Create zap for root vdev

And add it to the AVZ, this is not backwards compatible with older pools
due to an assertion in spa_sync() that verifies the number of ZAPs of
all vdevs matches the number of ZAPs in the AVZ.

Granted, the assertion only applies to #DEBUG builds - still, a feature
flag is introduced to avoid the assertion, com.klarasystems:vdev_zaps_v2

Notably, this allows to get/set properties on the root vdev:

    % zpool set user:prop=value <pool> root-0

Before this commit, it was already possible to get/set properties on
top-level vdevs with the syntax <type>-<vdev_id> (e.g. mirror-0):

    % zpool set user:prop=value <pool> mirror-0

This syntax also applies to the root vdev as it is is of type 'root'
with a vdev_id of 0, root-0. The keyword 'root' as an alias for
'root-0'.

The following tests have been added:

    - zpool get all properties from root vdev
    - zpool set a property on root vdev
    - verify root vdev ZAP is created

Reviewed-by: Brian Behlendorf <[email protected]>
Signed-off-by: Rob Wing <[email protected]>
Sponsored-by: Seagate Technology
Submitted-by: Klara, Inc.
Closes #14405

4 files changed, 50 insertions, 4 deletions

diff --git a/module/zcommon/zfeature_common.c b/module/zcommon/zfeature_common.c
index 6fe1da8ed..4c9b7ed72 100644
--- a/module/zcommon/zfeature_common.c
+++ b/module/zcommon/zfeature_common.c
@@ -731,6 +731,12 @@ zpool_feature_init(void)
 	    ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL,
 	    sfeatures);
 
+	zfeature_register(SPA_FEATURE_AVZ_V2,
+	    "com.klarasystems:vdev_zaps_v2", "vdev_zaps_v2",
+	    "Support for root vdev ZAP.",
+	    ZFEATURE_FLAG_MOS, ZFEATURE_TYPE_BOOLEAN, NULL,
+	    sfeatures);
+
 	zfs_mod_list_supported_free(sfeatures);
 }
 
diff --git a/module/zfs/spa.c b/module/zfs/spa.c
index e76700a9c..67601211d 100644
--- a/module/zfs/spa.c
+++ b/module/zfs/spa.c
@@ -3044,6 +3044,12 @@ vdev_count_verify_zaps(vdev_t *vd)
 	spa_t *spa = vd->vdev_spa;
 	uint64_t total = 0;
 
+	if (spa_feature_is_active(vd->vdev_spa, SPA_FEATURE_AVZ_V2) &&
+	    vd->vdev_root_zap != 0) {
+		total++;
+		ASSERT0(zap_lookup_int(spa->spa_meta_objset,
+		    spa->spa_all_vdev_zaps, vd->vdev_root_zap));
+	}
 	if (vd->vdev_top_zap != 0) {
 		total++;
 		ASSERT0(zap_lookup_int(spa->spa_meta_objset,
@@ -8626,6 +8632,11 @@ spa_avz_build(vdev_t *vd, uint64_t avz, dmu_tx_t *tx)
 {
 	spa_t *spa = vd->vdev_spa;
 
+	if (vd->vdev_root_zap != 0 &&
+	    spa_feature_is_active(spa, SPA_FEATURE_AVZ_V2)) {
+		VERIFY0(zap_add_int(spa->spa_meta_objset, avz,
+		    vd->vdev_root_zap, tx));
+	}
 	if (vd->vdev_top_zap != 0) {
 		VERIFY0(zap_add_int(spa->spa_meta_objset, avz,
 		    vd->vdev_top_zap, tx));
diff --git a/module/zfs/vdev.c b/module/zfs/vdev.c
index 241be8fd8..4bfd95861 100644
--- a/module/zfs/vdev.c
+++ b/module/zfs/vdev.c
@@ -397,7 +397,9 @@ vdev_prop_get_int(vdev_t *vd, vdev_prop_t prop, uint64_t *value)
 	uint64_t objid;
 	int err;
 
