Merge 4823cc2492 into 17dd66deda

cmd/zpool/zpool_main.c

@@ -512,7 +512,8 @@ get_usage(zpool_help_t idx)
 		return (gettext("\tinitialize [-c | -s | -u] [-w] <pool> "
 		    "[<device> ...]\n"));
 	case HELP_SCRUB:
-		return (gettext("\tscrub [-s | -p] [-w] [-e] <pool> ...\n"));
+		return (gettext("\tscrub [-s | -p] [-w] [-e] [-C] [-R] "
+		    "<pool> ...\n"));
 	case HELP_RESILVER:
 		return (gettext("\tresilver <pool> ...\n"));
 	case HELP_TRIM:
@@ -8399,11 +8400,13 @@ wait_callback(zpool_handle_t *zhp, void *data)
 }
 
 /*
- * zpool scrub [-s | -p] [-w] [-e] <pool> ...
+ * zpool scrub [-s | -p] [-w] [-e] [-C] <pool> ...
  *
+ *	-C	Scrub from last saved txg.
  *	-e	Only scrub blocks in the error log.
  *	-s	Stop.  Stops any in-progress scrub.
  *	-p	Pause. Pause in-progress scrub.
+ *	-R	Scrub only recent data.
  *	-w	Wait.  Blocks until scrub has completed.
  */
 int
@@ -8420,10 +8423,15 @@ zpool_do_scrub(int argc, char **argv)
 	boolean_t is_error_scrub = B_FALSE;
 	boolean_t is_pause = B_FALSE;
 	boolean_t is_stop = B_FALSE;
+	boolean_t is_txg_continue = B_FALSE;
+	boolean_t is_recent_scrub = B_FALSE;
 
 	/* check options */
-	while ((c = getopt(argc, argv, "spwe")) != -1) {
+	while ((c = getopt(argc, argv, "spweCR")) != -1) {
 		switch (c) {
+		case 'C':
+			is_txg_continue = B_TRUE;
+			break;
 		case 'e':
 			is_error_scrub = B_TRUE;
 			break;
@@ -8433,6 +8441,9 @@ zpool_do_scrub(int argc, char **argv)
 		case 'p':
 			is_pause = B_TRUE;
 			break;
+		case 'R':
+			is_recent_scrub = B_TRUE;
+			break;
 		case 'w':
 			wait = B_TRUE;
 			break;
@@ -8447,14 +8458,43 @@ zpool_do_scrub(int argc, char **argv)
 		(void) fprintf(stderr, gettext("invalid option "
 		    "combination :-s and -p are mutually exclusive\n"));
 		usage(B_FALSE);
+	} else if (is_pause && is_txg_continue) {
+		(void) fprintf(stderr, gettext("invalid option "
+		    "combination :-p and -C are mutually exclusive\n"));
+		usage(B_FALSE);
+	} else if (is_stop && is_txg_continue) {
+		(void) fprintf(stderr, gettext("invalid option "
+		    "combination :-s and -C are mutually exclusive\n"));
+		usage(B_FALSE);
+	} else if (is_error_scrub && is_txg_continue) {
+		(void) fprintf(stderr, gettext("invalid option "
+		    "combination :-e and -C are mutually exclusive\n"));
+		usage(B_FALSE);
+	} else if (is_pause && is_recent_scrub) {
+		(void) fprintf(stderr, gettext("invalid option "
+		    "combination :-p and -R are mutually exclusive\n"));
+		usage(B_FALSE);
+	} else if (is_stop && is_recent_scrub) {
+		(void) fprintf(stderr, gettext("invalid option "
+		    "combination :-s and -R are mutually exclusive\n"));
+		usage(B_FALSE);
+	} else if (is_error_scrub && is_recent_scrub) {
+		(void) fprintf(stderr, gettext("invalid option "
+		    "combination :-e and -R are mutually exclusive\n"));
+		usage(B_FALSE);
 	} else {
-		if (is_error_scrub)
+		if (is_error_scrub) {
 			cb.cb_type = POOL_SCAN_ERRORSCRUB;
+		}
 
