Re: [PATCH 3/6] btrfs: introduce try-lock semantics for exclusive op start

[Anand Jain <anand.jain@oracle.com>]

On 21/05/2021 20:06, David Sterba wrote:
> Add try-lock for exclusive operation start to allow callers to do more
> checks. The same operation must already be running. The try-lock and
> unlock must pair and are a substitute for btrfs_exclop_start, thus it
> must also pair with btrfs_exclop_finish to release the exclop context.
> 
> Signed-off-by: David Sterba <dsterba@suse.com>
> ---
>   fs/btrfs/ctree.h |  3 +++
>   fs/btrfs/ioctl.c | 26 ++++++++++++++++++++++++++
>   2 files changed, 29 insertions(+)
> 
> diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
> index 3dfc32a3ebab..0dffc06b5ad4 100644
> --- a/fs/btrfs/ctree.h
> +++ b/fs/btrfs/ctree.h
> @@ -3231,6 +3231,9 @@ void btrfs_update_ioctl_balance_args(struct btrfs_fs_info *fs_info,
>   			       struct btrfs_ioctl_balance_args *bargs);
>   bool btrfs_exclop_start(struct btrfs_fs_info *fs_info,
>   			enum btrfs_exclusive_operation type);
> +bool btrfs_exclop_start_try_lock(struct btrfs_fs_info *fs_info,
> +				 enum btrfs_exclusive_operation type);
> +void btrfs_exclop_start_unlock(struct btrfs_fs_info *fs_info);
>   void btrfs_exclop_finish(struct btrfs_fs_info *fs_info);
>   
>   /* file.c */
> diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
> index c4e710ea08ba..cacd6ee17d8e 100644
> --- a/fs/btrfs/ioctl.c
> +++ b/fs/btrfs/ioctl.c
> @@ -371,6 +371,32 @@ bool btrfs_exclop_start(struct btrfs_fs_info *fs_info,
>   	return ret;
>   }
>   
> +/*
> + * Conditionally allow to enter the exclusive operation in case it's compatible
> + * with the running one.  This must be paired with btrfs_exclop_start_unlock and
> + * btrfs_exclop_finish.
> + *
> + * Compatibility:
> + * - the same type is already running
> + * - not BTRFS_EXCLOP_NONE - this is intentionally incompatible and the caller
> + *   must check the condition first that would allow none -> @type
> + */
> +bool btrfs_exclop_start_try_lock(struct btrfs_fs_info *fs_info,
> +				 enum btrfs_exclusive_operation type)
> +{
> +	spin_lock(&fs_info->super_lock);
> +	if (fs_info->exclusive_operation == type)
> +		return true;
> +
> +	spin_unlock(&fs_info->super_lock);
> +	return false;
> +}
> +

  Sorry for the late comment.
  Nit:
  This function implements a conditional lock. But the function name
  misleads to some operation similar to spin_trylock() or
  mutex_trylock(). How about btrfs_exclop_start_cond_lock() instead?
  Otherwise, looks good.

  Reviewed-by: Anand Jain <anand.jain@oracle.com>

Thanks.

> +void btrfs_exclop_start_unlock(struct btrfs_fs_info *fs_info)
> +{
> +	spin_unlock(&fs_info->super_lock);
> +}
> +
>   void btrfs_exclop_finish(struct btrfs_fs_info *fs_info)
>   {
>   	spin_lock(&fs_info->super_lock);
>