Re: [PATCH 1/5] btrfs: free qgroup rsv on ioerr ordered_extent

[Qu Wenruo <quwenruo.btrfs@gmx.com>]

On 2023/12/2 07:30, Boris Burkov wrote:
> An ordered extent completing is a critical moment in qgroup rsv
> handling, as the ownership of the reservation is handed off from the
> ordered extent to the delayed ref. In the happy path we release (unlock)
> but do not free (decrement counter) the reservation, and the delayed ref
> drives the free. However, on an error, we don't create a delayed ref,
> since there is no ref to add. Therefore, free on the error path.

And I believe this would cause btrfs to be noisy at the unmount time,
due to unreleased qgroup rsv.

Have you hit any one during your tests? If so, mind to add some dmesg
output for it?

Or if no hit so far, would it be possible to add a new test case for it?

>
> Signed-off-by: Boris Burkov <boris@bur.io>

Reviewed-by: Qu Wenruo <wqu@suse.com>

Thanks,
Qu

> ---
>   fs/btrfs/ordered-data.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
> index 574e8a55e24a..8d4ab5ecfa5d 100644
> --- a/fs/btrfs/ordered-data.c
> +++ b/fs/btrfs/ordered-data.c
> @@ -599,7 +599,8 @@ void btrfs_remove_ordered_extent(struct btrfs_inode *btrfs_inode,
>   			release = entry->disk_num_bytes;
>   		else
>   			release = entry->num_bytes;
> -		btrfs_delalloc_release_metadata(btrfs_inode, release, false);
> +		btrfs_delalloc_release_metadata(btrfs_inode, release,
> +						test_bit(BTRFS_ORDERED_IOERR, &entry->flags));
>   	}
>
>   	percpu_counter_add_batch(&fs_info->ordered_bytes, -entry->num_bytes,