-		if (is_pause) {
+		if (is_recent_scrub) {
+			cb.cb_scrub_cmd = POOL_SCRUB_RECENT_TXGS;
+		} else if (is_pause) {
 			cb.cb_scrub_cmd = POOL_SCRUB_PAUSE;
 		} else if (is_stop) {
 			cb.cb_type = POOL_SCAN_NONE;
+		} else if (is_txg_continue) {
+			cb.cb_scrub_cmd = POOL_SCRUB_FROM_LAST_TXG;
 		} else {
 			cb.cb_scrub_cmd = POOL_SCRUB_NORMAL;
 		}

include/sys/dmu.h

@@ -381,6 +381,7 @@ typedef struct dmu_buf {
 #define	DMU_POOL_CREATION_VERSION	"creation_version"
 #define	DMU_POOL_SCAN			"scan"
 #define	DMU_POOL_ERRORSCRUB		"error_scrub"
+#define	DMU_POOL_LAST_SCRUBBED_TXG	"last_scrubbed_txg"
 #define	DMU_POOL_FREE_BPOBJ		"free_bpobj"
 #define	DMU_POOL_BPTREE_OBJ		"bptree_obj"
 #define	DMU_POOL_EMPTY_BPOBJ		"empty_bpobj"

include/sys/dsl_scan.h

@@ -44,6 +44,7 @@ struct dsl_pool;
 struct dmu_tx;
 
 extern int zfs_scan_suspend_progress;
+extern uint_t zfs_scrub_recent_txgs;
 
 /*
  * All members of this structure must be uint64_t, for byteswap
@@ -189,7 +190,8 @@ void dsl_scan_setup_sync(void *, dmu_tx_t *);
 void dsl_scan_fini(struct dsl_pool *dp);
 void dsl_scan_sync(struct dsl_pool *, dmu_tx_t *);
 int dsl_scan_cancel(struct dsl_pool *);
-int dsl_scan(struct dsl_pool *, pool_scan_func_t);
+int dsl_scan(struct dsl_pool *, pool_scan_func_t, uint64_t starttxg,
+    uint64_t txgend);
 void dsl_scan_assess_vdev(struct dsl_pool *dp, vdev_t *vd);
 boolean_t dsl_scan_scrubbing(const struct dsl_pool *dp);
 boolean_t dsl_errorscrubbing(const struct dsl_pool *dp);

include/sys/fs/zfs.h

@@ -261,6 +261,7 @@ typedef enum {
 	ZPOOL_PROP_DEDUP_TABLE_SIZE,
 	ZPOOL_PROP_DEDUP_TABLE_QUOTA,
 	ZPOOL_PROP_DEDUPCACHED,
+	ZPOOL_PROP_LAST_SCRUBBED_TXG,
 	ZPOOL_NUM_PROPS
 } zpool_prop_t;
 
@@ -1075,6 +1076,8 @@ typedef enum pool_scan_func {
 typedef enum pool_scrub_cmd {
 	POOL_SCRUB_NORMAL = 0,
 	POOL_SCRUB_PAUSE,
+	POOL_SCRUB_FROM_LAST_TXG,
+	POOL_SCRUB_RECENT_TXGS,
 	POOL_SCRUB_FLAGS_END
 } pool_scrub_cmd_t;
 

include/sys/spa.h

@@ -821,6 +821,8 @@ extern void spa_l2cache_drop(spa_t *spa);
 
 /* scanning */
 extern int spa_scan(spa_t *spa, pool_scan_func_t func);
+extern int spa_scan_range(spa_t *spa, pool_scan_func_t func, uint64_t txgstart,
+    uint64_t txgend);
 extern int spa_scan_stop(spa_t *spa);
 extern int spa_scrub_pause_resume(spa_t *spa, pool_scrub_cmd_t flag);
 
