[illumos-gate merge]

commit b4952e17e8858d3225793b28788278de9fe6038d
    3956 ::vdev -r should work with pipelines
    3957 ztest should update the cachefile before killing itself
    3958 multiple scans can lead to partial resilvering
    3959 ddt entries are not always resilvered
    3960 dsl_scan can skip over dedup-ed blocks if physical birth != logical birth
    3961 freed gang blocks are not resilvered and can cause pool to suspend
    3962 ztest should print out zfs debug buffer before exiting
commit be9000cc677e0a8d04e5be45c61d7370fc8c7b54
    3955 ztest failure: assertion refcount_count(&tx->tx_space_written) + delta <= tx->tx_space_towrite
commit 2c1e2b44148432fb7a509dd216a99299b6740250
    3949 ztest fault injection should avoid resilvering devices
    3950 ztest: deadman fires when we're doing a scan
    3951 ztest hang when running dedup test
    3952 ztest: ztest_reguid test and ztest_fault_inject don't place nice together
commit 98144673ce45bddc6d5dbe7e2afab720c660b5d7
    3992 mdb ::stacks segv

16 files changed, 289 insertions, 103 deletions

diff --git a/usr/src/cmd/mdb/common/modules/genunix/findstack.c b/usr/src/cmd/mdb/common/modules/genunix/findstack.c
index 223764eda9..12b90b8f7d 100644
--- a/usr/src/cmd/mdb/common/modules/genunix/findstack.c
+++ b/usr/src/cmd/mdb/common/modules/genunix/findstack.c
@@ -21,6 +21,7 @@
 
 /*
  * Copyright (c) 1999, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2013, Josef 'Jeff' Sipek <jeffpc@josefsipek.net>
  */
 
 #include <mdb/mdb_modapi.h>
@@ -340,6 +341,8 @@ stacks_cleanup(int force)
 			mdb_free(stacks_array,
 			    stacks_array_size * sizeof (*stacks_array));
 
+		mdb_free(stacks_hash, STACKS_HSIZE * sizeof (*stacks_hash));
+
 	} else if (stacks_array != NULL) {
 		for (idx = 0; idx < stacks_array_size; idx++) {
 			if ((cur = stacks_array[idx]) != NULL) {
@@ -360,6 +363,8 @@ stacks_cleanup(int force)
 
 	stacks_array_size = 0;
 	stacks_state = STACKS_STATE_CLEAN;
+	stacks_hash = NULL;
+	stacks_array = NULL;
 }
 
 /*ARGSUSED*/
diff --git a/usr/src/cmd/mdb/common/modules/zfs/zfs.c b/usr/src/cmd/mdb/common/modules/zfs/zfs.c
index b04b05d75a..b3cf1d5f0b 100644
--- a/usr/src/cmd/mdb/common/modules/zfs/zfs.c
+++ b/usr/src/cmd/mdb/common/modules/zfs/zfs.c
@@ -1140,7 +1140,7 @@ do_print_vdev(uintptr_t addr, int flags, int depth, int stats,
 	}
 
 	if (flags & DCMD_PIPE_OUT) {
-		mdb_printf("%#lr", addr);
+		mdb_printf("%#lr\n", addr);
 	} else {
 		if (vdev.vdev_path != NULL) {
 			if (mdb_readstr(desc, sizeof (desc),
diff --git a/usr/src/cmd/ztest/ztest.c b/usr/src/cmd/ztest/ztest.c
index 3aca1fe0c9..390061a8c5 100644
--- a/usr/src/cmd/ztest/ztest.c
+++ b/usr/src/cmd/ztest/ztest.c
@@ -20,7 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012 by Delphix. All rights reserved.
+ * Copyright (c) 2013 by Delphix. All rights reserved.
  * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
  * Copyright (c) 2013 Steven Hartland. All rights reserved.
  */
@@ -184,6 +184,7 @@ static const ztest_shared_opts_t ztest_opts_defaults = {
 
 extern uint64_t metaslab_gang_bang;
 extern uint64_t metaslab_df_alloc_threshold;
+extern uint64_t zfs_deadman_synctime;
 
 static ztest_shared_opts_t *ztest_shared_opts;
 static ztest_shared_opts_t ztest_opts;
@@ -363,7 +364,7 @@ ztest_info_t ztest_info[] = {
 	{ ztest_fault_inject,			1,	&zopt_sometimes	},
 	{ ztest_ddt_repair,			1,	&zopt_sometimes	},
 	{ ztest_dmu_snapshot_hold,		1,	&zopt_sometimes	},
-	{ ztest_reguid,				1,	&zopt_sometimes },
+	{ ztest_reguid,				1,	&zopt_rarely	},
 	{ ztest_spa_rename,			1,	&zopt_rarely	},
 	{ ztest_scrub,				1,	&zopt_rarely	},
 	{ ztest_spa_upgrade,			1,	&zopt_rarely	},
@@ -766,6 +767,16 @@ ztest_kill(ztest_shared_t *zs)
 {
 	zs->zs_alloc = metaslab_class_get_alloc(spa_normal_class(ztest_spa));
 	zs->zs_space = metaslab_class_get_space(spa_normal_class(ztest_spa));
+
+	/*
+	 * Before we kill off ztest, make sure that the config is updated.
+	 * See comment above spa_config_sync().
+	 */
+	mutex_enter(&spa_namespace_lock);
+	spa_config_sync(ztest_spa, B_FALSE, B_FALSE);
+	mutex_exit(&spa_namespace_lock);
+
+	zfs_dbgmsg_print(FTAG);
 	(void) kill(getpid(), SIGKILL);
 }
 
