[2.2] Additional backports

cmd/ztest.c

@@ -2457,8 +2457,7 @@ ztest_get_data(void *arg, uint64_t arg2, lr_write_t *lr, char *buf,
 		zgd->zgd_lr = (struct zfs_locked_range *)ztest_range_lock(zd,
 		    object, offset, size, RL_READER);
 
-		error = dmu_buf_hold(os, object, offset, zgd, &db,
-		    DMU_READ_NO_PREFETCH);
+		error = dmu_buf_hold_noread(os, object, offset, zgd, &db);
 
 		if (error == 0) {
 			blkptr_t *bp = &lr->lr_blkptr;
@@ -6379,6 +6378,7 @@ ztest_reguid(ztest_ds_t *zd, uint64_t id)
 	spa_t *spa = ztest_spa;
 	uint64_t orig, load;
 	int error;
+	ztest_shared_t *zs = ztest_shared;
 
 	if (ztest_opts.zo_mmp_test)
 		return;
@@ -6388,6 +6388,7 @@ ztest_reguid(ztest_ds_t *zd, uint64_t id)
 
 	(void) pthread_rwlock_wrlock(&ztest_name_lock);
 	error = spa_change_guid(spa);
+	zs->zs_guid = spa_guid(spa);
 	(void) pthread_rwlock_unlock(&ztest_name_lock);
 
 	if (error != 0)
@@ -6917,7 +6918,7 @@ ztest_trim(ztest_ds_t *zd, uint64_t id)
  * Verify pool integrity by running zdb.
  */
 static void
-ztest_run_zdb(const char *pool)
+ztest_run_zdb(uint64_t guid)
 {
 	int status;
 	char *bin;
@@ -6941,13 +6942,13 @@ ztest_run_zdb(const char *pool)
 	free(set_gvars_args);
 
 	size_t would = snprintf(zdb, len,
-	    "%s -bcc%s%s -G -d -Y -e -y %s -p %s %s",
+	    "%s -bcc%s%s -G -d -Y -e -y %s -p %s %"PRIu64,
 	    bin,
 	    ztest_opts.zo_verbose >= 3 ? "s" : "",
 	    ztest_opts.zo_verbose >= 4 ? "v" : "",
 	    set_gvars_args_joined,
 	    ztest_opts.zo_dir,
-	    pool);
+	    guid);
 	ASSERT3U(would, <, len);
 
 	umem_free(set_gvars_args_joined, strlen(set_gvars_args_joined) + 1);
@@ -7525,14 +7526,15 @@ ztest_import(ztest_shared_t *zs)
 	VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG));
 	zs->zs_metaslab_sz =
 	    1ULL << spa->spa_root_vdev->vdev_child[0]->vdev_ms_shift;
+	zs->zs_guid = spa_guid(spa);
 	spa_close(spa, FTAG);
 
 	kernel_fini();
 
 	if (!ztest_opts.zo_mmp_test) {
-		ztest_run_zdb(ztest_opts.zo_pool);
+		ztest_run_zdb(zs->zs_guid);
 		ztest_freeze();
-		ztest_run_zdb(ztest_opts.zo_pool);
+		ztest_run_zdb(zs->zs_guid);
 	}
 
 	(void) pthread_rwlock_destroy(&ztest_name_lock);
@@ -7603,7 +7605,6 @@ ztest_run(ztest_shared_t *zs)
 	dsl_pool_config_enter(dmu_objset_pool(os), FTAG);
 	dmu_objset_fast_stat(os, &dds);
 	dsl_pool_config_exit(dmu_objset_pool(os), FTAG);
-	zs->zs_guid = dds.dds_guid;
 	dmu_objset_disown(os, B_TRUE, FTAG);
 