@@ -1067,6 +1069,7 @@ extern uint64_t spa_get_deadman_failmode(spa_t *spa);
 extern void spa_set_deadman_failmode(spa_t *spa, const char *failmode);
 extern boolean_t spa_suspended(spa_t *spa);
 extern uint64_t spa_bootfs(spa_t *spa);
+extern uint64_t spa_get_last_scrubbed_txg(spa_t *spa);
 extern uint64_t spa_delegation(spa_t *spa);
 extern objset_t *spa_meta_objset(spa_t *spa);
 extern space_map_t *spa_syncing_log_sm(spa_t *spa);

include/sys/spa_impl.h

@@ -318,6 +318,7 @@ struct spa {
 	uint64_t	spa_scan_pass_scrub_spent_paused; /* total paused */
 	uint64_t	spa_scan_pass_exam;	/* examined bytes per pass */
 	uint64_t	spa_scan_pass_issued;	/* issued bytes per pass */
+	uint64_t	spa_scrubbed_last_txg;	/* last txg scrubbed */
 
 	/* error scrub pause time in milliseconds */
 	uint64_t	spa_scan_pass_errorscrub_pause;

lib/libzfs/libzfs.abi

@@ -2963,7 +2963,8 @@
       <enumerator name='ZPOOL_PROP_DEDUP_TABLE_SIZE' value='36'/>
       <enumerator name='ZPOOL_PROP_DEDUP_TABLE_QUOTA' value='37'/>
       <enumerator name='ZPOOL_PROP_DEDUPCACHED' value='38'/>
-      <enumerator name='ZPOOL_NUM_PROPS' value='39'/>
+      <enumerator name='ZPOOL_PROP_LAST_SCRUBBED_TXG' value='39'/>
+      <enumerator name='ZPOOL_NUM_PROPS' value='40'/>
     </enum-decl>
     <typedef-decl name='zpool_prop_t' type-id='af1ba157' id='5d0c23fb'/>
     <typedef-decl name='regoff_t' type-id='95e97e5e' id='54a2a2a8'/>
@@ -5815,7 +5816,9 @@
       <underlying-type type-id='9cac1fee'/>
       <enumerator name='POOL_SCRUB_NORMAL' value='0'/>
       <enumerator name='POOL_SCRUB_PAUSE' value='1'/>
-      <enumerator name='POOL_SCRUB_FLAGS_END' value='2'/>
+      <enumerator name='POOL_SCRUB_FROM_LAST_TXG' value='2'/>
+      <enumerator name='POOL_SCRUB_RECENT_TXGS' value='3'/>
+      <enumerator name='POOL_SCRUB_FLAGS_END' value='4'/>
     </enum-decl>
     <typedef-decl name='pool_scrub_cmd_t' type-id='a1474cbd' id='b51cf3c2'/>
     <enum-decl name='zpool_errata' id='d9abbf54'>

lib/libzfs/libzfs_pool.c

@@ -378,6 +378,7 @@ zpool_get_prop(zpool_handle_t *zhp, zpool_prop_t prop, char *buf,
 		case ZPOOL_PROP_BCLONEUSED:
 		case ZPOOL_PROP_DEDUP_TABLE_SIZE:
 		case ZPOOL_PROP_DEDUPCACHED:
+		case ZPOOL_PROP_LAST_SCRUBBED_TXG:
 			if (literal)
 				(void) snprintf(buf, len, "%llu",
 				    (u_longlong_t)intval);

man/man4/zfs.4

@@ -1986,6 +1986,9 @@ working on a scrub between TXG flushes.
 .It Sy zfs_scrub_error_blocks_per_txg Ns = Ns Sy 4096 Pq uint
 Error blocks to be scrubbed in one txg.
 .
+.It Sy zfs_scrub_recent_txgs Ns = Ns Sy 256 Pq uint
+Number of txgs to be considered as recent when performing a recent data scrub.
+.
 .It Sy zfs_scan_checkpoint_intval Ns = Ns Sy 7200 Ns s Po 2 hour Pc Pq uint
 To preserve progress across reboots, the sequential scan algorithm periodically
 needs to stop metadata scanning and issue all the verification I/O to disk.

