Illumos #2882, #2883, #2900

2882 implement libzfs_core
2883 changing "canmount" property to "on" should not always remount dataset
2900 "zfs snapshot" should be able to create multiple, arbitrary snapshots at once

Reviewed by: George Wilson <[email protected]>
Reviewed by: Chris Siden <[email protected]>
Reviewed by: Garrett D'Amore <[email protected]>
Reviewed by: Bill Pijewski <[email protected]>
Reviewed by: Dan Kruchinin <[email protected]>
Approved by: Eric Schrock <[email protected]>

References:
  https://www.illumos.org/issues/2882
  https://www.illumos.org/issues/2883
  https://www.illumos.org/issues/2900
  illumos/illumos-gate@4445fffbbb1ea25fd0e9ea68b9380dd7a6709025

Ported-by: Tim Chase <[email protected]>
Signed-off-by: Brian Behlendorf <[email protected]>
Closes #1293

Porting notes:

WARNING: This patch changes the user/kernel ABI.  That means that
the zfs/zpool utilities built from master are NOT compatible with
the 0.6.2 kernel modules.  Ensure you load the matching kernel
modules from master after updating the utilities.  Otherwise the
zfs/zpool commands will be unable to interact with your pool and
you will see errors similar to the following:

  $ zpool list
  failed to read pool configuration: bad address
  no pools available

  $ zfs list
  no datasets available

Add zvol minor device creation to the new zfs_snapshot_nvl function.

Remove the logging of the "release" operation in
dsl_dataset_user_release_sync().  The logging caused a null dereference
because ds->ds_dir is zeroed in dsl_dataset_destroy_sync() and the
logging functions try to get the ds name via the dsl_dataset_name()
function. I've got no idea why this particular code would have worked
in Illumos.  This code has subsequently been completely reworked in
Illumos commit 3b2aab1 (3464 zfs synctask code needs restructuring).

Squash some "may be used uninitialized" warning/erorrs.

Fix some printf format warnings for %lld and %llu.

Apply a few spa_writeable() changes that were made to Illumos in
illumos/illumos-gate.git@cd1c8b8 as part of the 3112, 3113, 3114 and
3115 fixes.

Add a missing call to fnvlist_free(nvl) in log_internal() that was added
in Illumos to fix issue 3085 but couldn't be ported to ZoL at the time
(zfsonlinux/zfs@9e11c73) because it depended on future work.

1 files changed, 68 insertions, 77 deletions

diff --git a/module/zfs/dsl_dataset.c b/module/zfs/dsl_dataset.c
index d9e8bd30f..2eca2b204 100644
--- a/module/zfs/dsl_dataset.c
+++ b/module/zfs/dsl_dataset.c
@@ -907,7 +907,8 @@ dsl_dataset_create_sync(dsl_dir_t *pdd, const char *lastname,
  * The snapshots must all be in the same pool.
  */
 int
-dmu_snapshots_destroy_nvl(nvlist_t *snaps, boolean_t defer, char *failed)
+dmu_snapshots_destroy_nvl(nvlist_t *snaps, boolean_t defer,
+    nvlist_t *errlist)
 {
 	int err;
 	dsl_sync_task_t *dst;
@@ -942,7 +943,7 @@ dmu_snapshots_destroy_nvl(nvlist_t *snaps, boolean_t defer, char *failed)
 		} else if (err == ENOENT) {
 			err = 0;
 		} else {
-			(void) strcpy(failed, nvpair_name(pair));
+			fnvlist_add_int32(errlist, nvpair_name(pair), err);
 			break;
 		}
 	}
@@ -956,10 +957,12 @@ dmu_snapshots_destroy_nvl(nvlist_t *snaps, boolean_t defer, char *failed)
 		dsl_dataset_t *ds = dsda->ds;
 
 		/*
-		 * Return the file system name that triggered the error
+		 * Return the snapshots that triggered the error.
 		 */
-		if (dst->dst_err) {
-			dsl_dataset_name(ds, failed);
+		if (dst->dst_err != 0) {
+			char name[ZFS_MAXNAMELEN];
+			dsl_dataset_name(ds, name);
+			fnvlist_add_int32(errlist, name, dst->dst_err);
 		}
 		ASSERT3P(dsda->rm_origin, ==, NULL);
 		dsl_dataset_disown(ds, dstg);
@@ -1038,7 +1041,6 @@ dsl_dataset_destroy(dsl_dataset_t *ds, void *tag, boolean_t defer)
 	dsl_dir_t *dd;
 	uint64_t obj;
 	struct dsl_ds_destroyarg dsda = { 0 };
-	dsl_dataset_t *dummy_ds;
 
 	dsda.ds = ds;
 
