[PATCH] btrfs: fix negative space_info->bytes_readonly

[PATCH] btrfs: fix negative space_info->bytes_readonly

[Naohiro Aota]

Consider we have a using block group on zoned btrfs.

|<- ZU ->|<- used ->|<---free--->|
                     `- Alloc offset
ZU: Zone unusable

Marking the block group read-only will migrate the zone unusable bytes
to the read-only bytes. So, we will have this.

|<- RO ->|<- used ->|<--- RO --->|
RO: Read only

When marking it back to read-write, btrfs_dec_block_group_ro()
subtracts the above "RO" bytes from the
space_info->bytes_readonly. And, it moves the zone unusable bytes back
and again subtracts those bytes from the space_info->bytes_readonly,
leading to negative bytes_readonly.

This commit fixes the issue by reordering the operations.

Link: https://github.com/naota/linux/issues/37
Fixes: 169e0da91a21 ("btrfs: zoned: track unusable bytes for zones")
Cc: stable@vger.kernel.org # 5.12+
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
---
 fs/btrfs/block-group.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 38885b29e6e5..c42b6528552f 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -2442,16 +2442,16 @@ void btrfs_dec_block_group_ro(struct btrfs_block_group *cache)
 	spin_lock(&sinfo->lock);
 	spin_lock(&cache->lock);
 	if (!--cache->ro) {
-		num_bytes = cache->length - cache->reserved -
-			    cache->pinned - cache->bytes_super -
-			    cache->zone_unusable - cache->used;
-		sinfo->bytes_readonly -= num_bytes;
 		if (btrfs_is_zoned(cache->fs_info)) {
 			/* Migrate zone_unusable bytes back */
 			cache->zone_unusable = cache->alloc_offset - cache->used;
 			sinfo->bytes_zone_unusable += cache->zone_unusable;
 			sinfo->bytes_readonly -= cache->zone_unusable;
 		}
+		num_bytes = cache->length - cache->reserved -
+			    cache->pinned - cache->bytes_super -
+			    cache->zone_unusable - cache->used;
+		sinfo->bytes_readonly -= num_bytes;
 		list_del_init(&cache->ro_list);
 	}
 	spin_unlock(&cache->lock);
-- 
2.32.0

Re: [PATCH] btrfs: fix negative space_info->bytes_readonly

[Johannes Thumshirn]

Looks good,
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>

Re: [PATCH] btrfs: fix negative space_info->bytes_readonly

[David Sterba]

On Thu, Jun 17, 2021 at 01:56:18PM +0900, Naohiro Aota wrote:
> Consider we have a using block group on zoned btrfs.
> 
> |<- ZU ->|<- used ->|<---free--->|
>                      `- Alloc offset
> ZU: Zone unusable
> 
> Marking the block group read-only will migrate the zone unusable bytes
> to the read-only bytes. So, we will have this.
> 
> |<- RO ->|<- used ->|<--- RO --->|
> RO: Read only
> 
> When marking it back to read-write, btrfs_dec_block_group_ro()
> subtracts the above "RO" bytes from the
> space_info->bytes_readonly. And, it moves the zone unusable bytes back
> and again subtracts those bytes from the space_info->bytes_readonly,
> leading to negative bytes_readonly.
> 
> This commit fixes the issue by reordering the operations.
> 
> Link: https://github.com/naota/linux/issues/37

I've copied the 'fi df' output to changelog.

> Fixes: 169e0da91a21 ("btrfs: zoned: track unusable bytes for zones")
> Cc: stable@vger.kernel.org # 5.12+
> Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>

Added to misc-next, thanks.