man/man7/zpoolprops.7

@@ -135,6 +135,14 @@ A unique identifier for the pool.
 The current health of the pool.
 Health can be one of
 .Sy ONLINE , DEGRADED , FAULTED , OFFLINE, REMOVED , UNAVAIL .
+.It Sy last_scrubbed_txg
+Indicates the transaction group (TXG) up to which the most recent scrub
+operation has checked and repaired the dataset.
+This provides insight into the data integrity status of their pool at
+a specific point in time.
+The
+.Xr zpool-scrub 8
+might be used to utilize this property.
 .It Sy leaked
 Space not released while
 .Sy freeing

man/man8/zpool-scrub.8

@@ -39,6 +39,8 @@
 .Op Fl s Ns | Ns Fl p
 .Op Fl w
 .Op Fl e
+.Op Fl C
+.Op Fl R
 .Ar pool Ns …
 .
 .Sh DESCRIPTION
@@ -114,6 +116,14 @@ The pool must have been scrubbed at least once with the
 feature enabled to use this option.
 Error scrubbing cannot be run simultaneously with regular scrubbing or
 resilvering, nor can it be run when a regular scrub is paused.
+.It Fl C
+Continue scrub from last saved txg (see zpool
+.Sy last_scrubbed_txg
+property).
+.It Fl R
+Scrub only recent data (this can be controlled by the
+.Sy zfs_scrub_recent_txgs
+parameter).
 .El
 .Sh EXAMPLES
 .Ss Example 1

module/zcommon/zpool_prop.c

@@ -129,6 +129,10 @@ zpool_prop_init(void)
 	    0, PROP_READONLY, ZFS_TYPE_POOL, "<size>", "DDTSIZE", B_FALSE,
 	    sfeatures);
 
+	zprop_register_number(ZPOOL_PROP_LAST_SCRUBBED_TXG,
+	    "last_scrubbed_txg", 0, PROP_READONLY, ZFS_TYPE_POOL, "<txg>",
+	    "LAST_SCRUBBED_TXG", B_FALSE, sfeatures);
+
 	/* default number properties */
 	zprop_register_number(ZPOOL_PROP_VERSION, "version", SPA_VERSION,
 	    PROP_DEFAULT, ZFS_TYPE_POOL, "<version>", "VERSION", B_FALSE,

module/zfs/dsl_scan.c

@@ -228,6 +228,9 @@ static int zfs_resilver_disable_defer = B_FALSE;
 	((scn)->scn_phys.scn_func == POOL_SCAN_SCRUB || \
 	(scn)->scn_phys.scn_func == POOL_SCAN_RESILVER)
 
+#define	DSL_SCAN_IS_SCRUB(scn)		\
+	((scn)->scn_phys.scn_func == POOL_SCAN_SCRUB)
+
 /*
  * Enable/disable the processing of the free_bpobj object.
  */
@@ -236,6 +239,9 @@ static int zfs_free_bpobj_enabled = 1;
 /* Error blocks to be scrubbed in one txg. */
 static uint_t zfs_scrub_error_blocks_per_txg = 1 << 12;
 
+/* The number of TXGs should be scrubbed while scrubbing recent data. */
+uint_t zfs_scrub_recent_txgs = 256;
+
 /* the order has to match pool_scan_type */
 static scan_cb_t *scan_funcs[POOL_SCAN_FUNCS] = {
 	NULL,
@@ -848,18 +854,24 @@ dsl_scan_setup_check(void *arg, dmu_tx_t *tx)
 	return (0);
 }
 