 	/*
@@ -7874,14 +7875,15 @@ ztest_init(ztest_shared_t *zs)
 	VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG));
 	zs->zs_metaslab_sz =
 	    1ULL << spa->spa_root_vdev->vdev_child[0]->vdev_ms_shift;
+	zs->zs_guid = spa_guid(spa);
 	spa_close(spa, FTAG);
 
 	kernel_fini();
 
 	if (!ztest_opts.zo_mmp_test) {
-		ztest_run_zdb(ztest_opts.zo_pool);
+		ztest_run_zdb(zs->zs_guid);
 		ztest_freeze();
-		ztest_run_zdb(ztest_opts.zo_pool);
+		ztest_run_zdb(zs->zs_guid);
 	}
 
 	(void) pthread_rwlock_destroy(&ztest_name_lock);
@@ -8304,7 +8306,7 @@ main(int argc, char **argv)
 		}
 
 		if (!ztest_opts.zo_mmp_test)
-			ztest_run_zdb(ztest_opts.zo_pool);
+			ztest_run_zdb(zs->zs_guid);
 	}
 
 	if (ztest_opts.zo_verbose >= 1) {

module/zfs/dsl_dir.c

@@ -26,6 +26,7 @@
  * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
  * Copyright (c) 2016 Actifio, Inc. All rights reserved.
  * Copyright (c) 2018, loli10K <[email protected]>. All rights reserved.
+ * Copyright (c) 2023 Hewlett Packard Enterprise Development LP.
  */
 
 #include <sys/dmu.h>
@@ -1358,30 +1359,23 @@ dsl_dir_tempreserve_impl(dsl_dir_t *dd, uint64_t asize, boolean_t netfree,
 		ext_quota = 0;
 
 	if (used_on_disk >= quota) {
+		if (retval == ENOSPC && (used_on_disk - quota) <
+		    dsl_pool_deferred_space(dd->dd_pool)) {
+			retval = SET_ERROR(ERESTART);
+		}
 		/* Quota exceeded */
 		mutex_exit(&dd->dd_lock);
 		DMU_TX_STAT_BUMP(dmu_tx_quota);
 		return (retval);
 	} else if (used_on_disk + est_inflight >= quota + ext_quota) {
-		if (est_inflight > 0 || used_on_disk < quota) {
-			retval = SET_ERROR(ERESTART);
-		} else {
-			ASSERT3U(used_on_disk, >=, quota);
-
-			if (retval == ENOSPC && (used_on_disk - quota) <
-			    dsl_pool_deferred_space(dd->dd_pool)) {
-				retval = SET_ERROR(ERESTART);
-			}
-		}
-
 		dprintf_dd(dd, "failing: used=%lluK inflight = %lluK "
-		    "quota=%lluK tr=%lluK err=%d\n",
+		    "quota=%lluK tr=%lluK\n",
 		    (u_longlong_t)used_on_disk>>10,
 		    (u_longlong_t)est_inflight>>10,
-		    (u_longlong_t)quota>>10, (u_longlong_t)asize>>10, retval);
+		    (u_longlong_t)quota>>10, (u_longlong_t)asize>>10);
 		mutex_exit(&dd->dd_lock);
 		DMU_TX_STAT_BUMP(dmu_tx_quota);
-		return (retval);
+		return (SET_ERROR(ERESTART));
 	}
 
 	/* We need to up our estimated delta before dropping dd_lock */

module/zfs/metaslab.c

@@ -3208,6 +3208,15 @@ metaslab_segment_weight(metaslab_t *msp)
 static boolean_t
 metaslab_should_allocate(metaslab_t *msp, uint64_t asize, boolean_t try_hard)
 {
+	/*
+	 * This case will usually but not always get caught by the checks below;
+	 * metaslabs can be loaded by various means, including the trim and
+	 * initialize code. Once that happens, without this check they are
+	 * allocatable even before they finish their first txg sync.
+	 */
+	if (unlikely(msp->ms_new))
+		return (B_FALSE);
+
 	/*
 	 * If the metaslab is loaded, ms_max_size is definitive and we can use
 	 * the fast check. If it's not, the ms_max_size is a lower bound (once