@@ -1058,9 +1060,6 @@ dsl_dataset_destroy(dsl_dataset_t *ds, void *tag, boolean_t defer)
 	}
 
 	dd = ds->ds_dir;
-	dummy_ds = kmem_zalloc(sizeof (dsl_dataset_t), KM_SLEEP);
-	dummy_ds->ds_dir = dd;
-	dummy_ds->ds_object = ds->ds_object;
 
 	if (!spa_feature_is_enabled(dsl_dataset_get_spa(ds),
 	    &spa_feature_table[SPA_FEATURE_ASYNC_DESTROY])) {
@@ -1072,11 +1071,11 @@ dsl_dataset_destroy(dsl_dataset_t *ds, void *tag, boolean_t defer)
 		    dsl_dataset_destroy_begin_check,
 		    dsl_dataset_destroy_begin_sync, ds, NULL, 0);
 		if (err)
-			goto out_free;
+			goto out;
 
 		err = dmu_objset_from_ds(ds, &os);
 		if (err)
-			goto out_free;
+			goto out;
 
 		/*
 		 * Remove all objects while in the open context so that
@@ -1091,7 +1090,7 @@ dsl_dataset_destroy(dsl_dataset_t *ds, void *tag, boolean_t defer)
 			(void) dmu_free_object(os, obj);
 		}
 		if (err != ESRCH)
-			goto out_free;
+			goto out;
 
 		/*
 		 * Sync out all in-flight IO.
@@ -1118,7 +1117,7 @@ dsl_dataset_destroy(dsl_dataset_t *ds, void *tag, boolean_t defer)
 	rw_exit(&dd->dd_pool->dp_config_rwlock);
 
 	if (err)
-		goto out_free;
+		goto out;
 
 	/*
 	 * Blow away the dsl_dir + head dataset.
@@ -1134,7 +1133,7 @@ dsl_dataset_destroy(dsl_dataset_t *ds, void *tag, boolean_t defer)
 			err = dsl_dataset_origin_rm_prep(&dsda, tag);
 			if (err) {
 				dsl_dir_close(dd, FTAG);
-				goto out_free;
+				goto out;
 			}
 		}
 
@@ -1142,7 +1141,7 @@ dsl_dataset_destroy(dsl_dataset_t *ds, void *tag, boolean_t defer)
 		dsl_sync_task_create(dstg, dsl_dataset_destroy_check,
 		    dsl_dataset_destroy_sync, &dsda, tag, 0);
 		dsl_sync_task_create(dstg, dsl_dir_destroy_check,
-		    dsl_dir_destroy_sync, dummy_ds, FTAG, 0);
+		    dsl_dir_destroy_sync, dd, FTAG, 0);
 		err = dsl_sync_task_group_wait(dstg);
 		dsl_sync_task_group_destroy(dstg);
 
@@ -1166,8 +1165,6 @@ dsl_dataset_destroy(dsl_dataset_t *ds, void *tag, boolean_t defer)
 	if (err)
 		dsl_dir_close(dd, FTAG);
 
-out_free:
-	kmem_free(dummy_ds, sizeof (dsl_dataset_t));
 out:
 	dsl_dataset_disown(ds, tag);
 	return (err);
@@ -1333,14 +1330,12 @@ static void
 dsl_dataset_destroy_begin_sync(void *arg1, void *arg2, dmu_tx_t *tx)
 {
 	dsl_dataset_t *ds = arg1;
-	dsl_pool_t *dp = ds->ds_dir->dd_pool;
 
 	/* Mark it as inconsistent on-disk, in case we crash */
 	dmu_buf_will_dirty(ds->ds_dbuf, tx);
 	ds->ds_phys->ds_flags |= DS_FLAG_INCONSISTENT;
 
-	spa_history_log_internal(LOG_DS_DESTROY_BEGIN, dp->dp_spa, tx,
-	    "dataset = %llu", ds->ds_object);
+	spa_history_log_internal_ds(ds, "destroy begin", tx, "");
 }
 
 static int
@@ -1665,9 +1660,13 @@ dsl_dataset_destroy_sync(void *arg1, void *tag, dmu_tx_t *tx)
 		ASSERT(spa_version(dp->dp_spa) >= SPA_VERSION_USERREFS);
 		dmu_buf_will_dirty(ds->ds_dbuf, tx);
 		ds->ds_phys->ds_flags |= DS_FLAG_DEFER_DESTROY;
+		spa_history_log_internal_ds(ds, "defer_destroy", tx, "");
 		return;
 	}
 