@@ -2730,7 +2741,7 @@ ztest_vdev_attach_detach(ztest_ds_t *zd, uint64_t id)
 	uint64_t leaf, top;
 	uint64_t ashift = ztest_get_ashift();
 	uint64_t oldguid, pguid;
-	size_t oldsize, newsize;
+	uint64_t oldsize, newsize;
 	char oldpath[MAXPATHLEN], newpath[MAXPATHLEN];
 	int replacing;
 	int oldvd_has_siblings = B_FALSE;
@@ -2889,8 +2900,8 @@ ztest_vdev_attach_detach(ztest_ds_t *zd, uint64_t id)
 	if (error != expected_error && expected_error != EBUSY) {
 		fatal(0, "attach (%s %llu, %s %llu, %d) "
 		    "returned %d, expected %d",
-		    oldpath, (longlong_t)oldsize, newpath,
-		    (longlong_t)newsize, replacing, error, expected_error);
+		    oldpath, oldsize, newpath,
+		    newsize, replacing, error, expected_error);
 	}
 
 	VERIFY(mutex_unlock(&ztest_vdev_lock) == 0);
@@ -3604,6 +3615,9 @@ ztest_dmu_read_write(ztest_ds_t *zd, uint64_t id)
 	else
 		dmu_tx_hold_write(tx, bigobj, bigoff, bigsize);
 
+	/* This accounts for setting the checksum/compression. */
+	dmu_tx_hold_bonus(tx, bigobj);
+
 	txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG);
 	if (txg == 0) {
 		umem_free(packbuf, packsize);
@@ -4754,6 +4768,14 @@ ztest_fault_inject(ztest_ds_t *zd, uint64_t id)
 	ASSERT(leaves >= 1);
 
 	/*
+	 * Grab the name lock as reader. There are some operations
+	 * which don't like to have their vdevs changed while
+	 * they are in progress (i.e. spa_change_guid). Those
+	 * operations will have grabbed the name lock as writer.
+	 */
+	(void) rw_rdlock(&ztest_name_lock);
+
+	/*
 	 * We need SCL_STATE here because we're going to look at vd0->vdev_tsd.
 	 */
 	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
@@ -4782,7 +4804,14 @@ ztest_fault_inject(ztest_ds_t *zd, uint64_t id)
 		if (vd0 != NULL && vd0->vdev_top->vdev_islog)
 			islog = B_TRUE;
 
-		if (vd0 != NULL && maxfaults != 1) {
+		/*
+		 * If the top-level vdev needs to be resilvered
+		 * then we only allow faults on the device that is
+		 * resilvering.
+		 */
+		if (vd0 != NULL && maxfaults != 1 &&
+		    (!vdev_resilver_needed(vd0->vdev_top, NULL, NULL) ||
+		    vd0->vdev_resilver_txg != 0)) {
 			/*
 			 * Make vd0 explicitly claim to be unreadable,
 			 * or unwriteable, or reach behind its back
@@ -4813,6 +4842,7 @@ ztest_fault_inject(ztest_ds_t *zd, uint64_t id)
 
 		if (sav->sav_count == 0) {
 			spa_config_exit(spa, SCL_STATE, FTAG);
+			(void) rw_unlock(&ztest_name_lock);
 			return;
 		}
 		vd0 = sav->sav_vdevs[ztest_random(sav->sav_count)];
@@ -4826,6 +4856,7 @@ ztest_fault_inject(ztest_ds_t *zd, uint64_t id)
 	}
 
 	spa_config_exit(spa, SCL_STATE, FTAG);
+	(void) rw_unlock(&ztest_name_lock);
 
 	/*
 	 * If we can tolerate two or more faults, or we're dealing
@@ -5290,16 +5321,33 @@ static void *
 ztest_deadman_thread(void *arg)
 {
 	ztest_shared_t *zs = arg;
-	int grace = 300;
-	hrtime_t delta;
+	spa_t *spa = ztest_spa;
+	hrtime_t delta, total = 0;
 
-	delta = (zs->zs_thread_stop - zs->zs_thread_start) / NANOSEC + grace;
+	for (;;) {
+		delta = (zs->zs_thread_stop - zs->zs_thread_start) /
+		    NANOSEC + zfs_deadman_synctime;
 
-	(void) poll(NULL, 0, (int)(1000 * delta));
+		(void) poll(NULL, 0, (int)(1000 * delta));
 
-	fatal(0, "failed to complete within %d seconds of deadline", grace);
+		/*
+		 * If the pool is suspended then fail immediately. Otherwise,
+		 * check to see if the pool is making any progress. If
+		 * vdev_deadman() discovers that there hasn't been any recent
+		 * I/Os then it will end up aborting the tests.
+		 */
+		if (spa_suspended(spa)) {
+			fatal(0, "aborting test after %llu seconds because "
+			    "pool has transitioned to a suspended state.",
+			    zfs_deadman_synctime);
+			return (NULL);
+		}
+		vdev_deadman(spa->spa_root_vdev);
 
-	return (NULL);
+		total += zfs_deadman_synctime;
+		(void) printf("ztest has been running for %lld seconds\n",
+		    total);
+	}
 }
 
 static void
@@ -5613,6 +5661,7 @@ ztest_run(ztest_shared_t *zs)
 
 	zs->zs_alloc = metaslab_class_get_alloc(spa_normal_class(spa));
 	zs->zs_space = metaslab_class_get_space(spa_normal_class(spa));
+	zfs_dbgmsg_print(FTAG);
 
 	umem_free(tid, ztest_opts.zo_threads * sizeof (thread_t));
 