+typedef struct {
+	pool_scan_func_t func;
+	uint64_t	 txgstart;
+	uint64_t	 txgend;
+} setup_sync_arg_t;
+
 void
 dsl_scan_setup_sync(void *arg, dmu_tx_t *tx)
 {
-	(void) arg;
+	setup_sync_arg_t *setup_sync_arg = (setup_sync_arg_t *)arg;
 	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(!dsl_scan_is_running(scn));
-	ASSERT(*funcp > POOL_SCAN_NONE && *funcp < POOL_SCAN_FUNCS);
+	ASSERT(setup_sync_arg->func > POOL_SCAN_NONE &&
+	    setup_sync_arg->func < POOL_SCAN_FUNCS);
 	memset(&scn->scn_phys, 0, sizeof (scn->scn_phys));
 
 	/*
@@ -869,10 +881,14 @@ dsl_scan_setup_sync(void *arg, dmu_tx_t *tx)
 	memset(&scn->errorscrub_phys, 0, sizeof (scn->errorscrub_phys));
 	dsl_errorscrub_sync_state(scn, tx);
 
-	scn->scn_phys.scn_func = *funcp;
+	scn->scn_phys.scn_func = setup_sync_arg->func;
 	scn->scn_phys.scn_state = DSS_SCANNING;
-	scn->scn_phys.scn_min_txg = 0;
+	scn->scn_phys.scn_min_txg = setup_sync_arg->txgstart;
-	scn->scn_phys.scn_max_txg = tx->tx_txg;
+	if (setup_sync_arg->txgend == 0) {
+		scn->scn_phys.scn_max_txg = tx->tx_txg;
+	} else {
+		scn->scn_phys.scn_max_txg = setup_sync_arg->txgend;
+	}
 	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;
@@ -959,7 +975,7 @@ dsl_scan_setup_sync(void *arg, dmu_tx_t *tx)
 
 	spa_history_log_internal(spa, "scan setup", tx,
 	    "func=%u mintxg=%llu maxtxg=%llu",
-	    *funcp, (u_longlong_t)scn->scn_phys.scn_min_txg,
+	    setup_sync_arg->func, (u_longlong_t)scn->scn_phys.scn_min_txg,
 	    (u_longlong_t)scn->scn_phys.scn_max_txg);
 }
 
@@ -969,10 +985,16 @@ dsl_scan_setup_sync(void *arg, dmu_tx_t *tx)
  * error scrub.
  */
 int
-dsl_scan(dsl_pool_t *dp, pool_scan_func_t func)
+dsl_scan(dsl_pool_t *dp, pool_scan_func_t func, uint64_t txgstart,
+    uint64_t txgend)
 {
 	spa_t *spa = dp->dp_spa;
 	dsl_scan_t *scn = dp->dp_scan;
+	setup_sync_arg_t setup_sync_arg;
+
+	if (func != POOL_SCAN_SCRUB && (txgstart != 0 || txgend != 0)) {
+		return (EINVAL);
+	}
 
 	/*
 	 * Purge all vdev caches and probe all devices.  We do this here
@@ -1023,8 +1045,13 @@ dsl_scan(dsl_pool_t *dp, pool_scan_func_t func)
 		return (SET_ERROR(err));
 	}
 
+	setup_sync_arg.func = func;
+	setup_sync_arg.txgstart = txgstart;
+	setup_sync_arg.txgend = txgend;
+
 	return (dsl_sync_task(spa_name(spa), dsl_scan_setup_check,
-	    dsl_scan_setup_sync, &func, 0, ZFS_SPACE_CHECK_EXTRA_RESERVED));
+	    dsl_scan_setup_sync, &setup_sync_arg, 0,
+	    ZFS_SPACE_CHECK_EXTRA_RESERVED));
 }
 
 static void
@@ -1112,15 +1139,24 @@ dsl_scan_done(dsl_scan_t *scn, boolean_t complete, dmu_tx_t *tx)
 
 	spa_notify_waiters(spa);
 