tests/test-runner/bin/zts-report.py.in

@@ -214,6 +214,7 @@ maybe = {
     'cli_root/zfs_get/zfs_get_009_pos': ['SKIP', 5479],
     'cli_root/zfs_rollback/zfs_rollback_001_pos': ['FAIL', known_reason],
     'cli_root/zfs_rollback/zfs_rollback_002_pos': ['FAIL', known_reason],
+    'cli_root/zfs_share/zfs_share_concurrent_shares': ['FAIL', known_reason],
     'cli_root/zfs_snapshot/zfs_snapshot_002_neg': ['FAIL', known_reason],
     'cli_root/zfs_unshare/zfs_unshare_006_pos': ['SKIP', na_reason],
     'cli_root/zpool_add/zpool_add_004_pos': ['FAIL', known_reason],
@@ -259,10 +260,9 @@ if sys.platform.startswith('freebsd'):
     maybe.update({
         'cli_root/zfs_copies/zfs_copies_002_pos': ['FAIL', known_reason],
         'cli_root/zfs_inherit/zfs_inherit_001_neg': ['FAIL', known_reason],
-        'cli_root/zfs_share/zfs_share_concurrent_shares':
-            ['FAIL', known_reason],
         'cli_root/zpool_import/zpool_import_012_pos': ['FAIL', known_reason],
         'delegate/zfs_allow_003_pos': ['FAIL', known_reason],
+        'delegate/zfs_allow_010_pos': ['FAIL', known_reason],
         'inheritance/inherit_001_pos': ['FAIL', 11829],
         'resilver/resilver_restart_001': ['FAIL', known_reason],
         'pool_checkpoint/checkpoint_big_rewind': ['FAIL', 12622],

tests/zfs-tests/tests/functional/block_cloning/block_cloning_copyfilerange_fallback_same_txg.ksh

@@ -48,10 +48,9 @@ function cleanup
 
 log_onexit cleanup
 
-log_must zpool create -o feature@block_cloning=enabled $TESTPOOL $DISKS
-
 log_must set_tunable64 TXG_TIMEOUT 5000
-log_must sync_pool $TESTPOOL true
+
+log_must zpool create -o feature@block_cloning=enabled $TESTPOOL $DISKS
 
 log_must dd if=/dev/urandom of=/$TESTPOOL/file bs=128K count=4
 log_must clonefile -f /$TESTPOOL/file /$TESTPOOL/clone 0 0 524288

tests/zfs-tests/tests/functional/cli_root/zpool_trim/zpool_trim_partial.ksh

@@ -60,7 +60,7 @@ log_must set_tunable64 VDEV_MIN_MS_COUNT 64
 
 # Minimum trim size is decreased to verify all trim sizes.
 typeset trim_extent_bytes_min=$(get_tunable TRIM_EXTENT_BYTES_MIN)
-log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 4096
+log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 512
 
 log_must mkdir "$TESTDIR"
 log_must truncate -s $LARGESIZE "$LARGEFILE"

tests/zfs-tests/tests/functional/cli_root/zpool_trim/zpool_trim_verify_trimmed.ksh

@@ -52,7 +52,7 @@ LARGEFILE="$TESTDIR/largefile"
 
 # Reduce trim size to allow for tighter tolerance below when checking.
 typeset trim_extent_bytes_min=$(get_tunable TRIM_EXTENT_BYTES_MIN)
-log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 4096
+log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 512
 
 log_must mkdir "$TESTDIR"
 log_must truncate -s $LARGESIZE "$LARGEFILE"

tests/zfs-tests/tests/functional/no_space/enospc_rm.ksh

@@ -17,6 +17,7 @@
 #
 # Copyright (c) 2014, 2016 by Delphix. All rights reserved.
 # Copyright (c) 2022 by Lawrence Livermore National Security, LLC.
+# Copyright (c) 2023 Hewlett Packard Enterprise Development LP.
 #
 