@@ -6024,6 +6073,7 @@ main(int argc, char **argv)
 	(void) setvbuf(stdout, NULL, _IOLBF, 0);
 
 	dprintf_setup(&argc, argv);
+	zfs_deadman_synctime = 300;
 
 	ztest_fd_rand = open("/dev/urandom", O_RDONLY);
 	ASSERT3S(ztest_fd_rand, >=, 0);
diff --git a/usr/src/lib/libzpool/common/llib-lzpool b/usr/src/lib/libzpool/common/llib-lzpool
index d3864d2a9a..7e61b55a91 100644
--- a/usr/src/lib/libzpool/common/llib-lzpool
+++ b/usr/src/lib/libzpool/common/llib-lzpool
@@ -64,3 +64,4 @@
 extern uint64_t metaslab_gang_bang;
 extern uint64_t metaslab_df_alloc_threshold;
 extern boolean_t zfeature_checks_disable;
+extern uint64_t zfs_deadman_synctime;
diff --git a/usr/src/uts/common/fs/zfs/dmu_tx.c b/usr/src/uts/common/fs/zfs/dmu_tx.c
index 84e518dfd1..d0dd730ed4 100644
--- a/usr/src/uts/common/fs/zfs/dmu_tx.c
+++ b/usr/src/uts/common/fs/zfs/dmu_tx.c
@@ -450,12 +450,12 @@ dmu_tx_count_free(dmu_tx_hold_t *txh, uint64_t off, uint64_t len)
 		blkid = off >> dn->dn_datablkshift;
 		nblks = (len + dn->dn_datablksz - 1) >> dn->dn_datablkshift;
 
-		if (blkid >= dn->dn_maxblkid) {
+		if (blkid > dn->dn_maxblkid) {
 			rw_exit(&dn->dn_struct_rwlock);
 			return;
 		}
 		if (blkid + nblks > dn->dn_maxblkid)
-			nblks = dn->dn_maxblkid - blkid;
+			nblks = dn->dn_maxblkid - blkid + 1;
 
 	}
 	l0span = nblks;    /* save for later use to calc level > 1 overhead */
diff --git a/usr/src/uts/common/fs/zfs/dsl_scan.c b/usr/src/uts/common/fs/zfs/dsl_scan.c
index 34816986a1..e1859cc34c 100644
--- a/usr/src/uts/common/fs/zfs/dsl_scan.c
+++ b/usr/src/uts/common/fs/zfs/dsl_scan.c
@@ -194,6 +194,7 @@ dsl_scan_setup_sync(void *arg, dmu_tx_t *tx)
 	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;
+	scn->scn_done_txg = 0;
 	spa_scan_stat_init(spa);
 
 	if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) {
@@ -769,7 +770,7 @@ dsl_scan_visitbp(blkptr_t *bp, const zbookmark_t *zb,
 	 * Don't scan it now unless we need to because something
 	 * under it was modified.
 	 */
-	if (bp->blk_birth <= scn->scn_phys.scn_cur_max_txg) {
+	if (BP_PHYSICAL_BIRTH(bp) <= scn->scn_phys.scn_cur_max_txg) {
 		scan_funcs[scn->scn_phys.scn_func](dp, bp, zb);
 	}
 	if (buf)
@@ -1214,7 +1215,7 @@ dsl_scan_ddt_entry(dsl_scan_t *scn, enum zio_checksum checksum,
 
 	for (int p = 0; p < DDT_PHYS_TYPES; p++, ddp++) {
 		if (ddp->ddp_phys_birth == 0 ||
-		    ddp->ddp_phys_birth > scn->scn_phys.scn_cur_max_txg)
+		    ddp->ddp_phys_birth > scn->scn_phys.scn_max_txg)
 			continue;
 		ddt_bp_create(checksum, ddk, ddp, &bp);
 
@@ -1457,6 +1458,16 @@ dsl_scan_sync(dsl_pool_t *dp, dmu_tx_t *tx)
 	if (scn->scn_phys.scn_state != DSS_SCANNING)
 		return;
 
+	if (scn->scn_done_txg == tx->tx_txg) {
+		ASSERT(!scn->scn_pausing);
+		/* finished with scan. */
+		zfs_dbgmsg("txg %llu scan complete", tx->tx_txg);
+		dsl_scan_done(scn, B_TRUE, tx);
+		ASSERT3U(spa->spa_scrub_inflight, ==, 0);
+		dsl_scan_sync_state(scn, tx);
+		return;
+	}
+
 	if (scn->scn_phys.scn_ddt_bookmark.ddb_class <=
 	    scn->scn_phys.scn_ddt_class_max) {
 		zfs_dbgmsg("doing scan sync txg %llu; "
@@ -1492,9 +1503,9 @@ dsl_scan_sync(dsl_pool_t *dp, dmu_tx_t *tx)
 	    (longlong_t)NSEC2MSEC(gethrtime() - scn->scn_sync_start_time));
 
 	if (!scn->scn_pausing) {
-		/* finished with scan. */
-		zfs_dbgmsg("finished scan txg %llu", (longlong_t)tx->tx_txg);
-		dsl_scan_done(scn, B_TRUE, tx);
+		scn->scn_done_txg = tx->tx_txg + 1;
+		zfs_dbgmsg("txg %llu traversal complete, waiting till txg %llu",
+		    tx->tx_txg, scn->scn_done_txg);
 	}
 
 	if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) {
diff --git a/usr/src/uts/common/fs/zfs/spa.c b/usr/src/uts/common/fs/zfs/spa.c
index 7334d39516..13eaaecbf7 100644
--- a/usr/src/uts/common/fs/zfs/spa.c
+++ b/usr/src/uts/common/fs/zfs/spa.c
@@ -757,6 +757,7 @@ spa_change_guid(spa_t *spa)
 	int error;
 	uint64_t guid;
 