-	if (dsl_scan_restarting(scn, tx))
+	if (dsl_scan_restarting(scn, tx)) {
 		spa_history_log_internal(spa, "scan aborted, restarting", tx,
 		    "errors=%llu", (u_longlong_t)spa_approx_errlog_size(spa));
-	else if (!complete)
+	} else if (!complete) {
 		spa_history_log_internal(spa, "scan cancelled", tx,
 		    "errors=%llu", (u_longlong_t)spa_approx_errlog_size(spa));
-	else
+	} else {
 		spa_history_log_internal(spa, "scan done", tx,
 		    "errors=%llu", (u_longlong_t)spa_approx_errlog_size(spa));
+		if (DSL_SCAN_IS_SCRUB(scn)) {
+			VERIFY0(zap_update(dp->dp_meta_objset,
+			    DMU_POOL_DIRECTORY_OBJECT,
+			    DMU_POOL_LAST_SCRUBBED_TXG,
+			    sizeof (uint64_t), 1,
+			    &scn->scn_phys.scn_max_txg, tx));
+			spa->spa_scrubbed_last_txg = scn->scn_phys.scn_max_txg;
+		}
+	}
 
 	if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) {
 		spa->spa_scrub_active = B_FALSE;
@@ -4301,13 +4337,16 @@ dsl_scan_sync(dsl_pool_t *dp, dmu_tx_t *tx)
 	 */
 	if (dsl_scan_restarting(scn, tx) ||
 	    (spa->spa_resilver_deferred && zfs_resilver_disable_defer)) {
-		pool_scan_func_t func = POOL_SCAN_SCRUB;
+		setup_sync_arg_t setup_sync_arg = {
+			.func = POOL_SCAN_SCRUB,
+		};
 		dsl_scan_done(scn, B_FALSE, tx);
 		if (vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL))
-			func = POOL_SCAN_RESILVER;
+			setup_sync_arg.func = POOL_SCAN_RESILVER;
 		zfs_dbgmsg("restarting scan func=%u on %s txg=%llu",
-		    func, dp->dp_spa->spa_name, (longlong_t)tx->tx_txg);
+		    setup_sync_arg.func, dp->dp_spa->spa_name,
-		dsl_scan_setup_sync(&func, tx);
+		    (longlong_t)tx->tx_txg);
+		dsl_scan_setup_sync(&setup_sync_arg, tx);
 	}
 
 	/*
@@ -5287,4 +5326,7 @@ ZFS_MODULE_PARAM(zfs, zfs_, resilver_disable_defer, INT, ZMOD_RW,
 
 ZFS_MODULE_PARAM(zfs, zfs_, scrub_error_blocks_per_txg, UINT, ZMOD_RW,
 	"Error blocks to be scrubbed in one txg");
+
+ZFS_MODULE_PARAM(zfs, zfs_, scrub_recent_txgs, UINT, ZMOD_RW,
+	"The number of TXGs should be scrubbed while scrubbing recent data");
 /* END CSTYLED */

module/zfs/spa.c

@@ -458,6 +458,9 @@ spa_prop_get_config(spa_t *spa, nvlist_t **nvp)
 		spa_prop_add_list(*nvp, ZPOOL_PROP_DEDUP_TABLE_SIZE, NULL,
 		    ddt_get_ddt_dsize(spa), src);
 
+		spa_prop_add_list(*nvp, ZPOOL_PROP_LAST_SCRUBBED_TXG, NULL,
+		    spa_get_last_scrubbed_txg(spa), src);
+
 		spa_prop_add_list(*nvp, ZPOOL_PROP_HEALTH, NULL,
 		    rvd->vdev_state, src);
 
@@ -4737,6 +4740,12 @@ spa_ld_get_props(spa_t *spa)
 	if (error != 0 && error != ENOENT)
 		return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
 
