc);
+
+ kmem_free(namebuf, MAXPATHLEN);
+
+ /* Iterate snapshots */
+ if (dsl_dataset_hold(nm, FTAG, &ds) == 0) {
+ for (zap_cursor_init(zc, os, ds->ds_phys->ds_snapnames_zapobj);
+ zap_cursor_retrieve(zc, za) == 0;
+ zap_cursor_advance(zc)) {
+ my_ss_cnt++;
+ }
+ zap_cursor_fini(zc);
+ dsl_dataset_rele(ds, FTAG);
+ }
+
+ kmem_free(zc, sizeof (zap_cursor_t));
+ kmem_free(za, sizeof (zap_attribute_t));
+
+ /* Add 1 for self */
+ my_fs_cnt++;
+
+#ifdef _KERNEL
+ extern void __dtrace_probe_zfs__fs__fix__count(char *, uint64_t,
+ uint64_t);
+ __dtrace_probe_zfs__fs__fix__count((char *)nm, my_fs_cnt, my_ss_cnt);
+#endif
+
+ /* save updated counts */
+ dmu_buf_will_dirty(dd->dd_dbuf, tx);
+ dd->dd_phys->dd_filesystem_count = my_fs_cnt;
+ dd->dd_phys->dd_snapshot_count = my_ss_cnt;
+
+ mutex_exit(&dd->dd_lock);
+
+ /* Return child dataset count plus self */
+ *fscnt = my_fs_cnt;
+ *sscnt = my_ss_cnt;
+}
+
+/*
+ * Return ENOSPC if new limit is less than the existing count, otherwise return
+ * -1 to force the zfs_set_prop_nvlist code down the default path to set the
+ * value in the nvlist.
+ */
+int
+dsl_dir_validate_fs_ss_limit(const char *ddname, uint64_t limit,
+ zfs_prop_t ptype)
+{
+ dsl_dir_t *dd;
+ dsl_dataset_t *ds;
+ int err = -1;
+ uint64_t count;
+ dmu_tx_t *tx;
+ uint64_t my_fs_cnt = 0;
+ uint64_t my_ss_cnt = 0;
+ uint64_t curr_limit;
+ spa_t *spa;
+ zfeature_info_t *limit_feat =
+ &spa_feature_table[SPA_FEATURE_FS_SS_LIMIT];
+
+ if (dsl_dataset_hold(ddname, FTAG, &ds))
+ return (EACCES);
+
+ spa = dsl_dataset_get_spa(ds);
+ if (!spa_feature_is_enabled(spa,
+ &spa_feature_table[SPA_FEATURE_FS_SS_LIMIT])) {
+ dsl_dataset_rele(ds, FTAG);
+ return (ENOTSUP);
+ }
+
+ if (dsl_dir_open(ddname, FTAG, &dd, NULL)) {
+ dsl_dataset_rele(ds, FTAG);
+ return (EACCES);
+ }
+
+ ASSERT(ds->ds_dir == dd);
+
+ if (dsl_prop_get_dd(dd, zfs_prop_to_name(ptype), 8, 1, &curr_limit,
+ NULL, B_FALSE) != 0)
+ curr_limit = MAXLIMIT;
+
+ tx = dmu_tx_create_dd(dd);
+ if (dmu_tx_assign(tx, TXG_WAIT)) {
+ dmu_tx_abort(tx);
+ dsl_dir_close(dd, FTAG);
+ dsl_dataset_rele(ds, FTAG);
+ return (ENOSPC);
+ }
+
+ if (limit == MAXLIMIT) {
+ /*
+ * If we had a limit, since we're now removing that limit,
+ * decrement the feature-active counter so that the feature
+ * becomes inactive (only enabled) if we remove the last limit.