+	mutex_enter(&spa->spa_vdev_top_lock);
 	mutex_enter(&spa_namespace_lock);
 	guid = spa_generate_guid(NULL);
 
@@ -769,6 +770,7 @@ spa_change_guid(spa_t *spa)
 	}
 
 	mutex_exit(&spa_namespace_lock);
+	mutex_exit(&spa->spa_vdev_top_lock);
 
 	return (error);
 }
@@ -4444,7 +4446,7 @@ spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing)
 	}
 
 	/* mark the device being resilvered */
-	newvd->vdev_resilvering = B_TRUE;
+	newvd->vdev_resilver_txg = txg;
 
 	/*
 	 * If the parent is not a mirror, or if we're replacing, insert the new
@@ -4674,7 +4676,6 @@ spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done)
 		if (pvd->vdev_ops == &vdev_spare_ops)
 			cvd->vdev_unspare = B_FALSE;
 		vdev_remove_parent(cvd);
-		cvd->vdev_resilvering = B_FALSE;
 	}
 
 
@@ -5391,6 +5392,8 @@ spa_vdev_resilver_done(spa_t *spa)
 			ASSERT(pvd->vdev_ops == &vdev_replacing_ops);
 			sguid = ppvd->vdev_child[1]->vdev_guid;
 		}
+		ASSERT(vd->vdev_resilver_txg == 0 || !vdev_dtl_required(vd));
+
 		spa_config_exit(spa, SCL_ALL, FTAG);
 		if (spa_vdev_detach(spa, guid, pguid, B_TRUE) != 0)
 			return;
diff --git a/usr/src/uts/common/fs/zfs/spa_config.c b/usr/src/uts/common/fs/zfs/spa_config.c
index d97fc32fbf..47bb595908 100644
--- a/usr/src/uts/common/fs/zfs/spa_config.c
+++ b/usr/src/uts/common/fs/zfs/spa_config.c
@@ -22,7 +22,7 @@
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
  * Copyright 2011 Nexenta Systems, Inc. All rights reserved.
- * Copyright (c) 2012 by Delphix. All rights reserved.
+ * Copyright (c) 2013 by Delphix. All rights reserved.
  */
 
 #include <sys/spa.h>
@@ -198,7 +198,12 @@ spa_config_write(spa_config_dirent_t *dp, nvlist_t *nvl)
 
 /*
  * Synchronize pool configuration to disk.  This must be called with the
- * namespace lock held.
+ * namespace lock held. Synchronizing the pool cache is typically done after
+ * the configuration has been synced to the MOS. This exposes a window where
+ * the MOS config will have been updated but the cache file has not. If
+ * the system were to crash at that instant then the cached config may not
+ * contain the correct information to open the pool and an explicity import
+ * would be required.
  */
 void
 spa_config_sync(spa_t *target, boolean_t removing, boolean_t postsysevent)
diff --git a/usr/src/uts/common/fs/zfs/sys/dsl_scan.h b/usr/src/uts/common/fs/zfs/sys/dsl_scan.h
index aae8c312af..bf8c5ac824 100644
--- a/usr/src/uts/common/fs/zfs/sys/dsl_scan.h
+++ b/usr/src/uts/common/fs/zfs/sys/dsl_scan.h
@@ -72,11 +72,42 @@ typedef enum dsl_scan_flags {
 	DSF_VISIT_DS_AGAIN = 1<<0,
 } dsl_scan_flags_t;
 
+/*
+ * Every pool will have one dsl_scan_t and this structure will contain
+ * in-memory information about the scan and a pointer to the on-disk
+ * representation (i.e. dsl_scan_phys_t). Most of the state of the scan
+ * is contained on-disk to allow the scan to resume in the event of a reboot
+ * or panic. This structure maintains information about the behavior of a
+ * running scan, some caching information, and how it should traverse the pool.
+ *
+ * The following members of this structure direct the behavior of the scan:
+ *
+ * scn_pausing -	a scan that cannot be completed in a single txg or
+ *			has exceeded its allotted time will need to pause.
+ *			When this flag is set the scanner will stop traversing
+ *			the pool and write out the current state to disk.
+ *
+ * scn_restart_txg -	directs the scanner to either restart or start a
+ *			a scan at the specified txg value.
+ *
+ * scn_done_txg -	when a scan completes its traversal it will set
+ *			the completion txg to the next txg. This is necessary
+ *			to ensure that any blocks that were freed during
+ *			the scan but have not yet been processed (i.e deferred
+ *			frees) are accounted for.
+ *
+ * This structure also maintains information about deferred frees which are
+ * a special kind of traversal. Deferred free can exist in either a bptree or
+ * a bpobj structure. The scn_is_bptree flag will indicate the type of
+ * deferred free that is in progress. If the deferred free is part of an
+ * asynchronous destroy then the scn_async_destroying flag will be set.
+ */
 typedef struct dsl_scan {
 	struct dsl_pool *scn_dp;
 
 	boolean_t scn_pausing;
 	uint64_t scn_restart_txg;
+	uint64_t scn_done_txg;
 	uint64_t scn_sync_start_time;
 	zio_t *scn_zio_root;
 
diff --git a/usr/src/uts/common/fs/zfs/sys/vdev_impl.h b/usr/src/uts/common/fs/zfs/sys/vdev_impl.h
index 02e3e838c3..9f68ac2968 100644
--- a/usr/src/uts/common/fs/zfs/sys/vdev_impl.h
+++ b/usr/src/uts/common/fs/zfs/sys/vdev_impl.h
@@ -20,7 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012 by Delphix. All rights reserved.
+ * Copyright (c) 2013 by Delphix. All rights reserved.
  */
 