+	/* Load the last scrubbed txg. */
+	error = spa_dir_prop(spa, DMU_POOL_LAST_SCRUBBED_TXG,
+	    &spa->spa_scrubbed_last_txg, B_FALSE);
+	if (error != 0 && error != ENOENT)
+		return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
+
 	/*
 	 * Load the livelist deletion field. If a livelist is queued for
 	 * deletion, indicate that in the spa
@@ -8877,6 +8886,13 @@ spa_scan_stop(spa_t *spa)
 
 int
 spa_scan(spa_t *spa, pool_scan_func_t func)
+{
+	return (spa_scan_range(spa, func, 0, 0));
+}
+
+int
+spa_scan_range(spa_t *spa, pool_scan_func_t func, uint64_t txgstart,
+    uint64_t txgend)
 {
 	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0);
 
@@ -8887,6 +8903,9 @@ spa_scan(spa_t *spa, pool_scan_func_t func)
 	    !spa_feature_is_enabled(spa, SPA_FEATURE_RESILVER_DEFER))
 		return (SET_ERROR(ENOTSUP));
 
+	if (func != POOL_SCAN_SCRUB && (txgstart != 0 || txgend != 0))
+		return (SET_ERROR(ENOTSUP));
+
 	/*
 	 * If a resilver was requested, but there is no DTL on a
 	 * writeable leaf device, we have nothing to do.
@@ -8901,7 +8920,7 @@ spa_scan(spa_t *spa, pool_scan_func_t func)
 	    !spa_feature_is_enabled(spa, SPA_FEATURE_HEAD_ERRLOG))
 		return (SET_ERROR(ENOTSUP));
 
-	return (dsl_scan(spa->spa_dsl_pool, func));
+	return (dsl_scan(spa->spa_dsl_pool, func, txgstart, txgend));
 }
 
 /*
@@ -10982,6 +11001,7 @@ EXPORT_SYMBOL(spa_l2cache_drop);
 
 /* scanning */
 EXPORT_SYMBOL(spa_scan);
+EXPORT_SYMBOL(spa_scan_range);
 EXPORT_SYMBOL(spa_scan_stop);
 
 /* spa syncing */

module/zfs/spa_misc.c

@@ -2679,6 +2679,12 @@ spa_mode(spa_t *spa)
 	return (spa->spa_mode);
 }
 
+uint64_t
+spa_get_last_scrubbed_txg(spa_t *spa)
+{
+	return (spa->spa_scrubbed_last_txg);
+}
+
 uint64_t
 spa_bootfs(spa_t *spa)
 {

module/zfs/zfs_ioctl.c

@@ -1719,6 +1719,19 @@ zfs_ioc_pool_scrub(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl)
 		error = spa_scrub_pause_resume(spa, POOL_SCRUB_PAUSE);
 	} else if (scan_type == POOL_SCAN_NONE) {
 		error = spa_scan_stop(spa);
+	} else if (scan_cmd == POOL_SCRUB_FROM_LAST_TXG) {
+		error = spa_scan_range(spa, scan_type,
+		    spa_get_last_scrubbed_txg(spa), 0);
+	} else if (scan_cmd == POOL_SCRUB_RECENT_TXGS) {
+		uint64_t start;
+
+		start = 0;
+		if (spa_last_synced_txg(spa) > zfs_scrub_recent_txgs) {
+			start = spa_last_synced_txg(spa) -
+			    zfs_scrub_recent_txgs;
+		}
+
+		error = spa_scan_range(spa, scan_type, start, 0);
 	} else {
 		error = spa_scan(spa, scan_type);
 	}