+ */
+ if (curr_limit != MAXLIMIT)
+ spa_feature_decr(spa, limit_feat, tx);
+
+ dmu_tx_commit(tx);
+ dsl_dir_close(dd, FTAG);
+ dsl_dataset_rele(ds, FTAG);
+ return (-1);
+ }
+
+ /*
+ * Since we are now setting a non-MAXLIMIT on the filesystem, we need
+ * to ensure the counts are correct. Descend down the tree from this
+ * point and update all of the counts to be accurate.
+ */
+ rw_enter(&dd->dd_pool->dp_config_rwlock, RW_READER);
+ dsl_dir_set_fs_ss_count(ddname, dd, tx, &my_fs_cnt, &my_ss_cnt);
+ rw_exit(&dd->dd_pool->dp_config_rwlock);
+
+ if (ptype == ZFS_PROP_FILESYSTEM_LIMIT)
+ count = dd->dd_phys->dd_filesystem_count;
+ else
+ count = dd->dd_phys->dd_snapshot_count;
+
+ if (limit < count) {
+ err = ENOSPC;
+ } else {
+ /*
+ * If we had no limit, since we're now setting a limit
+ * increment the feature-active counter so that the feature
+ * either becomes active for the first time, or the count
+ * simply increases so that we can decrement it when we remove
+ * the limit.
+ */
+ if (curr_limit == MAXLIMIT)
+ spa_feature_incr(spa, limit_feat, tx);
+ }
+
+ dmu_tx_commit(tx);
+
+ dsl_dir_close(dd, FTAG);
+ dsl_dataset_rele(ds, FTAG);
+
+ return (err);
+}
+
+/*
+ * Check if adding additional child filesystem(s) would exceed any filesystem
+ * limits. Note that all filesystem limits up to the root (or the highest
+ * initialized) filesystem or the given ancestor must be satisfied.
+ */
+int
+dsl_dir_fscount_check(dsl_dir_t *dd, uint64_t cnt, dsl_dir_t *ancestor)
+{
+ uint64_t limit;
+ int err = 0;
+
+ VERIFY(RW_LOCK_HELD(&dd->dd_pool->dp_config_rwlock));
+
+ /*
+ * The limit is never enforced for the admin user in global zone.
+ * If we're not in the global zone then we need to run this check in
+ * open context, since thats when we know what zone we're in and
+ * syncing is only performed in the global zone.
+ */
+ if (INGLOBALZONE(curproc))
+ return (0);
+
+ /*
+ * If an ancestor has been provided, stop checking the limit once we
+ * hit that dir. We need this during rename so that we don't overcount
+ * the check once we recurse up to the common ancestor.
+ */
+ if (ancestor == dd)
+ return (0);
+
+ /*
+ * If we hit an uninitialized node while recursing up the tree, we can
+ * stop since we know the counts are not valid on this node and we
+ * know we won't touch this node's counts.
+ */
+ if (dd->dd_phys->dd_filesystem_count == 0)
+ return (0);
+
+ /*
+ * If there's no value for this property, there's no need to enforce a
+ * filesystem limit.
+ */
+ err = dsl_prop_get_dd(dd, zfs_prop_to_name(ZFS_PROP_FILESYSTEM_LIMIT),
+ 8, 1, &limit, NULL, B_FALSE);
+ if (err == ENOENT)
+ return (0);
+ else if (err != 0)
+ return (err);
+
+#ifdef _KERNEL
+ extern void __dtrace_probe_zfs__fs__limit(uint64_t, uint64_t, char *);
+ __dtrace_probe_zfs__fs__limit(
+ (uint64_t)dd->dd_phys->dd_filesystem_count, (uint64_t)limit,
+ dd->dd_myname);
+#endif
+
+ if (limit != MAXLIMIT &&
+ (dd->dd_phys->dd_filesystem_count + cnt) > limit)
+ return (EDQUOT);
+
+ if (dd->dd_parent != NULL)
+ err = dsl_dir_fscount_check(dd->dd_parent, cnt, ancestor);
+
+ return (err);
+}
+
+/*
+ * Adjust the filesystem count for the specified dsl_dir_t and all parent
+ * filesystems. When a new filesystem is created, increment the count on all
+ * parents, and when a filesystem is destroyed, decrement the count.