 #ifndef _SYS_VDEV_IMPL_H
@@ -174,7 +174,7 @@ struct vdev {
 	uint64_t	vdev_faulted;	/* persistent faulted state	*/
 	uint64_t	vdev_degraded;	/* persistent degraded state	*/
 	uint64_t	vdev_removed;	/* persistent removed state	*/
-	uint64_t	vdev_resilvering; /* persistent resilvering state */
+	uint64_t	vdev_resilver_txg; /* persistent resilvering state */
 	uint64_t	vdev_nparity;	/* number of parity devices for raidz */
 	char		*vdev_path;	/* vdev path (if any)		*/
 	char		*vdev_devid;	/* vdev devid (if any)		*/
diff --git a/usr/src/uts/common/fs/zfs/sys/zfs_debug.h b/usr/src/uts/common/fs/zfs/sys/zfs_debug.h
index 14eb2abdc1..c4dcfaec65 100644
--- a/usr/src/uts/common/fs/zfs/sys/zfs_debug.h
+++ b/usr/src/uts/common/fs/zfs/sys/zfs_debug.h
@@ -20,7 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012 by Delphix. All rights reserved.
+ * Copyright (c) 2013 by Delphix. All rights reserved.
  */
 
 #ifndef _SYS_ZFS_DEBUG_H
@@ -77,6 +77,7 @@ typedef struct zfs_dbgmsg {
 extern void zfs_dbgmsg_init(void);
 extern void zfs_dbgmsg_fini(void);
 extern void zfs_dbgmsg(const char *fmt, ...);
+extern void zfs_dbgmsg_print(const char *tag);
 
 #ifndef _KERNEL
 extern int dprintf_find_string(const char *string);
diff --git a/usr/src/uts/common/fs/zfs/vdev.c b/usr/src/uts/common/fs/zfs/vdev.c
index 7a409bd7ed..dc5eb627f1 100644
--- a/usr/src/uts/common/fs/zfs/vdev.c
+++ b/usr/src/uts/common/fs/zfs/vdev.c
@@ -521,8 +521,8 @@ vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent, uint_t id,
 		(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_OFFLINE,
 		    &vd->vdev_offline);
 
-		(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_RESILVERING,
-		    &vd->vdev_resilvering);
+		(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_RESILVER_TXG,
+		    &vd->vdev_resilver_txg);
 
 		/*
 		 * When importing a pool, we want to ignore the persistent fault
@@ -1663,6 +1663,75 @@ vdev_dtl_empty(vdev_t *vd, vdev_dtl_type_t t)
 }
 
 /*
+ * Returns the lowest txg in the DTL range.
+ */
+static uint64_t
+vdev_dtl_min(vdev_t *vd)
+{
+	space_seg_t *ss;
+
+	ASSERT(MUTEX_HELD(&vd->vdev_dtl_lock));
+	ASSERT3U(vd->vdev_dtl[DTL_MISSING].sm_space, !=, 0);
+	ASSERT0(vd->vdev_children);
+
+	ss = avl_first(&vd->vdev_dtl[DTL_MISSING].sm_root);
+	return (ss->ss_start - 1);
+}
+
+/*
+ * Returns the highest txg in the DTL.
+ */
+static uint64_t
+vdev_dtl_max(vdev_t *vd)
+{
+	space_seg_t *ss;
+
+	ASSERT(MUTEX_HELD(&vd->vdev_dtl_lock));
+	ASSERT3U(vd->vdev_dtl[DTL_MISSING].sm_space, !=, 0);
+	ASSERT0(vd->vdev_children);
+
+	ss = avl_last(&vd->vdev_dtl[DTL_MISSING].sm_root);
+	return (ss->ss_end);
+}
+
+/*
+ * Determine if a resilvering vdev should remove any DTL entries from
+ * its range. If the vdev was resilvering for the entire duration of the
+ * scan then it should excise that range from its DTLs. Otherwise, this
+ * vdev is considered partially resilvered and should leave its DTL
+ * entries intact. The comment in vdev_dtl_reassess() describes how we
+ * excise the DTLs.
+ */
+static boolean_t
+vdev_dtl_should_excise(vdev_t *vd)
+{
+	spa_t *spa = vd->vdev_spa;
+	dsl_scan_t *scn = spa->spa_dsl_pool->dp_scan;
+
+	ASSERT0(scn->scn_phys.scn_errors);
+	ASSERT0(vd->vdev_children);
+
+	if (vd->vdev_resilver_txg == 0 ||
+	    vd->vdev_dtl[DTL_MISSING].sm_space == 0)
+		return (B_TRUE);
+
+	/*
+	 * When a resilver is initiated the scan will assign the scn_max_txg
+	 * value to the highest txg value that exists in all DTLs. If this
+	 * device's max DTL is not part of this scan (i.e. it is not in
+	 * the range (scn_min_txg, scn_max_txg] then it is not eligible
+	 * for excision.
+	 */
+	if (vdev_dtl_max(vd) <= scn->scn_phys.scn_max_txg) {
+		ASSERT3U(scn->scn_phys.scn_min_txg, <=, vdev_dtl_min(vd));
+		ASSERT3U(scn->scn_phys.scn_min_txg, <, vd->vdev_resilver_txg);
+		ASSERT3U(vd->vdev_resilver_txg, <=, scn->scn_phys.scn_max_txg);
+		return (B_TRUE);
+	}
+	return (B_FALSE);
+}
+
+/*
  * Reassess DTLs after a config change or scrub completion.
  */
 void
@@ -1685,9 +1754,17 @@ vdev_dtl_reassess(vdev_t *vd, uint64_t txg, uint64_t scrub_txg, int scrub_done)
 		dsl_scan_t *scn = spa->spa_dsl_pool->dp_scan;
 