tests/zfs-tests/tests/Makefile.am

@@ -1218,6 +1218,7 @@ nobase_dist_datadir_zfs_tests_tests_SCRIPTS += \
 	functional/cli_root/zpool_scrub/zpool_scrub_multiple_copies.ksh \
 	functional/cli_root/zpool_scrub/zpool_scrub_offline_device.ksh \
 	functional/cli_root/zpool_scrub/zpool_scrub_print_repairing.ksh \
+	functional/cli_root/zpool_scrub/zpool_scrub_txg_continue_from_last.ksh \
 	functional/cli_root/zpool_scrub/zpool_error_scrub_001_pos.ksh \
 	functional/cli_root/zpool_scrub/zpool_error_scrub_002_pos.ksh \
 	functional/cli_root/zpool_scrub/zpool_error_scrub_003_pos.ksh \

tests/zfs-tests/tests/functional/cli_root/zpool_scrub/zpool_scrub_txg_continue_from_last.ksh

@@ -0,0 +1,104 @@
+#!/bin/ksh -p
+#
+# 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) 2023, Klara Inc.
+#
+# This software was developed by
+# Mariusz Zaborski <mariusz.zaborski@klarasystems.com>
+# under sponsorship from Wasabi Technology, Inc. and Klara Inc.
+
+. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/tests/functional/cli_root/zpool_scrub/zpool_scrub.cfg
+. $STF_SUITE/tests/functional/cli_root/zpool_import/zpool_import.kshlib
+
+#
+# DESCRIPTION:
+#	Verify scrub -C
+#
+# STRATEGY:
+#      1. Create a pool and create one file.
+#      2. Verify that the last_txg_scrub is 0.
+#      3. Run scrub.
+#      4. Verify that the last_txg_scrub is set.
+#      5. Create second file.
+#      6. Invalidate both files.
+#      7. Run scrub only from last point.
+#      8. Verify that only one file, that was created with newer txg,
+#         was detected.
+#
+
+verify_runnable "global"
+
+function cleanup
+{
+	log_must zinject -c all
+	log_must rm -f $mntpnt/f1
+	log_must rm -f $mntpnt/f2
+}
+
+log_onexit cleanup
+
+log_assert "Verify scrub -C."
+
+# Create one file.
+mntpnt=$(get_prop mountpoint $TESTPOOL/$TESTFS)
+
+log_must file_write -b 1048576 -c 10 -o create -d 0 -f $mntpnt/f1
+log_must sync_pool $TESTPOOL true
+f1txg=$(get_last_txg_synced $TESTPOOL)
+
+# Verify that last_scrubbed_txg isn't set.
+zpoollasttxg=$(zpool get -H -o value last_scrubbed_txg $TESTPOOL)
+log_must [ $zpoollasttxg -eq 0 ]
+
+# Run scrub.
+log_must zpool scrub -w $TESTPOOL
+
+# Verify that last_scrubbed_txg is set.
+zpoollasttxg=$(zpool get -H -o value last_scrubbed_txg $TESTPOOL)
+log_must [ $zpoollasttxg -ne 0 ]
+
+# Create second file.
+log_must file_write -b 1048576 -c 10 -o create -d 0 -f $mntpnt/f2
+log_must sync_pool $TESTPOOL true
+f2txg=$(get_last_txg_synced $TESTPOOL)
+
+# Make sure that the sync txg are different.
+log_must [ $f1txg -ne $f2txg ]
+
+# Insert faults.
+log_must zinject -a -t data -e io -T read $mntpnt/f1
+log_must zinject -a -t data -e io -T read $mntpnt/f2
+
+# Run scrub from last saved point.
+log_must zpool scrub -w -C $TESTPOOL
+
+# Verify that only newer file was detected.
+log_mustnot eval "zpool status -v $TESTPOOL | grep '$mntpnt/f1'"
+log_must eval "zpool status -v $TESTPOOL | grep '$mntpnt/f2'"
+
+# Verify that both files are corrupted.
+log_must zpool scrub -w $TESTPOOL
+log_must eval "zpool status -v $TESTPOOL | grep '$mntpnt/f1'"
+log_must eval "zpool status -v $TESTPOOL | grep '$mntpnt/f2'"
+
+log_pass "Verified scrub -C show expected status."