 . $STF_SUITE/include/libtest.shlib
@@ -51,11 +52,20 @@ log_must zfs create $TESTPOOL/$TESTFS
 log_must zfs set mountpoint=$TESTDIR $TESTPOOL/$TESTFS
 log_must zfs set compression=off $TESTPOOL/$TESTFS
 
-log_note "Writing files until ENOSPC."
+log_note "Writing Big(1G) files until ENOSPC."
 log_mustnot_expect "No space left on device" fio --name=test \
     --fallocate=none --rw=write --bs=1M --size=1G --numjobs=4 \
     --sync=1 --directory=$TESTDIR/ --group_reporting
 
+log_must rm $TESTDIR/test.*
+log_must test -z "$(ls -A $TESTDIR)"
+sync_pool $TESTPOOL true
+
+log_note "Writing small(10M) files until ENOSPC."
+log_mustnot_expect "No space left on device" fio --name=test \
+    --fallocate=none --rw=write --bs=1M --size=10M --numjobs=200 \
+    --sync=1 --directory=$TESTDIR/ --group_reporting
+
 log_must rm $TESTDIR/test.*
 log_must test -z "$(ls -A $TESTDIR)"
 

tests/zfs-tests/tests/functional/trim/autotrim_config.ksh

@@ -57,7 +57,7 @@ log_onexit cleanup
 
 # Minimum trim size is decreased to verify all trim sizes.
 typeset trim_extent_bytes_min=$(get_tunable TRIM_EXTENT_BYTES_MIN)
-log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 4096
+log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 512
 
 # Reduced TRIM_TXG_BATCH to make trimming more frequent.
 typeset trim_txg_batch=$(get_tunable TRIM_TXG_BATCH)

tests/zfs-tests/tests/functional/trim/autotrim_integrity.ksh

@@ -54,7 +54,7 @@ log_onexit cleanup
 
 # Minimum trim size is decreased to verify all trim sizes.
 typeset trim_extent_bytes_min=$(get_tunable TRIM_EXTENT_BYTES_MIN)
-log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 4096
+log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 512
 
 # Reduced TRIM_TXG_BATCH to make trimming more frequent.
 typeset trim_txg_batch=$(get_tunable TRIM_TXG_BATCH)

tests/zfs-tests/tests/functional/trim/autotrim_trim_integrity.ksh

@@ -55,7 +55,7 @@ log_onexit cleanup
 
 # Minimum trim size is decreased to verify all trim sizes.
 typeset trim_extent_bytes_min=$(get_tunable TRIM_EXTENT_BYTES_MIN)
-log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 4096
+log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 512
 
 # Reduced TRIM_TXG_BATCH to make trimming more frequent.
 typeset trim_txg_batch=$(get_tunable TRIM_TXG_BATCH)

tests/zfs-tests/tests/functional/trim/trim_config.ksh

@@ -57,7 +57,7 @@ log_onexit cleanup
 
 # Minimum trim size is decreased to verify all trim sizes.
 typeset trim_extent_bytes_min=$(get_tunable TRIM_EXTENT_BYTES_MIN)
-log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 4096
+log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 512
 
 # Reduced TRIM_TXG_BATCH to make trimming more frequent.
 typeset trim_txg_batch=$(get_tunable TRIM_TXG_BATCH)

tests/zfs-tests/tests/functional/trim/trim_integrity.ksh

@@ -54,7 +54,7 @@ log_onexit cleanup
 
 # Minimum trim size is decreased to verify all trim sizes.
 typeset trim_extent_bytes_min=$(get_tunable TRIM_EXTENT_BYTES_MIN)
-log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 4096
+log_must set_tunable64 TRIM_EXTENT_BYTES_MIN 512
 
 # Reduced TRIM_TXG_BATCH to make trimming more frequent.
 typeset trim_txg_batch=$(get_tunable TRIM_TXG_BATCH)