1 files changed, 74 insertions, 33 deletions

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 4423d8b716a5..83a196521670 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -34,6 +34,19 @@ int btrfs_should_fragment_free_space(const struct btrfs_block_group *block_group
 }
 #endif
 
+static inline bool has_unwritten_metadata(struct btrfs_block_group *block_group)
+{
+	/* The meta_write_pointer is available only on the zoned setup. */
+	if (!btrfs_is_zoned(block_group->fs_info))
+		return false;
+
+	if (block_group->flags & BTRFS_BLOCK_GROUP_DATA)
+		return false;
+
+	return block_group->start + block_group->alloc_offset >
+		block_group->meta_write_pointer;
+}
+
 /*
  * Return target flags in extended format or 0 if restripe for this chunk_type
  * is not in progress
@@ -1249,6 +1262,15 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 		goto out;
 
 	spin_lock(&block_group->lock);
+	/*
+	 * Hitting this WARN means we removed a block group with an unwritten
+	 * region. It will cause "unable to find chunk map for logical" errors.
+	 */
+	if (WARN_ON(has_unwritten_metadata(block_group)))
+		btrfs_warn(fs_info,
+			   "block group %llu is removed before metadata write out",
+			   block_group->start);
+
 	set_bit(BLOCK_GROUP_FLAG_REMOVED, &block_group->runtime_flags);
 
 	/*
@@ -1460,6 +1482,32 @@ out:
 }
 
 /*
+ * Link the block_group to a list via bg_list.
+ *
+ * @bg:       The block_group to link to the list.
+ * @list:     The list to link it to.
+ *
+ * Use this rather than list_add_tail() directly to ensure proper respect
+ * to locking and refcounting.
+ *
+ * Returns: true if the bg was linked with a refcount bump and false otherwise.
+ */
+static bool btrfs_link_bg_list(struct btrfs_block_group *bg, struct list_head *list)
+{
+	struct btrfs_fs_info *fs_info = bg->fs_info;
+	bool added = false;
+
+	spin_lock(&fs_info->unused_bgs_lock);
+	if (list_empty(&bg->bg_list)) {
+		btrfs_get_block_group(bg);
+		list_add_tail(&bg->bg_list, list);
+		added = true;
+	}
+	spin_unlock(&fs_info->unused_bgs_lock);
+	return added;
+}
+
+/*
  * Process the unused_bgs list and remove any that don't have any allocated
  * space inside of them.
  */
@@ -1567,15 +1615,15 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		 * needing to allocate extents from the block group.
 		 */
 		used = btrfs_space_info_used(space_info, true);
-		if (space_info->total_bytes - block_group->length < used &&
-		    block_group->zone_unusable < block_group->length) {
+		if ((space_info->total_bytes - block_group->length < used &&
+		     block_group->zone_unusable < block_group->length) ||
+		    has_unwritten_metadata(block_group)) {
 			/*
 			 * Add a reference for the list, compensate for the ref
 			 * drop under the "next" label for the
 			 * fs_info->unused_bgs list.
 			 */
-			btrfs_get_block_group(block_group);
-			list_add_tail(&block_group->bg_list, &retry_list);
+			btrfs_link_bg_list(block_group, &retry_list);
 
 			trace_btrfs_skip_unused_block_group(block_group);
 			spin_unlock(&block_group->lock);
@@ -1598,8 +1646,10 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		ret = btrfs_zone_finish(block_group);
 		if (ret < 0) {
 			btrfs_dec_block_group_ro(block_group);
-			if (ret == -EAGAIN)
+			if (ret == -EAGAIN) {
+				btrfs_link_bg_list(block_group, &retry_list);
 				ret = 0;
+			}
 			goto next;
 		}
 
@@ -1891,6 +1941,17 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 			up_write(&space_info->groups_sem);
 			goto next;
 		}
+
+		/*
+		 * Cache the zone_unusable value before turning the block group
+		 * to read only. As soon as the block group is read only it's
+		 * zone_unusable value gets moved to the block group's read-only
+		 * bytes and isn't available for calculations anymore. We also
+		 * cache it before unlocking the block group, to prevent races
+		 * (reports from KCSAN and such tools) with tasks updating it.
+		 */
+		zone_unusable = bg->zone_unusable;
+
 		spin_unlock(&bg->lock);
 		spin_unlock(&space_info->lock);
 
@@ -1907,13 +1968,6 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 			goto next;
 		}
 
-		/*
-		 * Cache the zone_unusable value before turning the block group
-		 * to read only. As soon as the blog group is read only it's
-		 * zone_unusable value gets moved to the block group's read-only
-		 * bytes and isn't available for calculations anymore.
-		 */
-		zone_unusable = bg->zone_unusable;
 		ret = inc_block_group_ro(bg, 0);
 		up_write(&space_info->groups_sem);
 		if (ret < 0)
@@ -1944,20 +1998,8 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 		spin_unlock(&space_info->lock);
 
 next:
-		if (ret && !READ_ONCE(space_info->periodic_reclaim)) {
-			/* Refcount held by the reclaim_bgs list after splice. */
-			spin_lock(&fs_info->unused_bgs_lock);
-			/*
-			 * This block group might be added to the unused list
-			 * during the above process. Move it back to the
-			 * reclaim list otherwise.
-			 */
-			if (list_empty(&bg->bg_list)) {
-				btrfs_get_block_group(bg);
-				list_add_tail(&bg->bg_list, &retry_list);
-			}
-			spin_unlock(&fs_info->unused_bgs_lock);
-		}
+		if (ret && !READ_ONCE(space_info->periodic_reclaim))
+			btrfs_link_bg_list(bg, &retry_list);
 		btrfs_put_block_group(bg);
 
 		mutex_unlock(&fs_info->reclaim_bgs_lock);
@@ -1997,13 +2039,8 @@ void btrfs_mark_bg_to_reclaim(struct btrfs_block_group *bg)
 {
 	struct btrfs_fs_info *fs_info = bg->fs_info;
 
-	spin_lock(&fs_info->unused_bgs_lock);
-	if (list_empty(&bg->bg_list)) {
-		btrfs_get_block_group(bg);
+	if (btrfs_link_bg_list(bg, &fs_info->reclaim_bgs))
 		trace_btrfs_add_reclaim_block_group(bg);
-		list_add_tail(&bg->bg_list, &fs_info->reclaim_bgs);
-	}
-	spin_unlock(&fs_info->unused_bgs_lock);
 }
 
 static int read_bg_from_eb(struct btrfs_fs_info *fs_info, const struct btrfs_key *key,
@@ -2776,8 +2813,12 @@ void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans)
 		/* Already aborted the transaction if it failed. */
 next:
 		btrfs_dec_delayed_refs_rsv_bg_inserts(fs_info);
+
+		spin_lock(&fs_info->unused_bgs_lock);
 		list_del_init(&block_group->bg_list);
 		clear_bit(BLOCK_GROUP_FLAG_NEW, &block_group->runtime_flags);
+		btrfs_put_block_group(block_group);
+		spin_unlock(&fs_info->unused_bgs_lock);
 
 		/*
 		 * If the block group is still unused, add it to the list of
@@ -2915,7 +2956,7 @@ struct btrfs_block_group *btrfs_make_block_group(struct btrfs_trans_handle *tran
 	}
 #endif
 
-	list_add_tail(&cache->bg_list, &trans->new_bgs);
+	btrfs_link_bg_list(cache, &trans->new_bgs);
 	btrfs_inc_delayed_refs_rsv_bg_inserts(fs_info);
 
 	set_avail_alloc_bits(fs_info, type);