-	if (vd->vdev_top_zap != 0) {
+	if (vd->vdev_root_zap != 0) {
+		objid = vd->vdev_root_zap;
+	} else if (vd->vdev_top_zap != 0) {
 		objid = vd->vdev_top_zap;
 	} else if (vd->vdev_leaf_zap != 0) {
 		objid = vd->vdev_leaf_zap;
@@ -898,6 +900,14 @@ vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent, uint_t id,
 	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_CREATE_TXG,
 	    &vd->vdev_crtxg);
 
+	if (vd->vdev_ops == &vdev_root_ops &&
+	    (alloctype == VDEV_ALLOC_LOAD ||
+	    alloctype == VDEV_ALLOC_SPLIT ||
+	    alloctype == VDEV_ALLOC_ROOTPOOL)) {
+		(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_VDEV_ROOT_ZAP,
+		    &vd->vdev_root_zap);
+	}
+
 	/*
 	 * If we're a top-level vdev, try to load the allocation parameters.
 	 */
@@ -3347,6 +3357,12 @@ vdev_construct_zaps(vdev_t *vd, dmu_tx_t *tx)
 				vdev_zap_allocation_data(vd, tx);
 		}
 	}
+	if (vd->vdev_ops == &vdev_root_ops && vd->vdev_root_zap == 0 &&
+	    spa_feature_is_enabled(vd->vdev_spa, SPA_FEATURE_AVZ_V2)) {
+		if (!spa_feature_is_active(vd->vdev_spa, SPA_FEATURE_AVZ_V2))
+			spa_feature_incr(vd->vdev_spa, SPA_FEATURE_AVZ_V2, tx);
+		vd->vdev_root_zap = vdev_create_link_zap(vd, tx);
+	}
 
 	for (uint64_t i = 0; i < vd->vdev_children; i++) {
 		vdev_construct_zaps(vd->vdev_child[i], tx);
@@ -5683,12 +5699,17 @@ vdev_props_set_sync(void *arg, dmu_tx_t *tx)
 		/*
 		 * Set vdev property values in the vdev props mos object.
 		 */
-		if (vd->vdev_top_zap != 0) {
+		if (vd->vdev_root_zap != 0) {
+			objid = vd->vdev_root_zap;
+		} else if (vd->vdev_top_zap != 0) {
 			objid = vd->vdev_top_zap;
 		} else if (vd->vdev_leaf_zap != 0) {
 			objid = vd->vdev_leaf_zap;
 		} else {
-			panic("vdev not top or leaf");
+			/*
+			 * XXX: implement vdev_props_set_check()
+			 */
+			panic("vdev not root/top/leaf");
 		}
 
 		switch (prop = vdev_name_to_prop(propname)) {
@@ -5891,7 +5912,9 @@ vdev_prop_get(vdev_t *vd, nvlist_t *innvl, nvlist_t *outnvl)
 
 	nvlist_lookup_nvlist(innvl, ZPOOL_VDEV_PROPS_GET_PROPS, &nvprops);
 
-	if (vd->vdev_top_zap != 0) {
+	if (vd->vdev_root_zap != 0) {
+		objid = vd->vdev_root_zap;
+	} else if (vd->vdev_top_zap != 0) {
 		objid = vd->vdev_top_zap;
 	} else if (vd->vdev_leaf_zap != 0) {
 		objid = vd->vdev_leaf_zap;
diff --git a/module/zfs/vdev_label.c b/module/zfs/vdev_label.c
index f61be65a2..85c7134ca 100644
--- a/module/zfs/vdev_label.c
+++ b/module/zfs/vdev_label.c
@@ -573,6 +573,12 @@ vdev_config_generate(spa_t *spa, vdev_t *vd, boolean_t getstats,
 			    vd->vdev_top_zap);
 		}
 
+		if (vd->vdev_ops == &vdev_root_ops && vd->vdev_root_zap != 0 &&
+		    spa_feature_is_active(vd->vdev_spa, SPA_FEATURE_AVZ_V2)) {
+			fnvlist_add_uint64(nv, ZPOOL_CONFIG_VDEV_ROOT_ZAP,
+			    vd->vdev_root_zap);
+		}
+
 		if (vd->vdev_resilver_deferred) {
 			ASSERT(vd->vdev_ops->vdev_op_leaf);
 			ASSERT(spa->spa_resilver_deferred);