btrfs: skip reclaim if block_group is empty

As we delete extents from a block group, at some deletion we cross below
the reclaim threshold. It is possible we are still in the middle of
deleting more extents and might soon hit 0. If the block group is empty
by the time the reclaim worker runs, we will still relocate it.

This works just fine, as relocating an empty block group ultimately
results in properly deleting it. However, we have more direct ways of
removing empty block groups in the cleaner thread. Those are either
async discard or the unused_bgs list. In fact, when we decide whether to
relocate a block group during extent deletion, we do check for emptiness
and prefer the discard/unused_bgs mechanisms when possible.

Not using relocation for this case reduces some modest overhead from
empty bg relocation:

- extra transactions
- extra metadata use/churn for creating relocation metadata
- trying to read the extent tree to look for extents (and in this case
  finding none)

Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Boris Burkov <boris@bur.io>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 3f8b1cb..684401a 100644 (file)
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -1606,6 +1606,24 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
                        up_write(&space_info->groups_sem);
                        goto next;
                }
+               if (bg->used == 0) {
+                       /*
+                        * It is possible that we trigger relocation on a block
+                        * group as its extents are deleted and it first goes
+                        * below the threshold, then shortly after goes empty.
+                        *
+                        * In this case, relocating it does delete it, but has
+                        * some overhead in relocation specific metadata, looking
+                        * for the non-existent extents and running some extra
+                        * transactions, which we can avoid by using one of the
+                        * other mechanisms for dealing with empty block groups.
+                        */
+                       if (!btrfs_test_opt(fs_info, DISCARD_ASYNC))
+                               btrfs_mark_bg_unused(bg);
+                       spin_unlock(&bg->lock);
+                       up_write(&space_info->groups_sem);
+                       goto next;
+               }
                spin_unlock(&bg->lock);
 
                /* Get out fast, in case we're unmounting the filesystem */