+ */
+void
+dsl_dir_fscount_adjust(dsl_dir_t *dd, dmu_tx_t *tx, int64_t delta,
+ boolean_t syncing, boolean_t first)
+{
+ VERIFY(RW_LOCK_HELD(&dd->dd_pool->dp_config_rwlock));
+ if (syncing)
+ VERIFY(dmu_tx_is_syncing(tx));
+
+ /*
+ * There is a special case where we are receiving a filesystem that
+ * already exists. In this case a temporary clone name of %X is created
+ * (see dmu_recv_begin). In dmu_recv_existing_end we destroy this
+ * temporary clone. We never update the filesystem counts for temporary
+ * clones. To detect this case we check the filesystem name to see if
+ * its a hidden filesystem (%X).
+ */
+ if (dd->dd_myname[0] == '%')
+ return;
+
+ /*
+ * If we hit an uninitialized node while recursing up the tree, we can
+ * stop since we know the counts are not valid on this node and we
+ * know we shouldn't touch this node's counts. An uninitialized count
+ * on the node indicates that either the feature has not yet been
+ * activated or there are no limits on this part of the tree.
+ */
+ if (dd->dd_phys->dd_filesystem_count == 0)
+ return;
+
+ /*
+ * The feature might have previously been active, so there could be
+ * non-0 counts on the nodes, but it might now be inactive.
+ *
+ * On initial entry we need to check if this feature is active, but
+ * we don't want to re-check this on each recursive call. Note: the
+ * feature cannot be active if its not enabled. If the feature is not
+ * active, don't touch the on-disk count fields.
+ */
+ if (first) {
+ dsl_dataset_t *ds = NULL;
+ spa_t *spa;
+ zfeature_info_t *quota_feat =
+ &spa_feature_table[SPA_FEATURE_FS_SS_LIMIT];
+
+ VERIFY(0 == dsl_dataset_hold_obj(dd->dd_pool,
+ dd->dd_phys->dd_head_dataset_obj, FTAG, &ds));
+ spa = dsl_dataset_get_spa(ds);
+ dsl_dataset_rele(ds, FTAG);
+ if (!spa_feature_is_active(spa, quota_feat))
+ return;
+ }
+
+ dmu_buf_will_dirty(dd->dd_dbuf, tx);
+
+ mutex_enter(&dd->dd_lock);
+
+ dd->dd_phys->dd_filesystem_count += delta;
+
+ if (dd->dd_parent != NULL)
+ dsl_dir_fscount_adjust(dd->dd_parent, tx, delta, syncing,
+ B_FALSE);
+
+ mutex_exit(&dd->dd_lock);
+}
+
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 *ddphys;
dmu_buf_t *dbuf;
+ zfeature_info_t *limit_feat =
+ &spa_feature_table[SPA_FEATURE_FS_SS_LIMIT];
+
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));
@@ -429,10 +865,13 @@
VERIFY(0 == dmu_bonus_hold(mos, ddobj, FTAG, &dbuf));
dmu_buf_will_dirty(dbuf, tx);
ddphys = dbuf->db_data;
ddphys->dd_creation_time = gethrestime_sec();
+ /* Only initialize the count if the limit feature is active */
+ if (spa_feature_is_active(dp->dp_spa, limit_feat))
+ ddphys->dd_filesystem_count = 1;
if (pds)
ddphys->dd_parent_obj = pds->dd_object;
ddphys->dd_props_zapobj = zap_create(mos,
DMU_OT_DSL_PROPS, DMU_OT_NONE, 0, tx);
ddphys->dd_child_dir_zapobj = zap_create(mos,
@@ -485,10 +924,14 @@
dd_used_t t;
ASSERT(RW_WRITE_HELD(&dd->dd_pool->dp_config_rwlock));
ASSERT(dd->dd_phys->dd_head_dataset_obj == 0);
+ /* Decrement the filesystem count for all parent filesystems. */
+ if (dd->dd_parent != NULL)
+ dsl_dir_fscount_adjust(dd->dd_parent, tx, -1, B_TRUE, B_TRUE);
+
/*
* Remove our reservation. The impl() routine avoids setting the
* actual property, which would require the (already destroyed) ds.
