Fix ddtprune causing space leak

In zio_ddt_free, if a pruned dde is still in ddt, it would do nothing
and cause space leak.

Reviewed-by: Rob Norris <robn@despairlabs.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Allan Jude <allan@klarasystems.com>
Signed-off-by: Chunwei Chen <david.chen@nutanix.com>
Closes #17982
Closes #17983

module/zfs/zio.c

@@ -4157,6 +4157,17 @@ zio_ddt_free(zio_t *zio)
 		ddt_phys_variant_t v = ddt_phys_select(ddt, dde, bp);
 		if (v != DDT_PHYS_NONE)
 			ddt_phys_decref(dde->dde_phys, v);
+		else
+			/*
+			 * If the entry was found but the phys was not, then
+			 * this block must have been pruned from the dedup
+			 * table, and the entry refers to a later version of
+			 * this data. Therefore, the caller is trying to delete
+			 * the only stored instance of this block, and so we
+			 * need to do a normal (not dedup) free. Clear dde so
+			 * we fall into the block below.
+			 */
+			dde = NULL;
 	}
 	ddt_exit(ddt);
 

tests/runfiles/common.run

@@ -706,7 +706,8 @@ tags = ['functional', 'deadman']
 [tests/functional/dedup]
 tests = ['dedup_fdt_create', 'dedup_fdt_import', 'dedup_fdt_pacing',
     'dedup_legacy_create', 'dedup_legacy_import', 'dedup_legacy_fdt_upgrade',
-    'dedup_legacy_fdt_mixed', 'dedup_quota', 'dedup_prune', 'dedup_zap_shrink']
+    'dedup_legacy_fdt_mixed', 'dedup_quota', 'dedup_prune', 'dedup_prune_leak',
+    'dedup_zap_shrink']
 pre =
 post =
 tags = ['functional', 'dedup']

tests/zfs-tests/tests/Makefile.am

@@ -1482,6 +1482,7 @@ nobase_dist_datadir_zfs_tests_tests_SCRIPTS += \
 	functional/dedup/dedup_legacy_fdt_upgrade.ksh \
 	functional/dedup/dedup_legacy_fdt_mixed.ksh \
 	functional/dedup/dedup_prune.ksh \
+	functional/dedup/dedup_prune_leak.ksh \
 	functional/dedup/dedup_quota.ksh \
 	functional/dedup/dedup_zap_shrink.ksh \
 	functional/delegate/cleanup.ksh \

tests/zfs-tests/tests/functional/dedup/dedup_prune_leak.ksh

@@ -0,0 +1,86 @@
+#!/bin/ksh -p
+# SPDX-License-Identifier: CDDL-1.0
+# 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 https://opensource.org/licenses/CDDL-1.0.
+# 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) 2025, Klara Inc.
+# Copyright (c) 2025, Nutanix Inc.
+#
+
+# DESCRIPTION:
+#	Verify that zpool ddtprune successfully reduces the number of entries
+#	in the DDT.
+#
+# STRATEGY:
+#	1. Create a pool with dedup=on
+#	2. Add non-duplicate entries to the DDT
+#	3. ddtprune all entries
+#	4. Remove the file
+#	5. Verify there's no space leak
+#
+
+. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/tests/functional/events/events_common.kshlib
+
+verify_runnable "both"
+
+log_assert "Verify DDT pruning does not cause space leak"
+
+# We set the dedup log txg interval to 1, to get a log flush every txg,
+# effectively disabling the log. Without this it's hard to predict when
+# entries appear in the DDT ZAP
+log_must save_tunable DEDUP_LOG_TXG_MAX
+log_must set_tunable32 DEDUP_LOG_TXG_MAX 1
+log_must save_tunable DEDUP_LOG_FLUSH_ENTRIES_MIN
+log_must set_tunable32 DEDUP_LOG_FLUSH_ENTRIES_MIN 100000
+function cleanup
+{
+	if poolexists $TESTPOOL ; then
+		destroy_pool $TESTPOOL
+	fi
+	log_must restore_tunable DEDUP_LOG_TXG_MAX
+	log_must restore_tunable DEDUP_LOG_FLUSH_ENTRIES_MIN
+}
+
+log_onexit cleanup
+
+log_must zpool create -f $TESTPOOL $DISKS
+
+log_must zfs create -o dedup=on $TESTPOOL/$TESTFS
+typeset mountpoint=$(get_prop mountpoint $TESTPOOL/$TESTFS)
+log_must dd if=/dev/urandom of=$mountpoint/f1 bs=1M count=16
+# We seems to need some amount of txg sync here to make it more consistently
+# reproducible
+for i in $(seq 50); do
+	zpool sync $TESTPOOL
+done
+
+log_must zpool ddtprune -p 100 $TESTPOOL
+log_must rm $mountpoint/f1
+sync_pool $TESTPOOL
+
+zdb_out=$(zdb -bcc $TESTPOOL)
+echo "$zdb_out"
+if echo "$zdb_out" | grep -q "leaked space"; then
+	log_fail "DDT pruning causes space leak"
+fi
+
+log_pass "DDT pruning does not cause space leak"