+	/* We need to log before removing it from the namespace. */
+	spa_history_log_internal_ds(ds, "destroy", tx, "");
+
 	/* signal any waiters that this dataset is going away */
 	mutex_enter(&ds->ds_lock);
 	ds->ds_owner = dsl_reaper;
@@ -1965,8 +1964,6 @@ dsl_dataset_destroy_sync(void *arg1, void *tag, dmu_tx_t *tx)
 		dsl_dataset_rele(ds_prev, FTAG);
 
 	spa_prop_clear_bootfs(dp->dp_spa, ds->ds_object, tx);
-	spa_history_log_internal(LOG_DS_DESTROY, dp->dp_spa, tx,
-	    "dataset = %llu", ds->ds_object);
 
 	if (ds->ds_phys->ds_next_clones_obj != 0) {
 		ASSERTV(uint64_t count);
@@ -2014,7 +2011,7 @@ dsl_dataset_snapshot_reserve_space(dsl_dataset_t *ds, dmu_tx_t *tx)
 		return (ENOSPC);
 
 	/*
-	 * Propogate any reserved space for this snapshot to other
+	 * Propagate any reserved space for this snapshot to other
 	 * snapshot checks in this sync group.
 	 */
 	if (asize > 0)
@@ -2024,10 +2021,9 @@ dsl_dataset_snapshot_reserve_space(dsl_dataset_t *ds, dmu_tx_t *tx)
 }
 
 int
-dsl_dataset_snapshot_check(void *arg1, void *arg2, dmu_tx_t *tx)
+dsl_dataset_snapshot_check(dsl_dataset_t *ds, const char *snapname,
+    dmu_tx_t *tx)
 {
-	dsl_dataset_t *ds = arg1;
-	const char *snapname = arg2;
 	int err;
 	uint64_t value;
 
@@ -2039,7 +2035,7 @@ dsl_dataset_snapshot_check(void *arg1, void *arg2, dmu_tx_t *tx)
 		return (EAGAIN);
 
 	/*
-	 * Check for conflicting name snapshot name.
+	 * Check for conflicting snapshot name.
 	 */
 	err = dsl_dataset_snap_lookup(ds, snapname, &value);
 	if (err == 0)
@@ -2063,10 +2059,9 @@ dsl_dataset_snapshot_check(void *arg1, void *arg2, dmu_tx_t *tx)
 }
 
 void
-dsl_dataset_snapshot_sync(void *arg1, void *arg2, dmu_tx_t *tx)
+dsl_dataset_snapshot_sync(dsl_dataset_t *ds, const char *snapname,
+    dmu_tx_t *tx)
 {
-	dsl_dataset_t *ds = arg1;
-	const char *snapname = arg2;
 	dsl_pool_t *dp = ds->ds_dir->dd_pool;
 	dmu_buf_t *dbuf;
 	dsl_dataset_phys_t *dsphys;
@@ -2172,8 +2167,7 @@ dsl_dataset_snapshot_sync(void *arg1, void *arg2, dmu_tx_t *tx)
 
 	dsl_dir_snap_cmtime_update(ds->ds_dir);
 
-	spa_history_log_internal(LOG_DS_SNAPSHOT, dp->dp_spa, tx,
-	    "dataset = %llu", dsobj);
+	spa_history_log_internal_ds(ds->ds_prev, "snapshot", tx, "");
 }
 
 void
@@ -2259,7 +2253,20 @@ dsl_dataset_stats(dsl_dataset_t *ds, nvlist_t *nv)
 {
 	uint64_t refd, avail, uobjs, aobjs, ratio;
 
-	dsl_dir_stats(ds->ds_dir, nv);
+	ratio = ds->ds_phys->ds_compressed_bytes == 0 ? 100 :
+	    (ds->ds_phys->ds_uncompressed_bytes * 100 /
+	    ds->ds_phys->ds_compressed_bytes);
+
+	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_REFRATIO, ratio);
+
+	if (dsl_dataset_is_snapshot(ds)) {
+		dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_COMPRESSRATIO, ratio);
+		dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_USED,
+		    ds->ds_phys->ds_unique_bytes);
+		get_clones_stat(ds, nv);
+	} else {
+		dsl_dir_stats(ds->ds_dir, nv);
+	}
 
 	dsl_dataset_space(ds, &refd, &avail, &uobjs, &aobjs);
 	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_AVAILABLE, avail);
@@ -2305,22 +2312,6 @@ dsl_dataset_stats(dsl_dataset_t *ds, nvlist_t *nv)
 		}
 	}
 