 		mutex_enter(&vd->vdev_dtl_lock);
+
+		/*
+		 * If we've completed a scan cleanly then determine
+		 * if this vdev should remove any DTLs. We only want to
+		 * excise regions on vdevs that were available during
+		 * the entire duration of this scan.
+		 */
 		if (scrub_txg != 0 &&
 		    (spa->spa_scrub_started ||
-		    (scn && scn->scn_phys.scn_errors == 0))) {
+		    (scn != NULL && scn->scn_phys.scn_errors == 0)) &&
+		    vdev_dtl_should_excise(vd)) {
 			/*
 			 * We completed a scrub up to scrub_txg.  If we
 			 * did it without rebooting, then the scrub dtl
@@ -1726,6 +1803,16 @@ vdev_dtl_reassess(vdev_t *vd, uint64_t txg, uint64_t scrub_txg, int scrub_done)
 		else
 			space_map_walk(&vd->vdev_dtl[DTL_MISSING],
 			    space_map_add, &vd->vdev_dtl[DTL_OUTAGE]);
+
+		/*
+		 * If the vdev was resilvering and no longer has any
+		 * DTLs then reset its resilvering flag.
+		 */
+		if (vd->vdev_resilver_txg != 0 &&
+		    vd->vdev_dtl[DTL_MISSING].sm_space == 0 &&
+		    vd->vdev_dtl[DTL_OUTAGE].sm_space == 0)
+			vd->vdev_resilver_txg = 0;
+
 		mutex_exit(&vd->vdev_dtl_lock);
 
 		if (txg != 0)
@@ -1902,12 +1989,9 @@ vdev_resilver_needed(vdev_t *vd, uint64_t *minp, uint64_t *maxp)
 		mutex_enter(&vd->vdev_dtl_lock);
 		if (vd->vdev_dtl[DTL_MISSING].sm_space != 0 &&
 		    vdev_writeable(vd)) {
-			space_seg_t *ss;
 
-			ss = avl_first(&vd->vdev_dtl[DTL_MISSING].sm_root);
-			thismin = ss->ss_start - 1;
-			ss = avl_last(&vd->vdev_dtl[DTL_MISSING].sm_root);
-			thismax = ss->ss_end;
+			thismin = vdev_dtl_min(vd);
+			thismax = vdev_dtl_max(vd);
 			needed = B_TRUE;
 		}
 		mutex_exit(&vd->vdev_dtl_lock);
diff --git a/usr/src/uts/common/fs/zfs/vdev_file.c b/usr/src/uts/common/fs/zfs/vdev_file.c
index 2e67544345..a05abeb9d9 100644
--- a/usr/src/uts/common/fs/zfs/vdev_file.c
+++ b/usr/src/uts/common/fs/zfs/vdev_file.c
@@ -185,7 +185,6 @@ vdev_file_io_strategy(void *arg)
 static int
 vdev_file_io_start(zio_t *zio)
 {
-	spa_t *spa = zio->io_spa;
 	vdev_t *vd = zio->io_vd;
 	vdev_file_t *vf = vd->vdev_tsd;
 	vdev_buf_t *vb;
@@ -224,8 +223,8 @@ vdev_file_io_start(zio_t *zio)
 	bp->b_private = vf->vf_vnode;
 	bp->b_iodone = (int (*)())vdev_file_io_intr;
 
-	spa_taskq_dispatch_ent(spa, ZIO_TYPE_FREE, ZIO_TASKQ_ISSUE,
-	    vdev_file_io_strategy, bp, 0, &zio->io_tqent);
+	VERIFY3U(taskq_dispatch(system_taskq, vdev_file_io_strategy, bp,
+	    TQ_SLEEP), !=, 0);
 
 	return (ZIO_PIPELINE_STOP);
 }
diff --git a/usr/src/uts/common/fs/zfs/vdev_label.c b/usr/src/uts/common/fs/zfs/vdev_label.c
index 904918c3a4..ee9921f94a 100644
--- a/usr/src/uts/common/fs/zfs/vdev_label.c
+++ b/usr/src/uts/common/fs/zfs/vdev_label.c
@@ -216,30 +216,25 @@ vdev_config_generate(spa_t *spa, vdev_t *vd, boolean_t getstats,
 {
 	nvlist_t *nv = NULL;
 
-	VERIFY(nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_SLEEP) == 0);
+	nv = fnvlist_alloc();
 