*/
dsl_dir_set_reservation_sync_impl(dd, 0, tx);
@@ -1034,12 +1477,10 @@
}
mutex_exit(&dd->dd_lock);
return (err);
}
-extern dsl_syncfunc_t dsl_prop_set_sync;
-
static void
dsl_dir_set_quota_sync(void *arg1, void *arg2, dmu_tx_t *tx)
{
dsl_dataset_t *ds = arg1;
dsl_dir_t *dd = ds->ds_dir;
@@ -1276,11 +1717,11 @@
/* no rename into our descendant */
if (closest_common_ancestor(dd, ra->newparent) == dd)
return (EINVAL);
if (err = dsl_dir_transfer_possible(dd->dd_parent,
- ra->newparent, myspace))
+ ra->newparent, dd, myspace, tx))
return (err);
}
return (0);
}
@@ -1301,10 +1742,24 @@
dsl_dir_name(ra->newparent, namebuf);
spa_history_log_internal_dd(dd, "rename", tx,
"-> %s/%s", namebuf, ra->mynewname);
if (ra->newparent != dd->dd_parent) {
+ int cnt;
+
+ mutex_enter(&dd->dd_lock);
+
+ cnt = dd->dd_phys->dd_filesystem_count;
+ dsl_dir_fscount_adjust(dd->dd_parent, tx, -cnt, B_TRUE, B_TRUE);
+ dsl_dir_fscount_adjust(ra->newparent, tx, cnt, B_TRUE, B_TRUE);
+
+ cnt = dd->dd_phys->dd_snapshot_count;
+ dsl_snapcount_adjust(dd->dd_parent, tx, -cnt, B_TRUE);
+ dsl_snapcount_adjust(ra->newparent, tx, cnt, B_TRUE);
+
+ mutex_exit(&dd->dd_lock);
+
dsl_dir_diduse_space(dd->dd_parent, DD_USED_CHILD,
-dd->dd_phys->dd_used_bytes,
-dd->dd_phys->dd_compressed_bytes,
-dd->dd_phys->dd_uncompressed_bytes, tx);
dsl_dir_diduse_space(ra->newparent, DD_USED_CHILD,
@@ -1373,22 +1828,35 @@
dsl_dir_close(ra.newparent, FTAG);
return (err);
}
int
-dsl_dir_transfer_possible(dsl_dir_t *sdd, dsl_dir_t *tdd, uint64_t space)
+dsl_dir_transfer_possible(dsl_dir_t *sdd, dsl_dir_t *tdd, dsl_dir_t *moving_dd,
+ uint64_t space, dmu_tx_t *tx)
{
dsl_dir_t *ancestor;
int64_t adelta;
uint64_t avail;
+ int err;
ancestor = closest_common_ancestor(sdd, tdd);
adelta = would_change(sdd, -space, ancestor);
avail = dsl_dir_space_available(tdd, ancestor, adelta, FALSE);
if (avail < space)
return (ENOSPC);
+ if (sdd != moving_dd) {
+ err = dsl_dir_fscount_check(tdd,
+ moving_dd->dd_phys->dd_filesystem_count, ancestor);
+ if (err != 0)
+ return (err);
+ }
+ err = dsl_snapcount_check(tdd, moving_dd->dd_phys->dd_snapshot_count,
+ ancestor);
+ if (err != 0)
+ return (err);
+
return (0);
}
timestruc_t
dsl_dir_snap_cmtime(dsl_dir_t *dd)