-	ratio = ds->ds_phys->ds_compressed_bytes == 0 ? 100 :
-	    (ds->ds_phys->ds_uncompressed_bytes * 100 /
-	    ds->ds_phys->ds_compressed_bytes);
-	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_REFRATIO, ratio);
-
-	if (ds->ds_phys->ds_next_snap_obj) {
-		/*
-		 * This is a snapshot; override the dd's space used with
-		 * our unique space and compression ratio.
-		 */
-		dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_USED,
-		    ds->ds_phys->ds_unique_bytes);
-		dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_COMPRESSRATIO, ratio);
-
-		get_clones_stat(ds, nv);
-	}
 }
 
 void
@@ -2329,27 +2320,25 @@ dsl_dataset_fast_stat(dsl_dataset_t *ds, dmu_objset_stats_t *stat)
 	stat->dds_creation_txg = ds->ds_phys->ds_creation_txg;
 	stat->dds_inconsistent = ds->ds_phys->ds_flags & DS_FLAG_INCONSISTENT;
 	stat->dds_guid = ds->ds_phys->ds_guid;
-	if (ds->ds_phys->ds_next_snap_obj) {
+	stat->dds_origin[0] = '\0';
+	if (dsl_dataset_is_snapshot(ds)) {
 		stat->dds_is_snapshot = B_TRUE;
 		stat->dds_num_clones = ds->ds_phys->ds_num_children - 1;
 	} else {
 		stat->dds_is_snapshot = B_FALSE;
 		stat->dds_num_clones = 0;
-	}
 
-	/* clone origin is really a dsl_dir thing... */
-	rw_enter(&ds->ds_dir->dd_pool->dp_config_rwlock, RW_READER);
-	if (dsl_dir_is_clone(ds->ds_dir)) {
-		dsl_dataset_t *ods;
+		rw_enter(&ds->ds_dir->dd_pool->dp_config_rwlock, RW_READER);
+		if (dsl_dir_is_clone(ds->ds_dir)) {
+			dsl_dataset_t *ods;
 
-		VERIFY(0 == dsl_dataset_get_ref(ds->ds_dir->dd_pool,
-		    ds->ds_dir->dd_phys->dd_origin_obj, FTAG, &ods));
-		dsl_dataset_name(ods, stat->dds_origin);
-		dsl_dataset_drop_ref(ods, FTAG);
-	} else {
-		stat->dds_origin[0] = '\0';
+			VERIFY(0 == dsl_dataset_get_ref(ds->ds_dir->dd_pool,
+			    ds->ds_dir->dd_phys->dd_origin_obj, FTAG, &ods));
+			dsl_dataset_name(ods, stat->dds_origin);
+			dsl_dataset_drop_ref(ods, FTAG);
+		}
+		rw_exit(&ds->ds_dir->dd_pool->dp_config_rwlock);
 	}
-	rw_exit(&ds->ds_dir->dd_pool->dp_config_rwlock);
 }
 
 uint64_t
@@ -2466,8 +2455,8 @@ dsl_dataset_snapshot_rename_sync(void *arg1, void *arg2, dmu_tx_t *tx)
 	    ds->ds_snapname, 8, 1, &ds->ds_object, tx);
 	ASSERT0(err);
 
-	spa_history_log_internal(LOG_DS_RENAME, dd->dd_pool->dp_spa, tx,
-	    "dataset = %llu", ds->ds_object);
+	spa_history_log_internal_ds(ds, "rename", tx,
+	    "-> @%s", newsnapname);
 	dsl_dataset_rele(hds, FTAG);
 }
 
@@ -2946,8 +2935,7 @@ dsl_dataset_promote_sync(void *arg1, void *arg2, dmu_tx_t *tx)
 	origin_ds->ds_phys->ds_unique_bytes = pa->unique;
 