-	VERIFY(nvlist_add_string(nv, ZPOOL_CONFIG_TYPE,
-	    vd->vdev_ops->vdev_op_type) == 0);
+	fnvlist_add_string(nv, ZPOOL_CONFIG_TYPE, vd->vdev_ops->vdev_op_type);
 	if (!(flags & (VDEV_CONFIG_SPARE | VDEV_CONFIG_L2CACHE)))
-		VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_ID, vd->vdev_id)
-		    == 0);
-	VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_GUID, vd->vdev_guid) == 0);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_ID, vd->vdev_id);
+	fnvlist_add_uint64(nv, ZPOOL_CONFIG_GUID, vd->vdev_guid);
 
 	if (vd->vdev_path != NULL)
-		VERIFY(nvlist_add_string(nv, ZPOOL_CONFIG_PATH,
-		    vd->vdev_path) == 0);
+		fnvlist_add_string(nv, ZPOOL_CONFIG_PATH, vd->vdev_path);
 
 	if (vd->vdev_devid != NULL)
-		VERIFY(nvlist_add_string(nv, ZPOOL_CONFIG_DEVID,
-		    vd->vdev_devid) == 0);
+		fnvlist_add_string(nv, ZPOOL_CONFIG_DEVID, vd->vdev_devid);
 
 	if (vd->vdev_physpath != NULL)
-		VERIFY(nvlist_add_string(nv, ZPOOL_CONFIG_PHYS_PATH,
-		    vd->vdev_physpath) == 0);
+		fnvlist_add_string(nv, ZPOOL_CONFIG_PHYS_PATH,
+		    vd->vdev_physpath);
 
 	if (vd->vdev_fru != NULL)
-		VERIFY(nvlist_add_string(nv, ZPOOL_CONFIG_FRU,
-		    vd->vdev_fru) == 0);
+		fnvlist_add_string(nv, ZPOOL_CONFIG_FRU, vd->vdev_fru);
 
 	if (vd->vdev_nparity != 0) {
 		ASSERT(strcmp(vd->vdev_ops->vdev_op_type,
@@ -260,59 +255,54 @@ vdev_config_generate(spa_t *spa, vdev_t *vd, boolean_t getstats,
 		 * that only support a single parity device -- older software
 		 * will just ignore it.
 		 */
-		VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_NPARITY,
-		    vd->vdev_nparity) == 0);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_NPARITY, vd->vdev_nparity);
 	}
 
 	if (vd->vdev_wholedisk != -1ULL)
-		VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK,
-		    vd->vdev_wholedisk) == 0);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK,
+		    vd->vdev_wholedisk);
 
 	if (vd->vdev_not_present)
-		VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT, 1) == 0);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT, 1);
 
 	if (vd->vdev_isspare)
-		VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_IS_SPARE, 1) == 0);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_IS_SPARE, 1);
 
 	if (!(flags & (VDEV_CONFIG_SPARE | VDEV_CONFIG_L2CACHE)) &&
 	    vd == vd->vdev_top) {
-		VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_METASLAB_ARRAY,
-		    vd->vdev_ms_array) == 0);
-		VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_METASLAB_SHIFT,
-		    vd->vdev_ms_shift) == 0);
-		VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_ASHIFT,
-		    vd->vdev_ashift) == 0);
-		VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_ASIZE,
-		    vd->vdev_asize) == 0);
-		VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_IS_LOG,
-		    vd->vdev_islog) == 0);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_METASLAB_ARRAY,
+		    vd->vdev_ms_array);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_METASLAB_SHIFT,
+		    vd->vdev_ms_shift);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_ASHIFT, vd->vdev_ashift);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_ASIZE,
+		    vd->vdev_asize);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_IS_LOG, vd->vdev_islog);
 		if (vd->vdev_removing)
-			VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_REMOVING,
-			    vd->vdev_removing) == 0);
+			fnvlist_add_uint64(nv, ZPOOL_CONFIG_REMOVING,
+			    vd->vdev_removing);
 	}
 
 	if (vd->vdev_dtl_smo.smo_object != 0)
-		VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_DTL,
-		    vd->vdev_dtl_smo.smo_object) == 0);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_DTL,
+		    vd->vdev_dtl_smo.smo_object);
 
 	if (vd->vdev_crtxg)
-		VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_CREATE_TXG,
-		    vd->vdev_crtxg) == 0);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_CREATE_TXG, vd->vdev_crtxg);
 
 	if (getstats) {
 		vdev_stat_t vs;
 		pool_scan_stat_t ps;
 
 		vdev_get_stats(vd, &vs);
-		VERIFY(nvlist_add_uint64_array(nv, ZPOOL_CONFIG_VDEV_STATS,
-		    (uint64_t *)&vs, sizeof (vs) / sizeof (uint64_t)) == 0);
+		fnvlist_add_uint64_array(nv, ZPOOL_CONFIG_VDEV_STATS,
+		    (uint64_t *)&vs, sizeof (vs) / sizeof (uint64_t));
 
 		/* provide either current or previous scan information */
 		if (spa_scan_get_stats(spa, &ps) == 0) {
-			VERIFY(nvlist_add_uint64_array(nv,
+			fnvlist_add_uint64_array(nv,
 			    ZPOOL_CONFIG_SCAN_STATS, (uint64_t *)&ps,
-			    sizeof (pool_scan_stat_t) / sizeof (uint64_t))
-			    == 0);
+			    sizeof (pool_scan_stat_t) / sizeof (uint64_t));
 		}
 	}
 
