Re: [PATCH 1/2] btrfs: introduce conditional wakeup helpers

[Nikolay Borisov <n.borisov.lkml@gmail.com>]

On  8.03.2018 13:49, David Sterba wrote:
> Add convenience wrappers for the waitqueue management that involves
> memory barriers to prevent deadlocks. The helpers will let us remove
> barriers and the necessary comments in several places.
> 
> Signed-off-by: David Sterba <dsterba@suse.com>

Reviewed-by: Nikolay Borisov <nborisov@suse.com>

> ---
>  fs/btrfs/ctree.h | 23 +++++++++++++++++++++++
>  1 file changed, 23 insertions(+)
> 
> diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
> index bf545e7552b8..c9827a1b676d 100644
> --- a/fs/btrfs/ctree.h
> +++ b/fs/btrfs/ctree.h
> @@ -3748,4 +3748,27 @@ static inline int btrfs_is_testing(struct btrfs_fs_info *fs_info)
>  #endif
>  	return 0;
>  }
> +
> +static inline void cond_wake_up(struct wait_queue_head *wq)
> +{
> +	/*
> +	 * This implies a full smp_mb barrier, see comments for
> +	 * waitqueue_active why.
> +	 */
> +	if (wq_has_sleeper(wq))
> +		wake_up(wq);
> +}
> +
> +static inline void cond_wake_up_nomb(struct wait_queue_head *wq)
> +{
> +	/*
> +	 * Special case for conditional wakeup where the barrier required for
> +	 * waitqueue_active is implied by some of the preceding code. Eg. one
> +	 * of such atomic operations (atomic_dec_and_return, ...), or a
> +	 * unlock/lock sequence, etc.
> +	 */
> +	if (waitqueue_active(wq))
> +		wake_up(wq);
> +}
> +
>  #endif
>