On 18/10/2021 17:42, David Sterba wrote:

On Thu, Oct 14, 2021 at 06:39:02PM +0900, Johannes Thumshirn wrote:

quoted

Currently auto reclaim of unusable zones reclaims the block-groups in the
order they have been added to the reclaim list.

Change this to a greedy algorithm by sorting the list so we have the
block-groups with the least amount of valid bytes reclaimed first.

Signed-off-by: Johannes Thumshirn <redacted>

---
Changes since v2:
- Go back to the RFC state, as we must not access ->bg_list
  without taking the lock. (Nikolay)

Changes since v1:
-  Changed list_sort() comparator to 'boolean' style

Changes since RFC:
- Updated the patch description
- Don't sort the list under the spin_lock (David)
---
 fs/btrfs/block-group.c | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 7dba9028c80c..77e6224866c1 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c

@@ -1,5 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 
+#include <linux/list_sort.h>
+
 #include "misc.h"
 #include "ctree.h"
 #include "block-group.h"

@@ -1486,6 +1488,21 @@ void btrfs_mark_bg_unused(struct btrfs_block_group *bg)
 	spin_unlock(&fs_info->unused_bgs_lock);
 }
 
+/*
+ * We want block groups with a low number of used bytes to be in the beginning
+ * of the list, so they will get reclaimed first.
+ */
+static int reclaim_bgs_cmp(void *unused, const struct list_head *a,
+			   const struct list_head *b)
+{
+	const struct btrfs_block_group *bg1, *bg2;
+
+	bg1 = list_entry(a, struct btrfs_block_group, bg_list);
+	bg2 = list_entry(b, struct btrfs_block_group, bg_list);
+
+	return bg1->used - bg2->used;

So you also reverted to v1 the compare condition, this should be < so
it's the valid stable sort condition.

Ah damn, want me to resend with fixed commit message, subject and compare function?