@@ -342,8 +332,8 @@ vdev_config_generate(spa_t *spa, vdev_t *vd, boolean_t getstats,
 		}
 
 		if (idx) {
-			VERIFY(nvlist_add_nvlist_array(nv,
-			    ZPOOL_CONFIG_CHILDREN, child, idx) == 0);
+			fnvlist_add_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
+			    child, idx);
 		}
 
 		for (c = 0; c < idx; c++)
@@ -355,26 +345,20 @@ vdev_config_generate(spa_t *spa, vdev_t *vd, boolean_t getstats,
 		const char *aux = NULL;
 
 		if (vd->vdev_offline && !vd->vdev_tmpoffline)
-			VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_OFFLINE,
-			    B_TRUE) == 0);
-		if (vd->vdev_resilvering)
-			VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_RESILVERING,
-			    B_TRUE) == 0);
+			fnvlist_add_uint64(nv, ZPOOL_CONFIG_OFFLINE, B_TRUE);
+		if (vd->vdev_resilver_txg != 0)
+			fnvlist_add_uint64(nv, ZPOOL_CONFIG_RESILVER_TXG,
+			    vd->vdev_resilver_txg);
 		if (vd->vdev_faulted)
-			VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_FAULTED,
-			    B_TRUE) == 0);
+			fnvlist_add_uint64(nv, ZPOOL_CONFIG_FAULTED, B_TRUE);
 		if (vd->vdev_degraded)
-			VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_DEGRADED,
-			    B_TRUE) == 0);
+			fnvlist_add_uint64(nv, ZPOOL_CONFIG_DEGRADED, B_TRUE);
 		if (vd->vdev_removed)
-			VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_REMOVED,
-			    B_TRUE) == 0);
+			fnvlist_add_uint64(nv, ZPOOL_CONFIG_REMOVED, B_TRUE);
 		if (vd->vdev_unspare)
-			VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_UNSPARE,
-			    B_TRUE) == 0);
+			fnvlist_add_uint64(nv, ZPOOL_CONFIG_UNSPARE, B_TRUE);
 		if (vd->vdev_ishole)
-			VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_IS_HOLE,
-			    B_TRUE) == 0);
+			fnvlist_add_uint64(nv, ZPOOL_CONFIG_IS_HOLE, B_TRUE);
 
 		switch (vd->vdev_stat.vs_aux) {
 		case VDEV_AUX_ERR_EXCEEDED:
@@ -387,12 +371,11 @@ vdev_config_generate(spa_t *spa, vdev_t *vd, boolean_t getstats,
 		}
 
 		if (aux != NULL)
-			VERIFY(nvlist_add_string(nv, ZPOOL_CONFIG_AUX_STATE,
-			    aux) == 0);
+			fnvlist_add_string(nv, ZPOOL_CONFIG_AUX_STATE, aux);
 
 		if (vd->vdev_splitting && vd->vdev_orig_guid != 0LL) {
-			VERIFY(nvlist_add_uint64(nv, ZPOOL_CONFIG_ORIG_GUID,
-			    vd->vdev_orig_guid) == 0);
+			fnvlist_add_uint64(nv, ZPOOL_CONFIG_ORIG_GUID,
+			    vd->vdev_orig_guid);
 		}
 	}
 
diff --git a/usr/src/uts/common/fs/zfs/zfs_debug.c b/usr/src/uts/common/fs/zfs/zfs_debug.c
index 44824e15a0..26ea561eb1 100644
--- a/usr/src/uts/common/fs/zfs/zfs_debug.c
+++ b/usr/src/uts/common/fs/zfs/zfs_debug.c
@@ -20,7 +20,7 @@
  */
 /*
  * Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012 by Delphix. All rights reserved.
+ * Copyright (c) 2013 by Delphix. All rights reserved.
  */
 
 #include <sys/zfs_context.h>
@@ -94,3 +94,16 @@ zfs_dbgmsg(const char *fmt, ...)
 	}
 	mutex_exit(&zfs_dbgmsgs_lock);
 }
+
+void
+zfs_dbgmsg_print(const char *tag)
+{
+	zfs_dbgmsg_t *zdm;
+
+	(void) printf("ZFS_DBGMSG(%s):\n", tag);
+	mutex_enter(&zfs_dbgmsgs_lock);
+	for (zdm = list_head(&zfs_dbgmsgs); zdm;
+	    zdm = list_next(&zfs_dbgmsgs, zdm))
+		(void) printf("%s\n", zdm->zdm_msg);
+	mutex_exit(&zfs_dbgmsgs_lock);
+}
diff --git a/usr/src/uts/common/sys/fs/zfs.h b/usr/src/uts/common/sys/fs/zfs.h
index 6be8aee742..f6eb9aa490 100644
--- a/usr/src/uts/common/sys/fs/zfs.h
+++ b/usr/src/uts/common/sys/fs/zfs.h
@@ -21,7 +21,7 @@
 
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012 by Delphix. All rights reserved.
+ * Copyright (c) 2013 by Delphix. All rights reserved.
  * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
  * Copyright (c) 2012, Joyent, Inc. All rights reserved.
  */
@@ -522,7 +522,7 @@ typedef struct zpool_rewind_policy {
 #define	ZPOOL_CONFIG_SPLIT_GUID		"split_guid"
 #define	ZPOOL_CONFIG_SPLIT_LIST		"guid_list"
 #define	ZPOOL_CONFIG_REMOVING		"removing"
-#define	ZPOOL_CONFIG_RESILVERING	"resilvering"
+#define	ZPOOL_CONFIG_RESILVER_TXG	"resilver_txg"
 #define	ZPOOL_CONFIG_COMMENT		"comment"
 #define	ZPOOL_CONFIG_SUSPENDED		"suspended"	/* not stored on disk */
 #define	ZPOOL_CONFIG_TIMESTAMP		"timestamp"	/* not stored on disk */