Re: [PATCH 3/5] locking/lockdep: Add a faster path in __lock_release()

[Ingo Molnar]

* Waiman Long <longman@xxxxxxxxxx> wrote:

> When __lock_release() is called, the most likely unlock scenario is
> on the innermost lock in the chain.  In this case, we can skip some of
> the checks and provide a faster path to completion.
> 
> Signed-off-by: Waiman Long <longman@xxxxxxxxxx>
> ---
>  kernel/locking/lockdep.c | 17 ++++++++++++++---
>  1 file changed, 14 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c
> index add0468..ca002c0 100644
> --- a/kernel/locking/lockdep.c
> +++ b/kernel/locking/lockdep.c
> @@ -3625,6 +3625,13 @@ static int __lock_downgrade(struct lockdep_map *lock, unsigned long ip)
>  	curr->lockdep_depth = i;
>  	curr->curr_chain_key = hlock->prev_chain_key;
>  
> +	/*
> +	 * The most likely case is when the unlock is on the innermost
> +	 * lock. In this case, we are done!
> +	 */
> +	if (i == depth - 1)
> +		return 1;
> +
>  	if (reacquire_held_locks(curr, depth, i + 1))
>  		return 0;
>  
> @@ -3632,10 +3639,14 @@ static int __lock_downgrade(struct lockdep_map *lock, unsigned long ip)
>  	 * We had N bottles of beer on the wall, we drank one, but now
>  	 * there's not N-1 bottles of beer left on the wall...
>  	 */
> -	if (DEBUG_LOCKS_WARN_ON(curr->lockdep_depth != depth - 1))
> -		return 0;
> +	DEBUG_LOCKS_WARN_ON(curr->lockdep_depth != depth - 1);
>  
> -	return 1;
> +	/*
> +	 * Since reacquire_held_locks() would have called check_chain_key()
> +	 * indirectly via __lock_acquire(), we don't need to do it again
> +	 * on return.
> +	 */
> +	return 0;

Minor nit:

	s/depth - 1/depth-1

for slightly better readability.

Thanks,

	Ingo