 	/* log history record */
-	spa_history_log_internal(LOG_DS_PROMOTE, dd->dd_pool->dp_spa, tx,
-	    "dataset = %llu", hds->ds_object);
+	spa_history_log_internal_ds(hds, "promote", tx, "");
 
 	dsl_dir_close(odd, FTAG);
 }
@@ -3306,6 +3294,9 @@ dsl_dataset_clone_swap_sync(void *arg1, void *arg2, dmu_tx_t *tx)
 	    csa->ohds->ds_phys->ds_deadlist_obj);
 
 	dsl_scan_ds_clone_swapped(csa->ohds, csa->cds, tx);
+
+	spa_history_log_internal_ds(csa->cds, "clone swap", tx,
+	    "parent=%s", csa->ohds->ds_dir->dd_myname);
 }
 
 /*
@@ -3463,6 +3454,9 @@ dsl_dataset_set_quota_sync(void *arg1, void *arg2, dmu_tx_t *tx)
 	if (ds->ds_quota != effective_value) {
 		dmu_buf_will_dirty(ds->ds_dbuf, tx);
 		ds->ds_quota = effective_value;
+
+		spa_history_log_internal_ds(ds, "set refquota", tx,
+		    "refquota=%lld", (longlong_t)ds->ds_quota);
 	}
 }
 
@@ -3566,6 +3560,9 @@ dsl_dataset_set_reservation_sync(void *arg1, void *arg2, dmu_tx_t *tx)
 
 	dsl_dir_diduse_space(ds->ds_dir, DD_USED_REFRSRV, delta, 0, 0, tx);
 	mutex_exit(&ds->ds_dir->dd_lock);
+
+	spa_history_log_internal_ds(ds, "set refreservation", tx,
+	    "refreservation=%lld", (longlong_t)effective_value);
 }
 
 int
@@ -3631,7 +3628,7 @@ dsl_dataset_user_hold_check(void *arg1, void *arg2, dmu_tx_t *tx)
 {
 	dsl_dataset_t *ds = arg1;
 	struct dsl_ds_holdarg *ha = arg2;
-	char *htag = ha->htag;
+	const char *htag = ha->htag;
 	objset_t *mos = ds->ds_dir->dd_pool->dp_meta_objset;
 	int error = 0;
 
@@ -3665,7 +3662,7 @@ dsl_dataset_user_hold_sync(void *arg1, void *arg2, dmu_tx_t *tx)
 {
 	dsl_dataset_t *ds = arg1;
 	struct dsl_ds_holdarg *ha = arg2;
-	char *htag = ha->htag;
+	const char *htag = ha->htag;
 	dsl_pool_t *dp = ds->ds_dir->dd_pool;
 	objset_t *mos = dp->dp_meta_objset;
 	uint64_t now = gethrestime_sec();
@@ -3693,9 +3690,9 @@ dsl_dataset_user_hold_sync(void *arg1, void *arg2, dmu_tx_t *tx)
 		    htag, &now, tx));
 	}
 
-	spa_history_log_internal(LOG_DS_USER_HOLD,
-	    dp->dp_spa, tx, "<%s> temp = %d dataset = %llu", htag,
-	    (int)ha->temphold, ds->ds_object);
+	spa_history_log_internal_ds(ds, "hold", tx,
+	    "tag = %s temp = %d holds now = %llu",
+	    htag, (int)ha->temphold, ds->ds_userrefs);
 }
 
 static int
@@ -3902,7 +3899,6 @@ dsl_dataset_user_release_sync(void *arg1, void *tag, dmu_tx_t *tx)
 	dsl_pool_t *dp = ds->ds_dir->dd_pool;
 	objset_t *mos = dp->dp_meta_objset;
 	uint64_t zapobj;
-	uint64_t dsobj = ds->ds_object;
 	uint64_t refs;
 	int error;
 
@@ -3914,11 +3910,6 @@ dsl_dataset_user_release_sync(void *arg1, void *tag, dmu_tx_t *tx)
 	VERIFY(error == 0 || error == ENOENT);
 	zapobj = ds->ds_phys->ds_userrefs_obj;
 	VERIFY(0 == zap_remove(mos, zapobj, ra->htag, tx));
-
-	spa_history_log_internal(LOG_DS_USER_RELEASE,
-	    dp->dp_spa, tx, "<%s> %lld dataset = %llu",
-	    ra->htag, (longlong_t)refs, dsobj);
-
 	if (ds->ds_userrefs == 0 && ds->ds_phys->ds_num_children == 1 &&
 	    DS_IS_DEFER_DESTROY(ds)) {
 		struct dsl_ds_destroyarg dsda = {0};