Re: [PATCH 1/3] maple_tree: use ma_data_end() in mas_data_end()

["Liam R. Howlett" <Liam.Howlett@oracle.com>]

* Wei Yang <richard.weiyang@gmail.com> [240830 20:11]:
> These two function share almost the same code and do the same thing.

Please stop trying to optimise code like this.  I don't have time to
verify you are not making things worse and you haven't done any of the
testing required.

I went through with perf to optimise dereferences and static inline
functions, so unless you are prepared to look at the generated code and
actually benchmark, please stop sending patches that I need to verify
like this myself.

> 
> Let's just call ma_data_end() in mas_data_end() to reduce duplicate
> code.
> 
> Signed-off-by: Wei Yang <richard.weiyang@gmail.com>
> ---
>  lib/maple_tree.c | 22 ++++------------------
>  1 file changed, 4 insertions(+), 18 deletions(-)
> 
> diff --git a/lib/maple_tree.c b/lib/maple_tree.c
> index b7d747a7938e..85668246f944 100644
> --- a/lib/maple_tree.c
> +++ b/lib/maple_tree.c
> @@ -1435,28 +1435,14 @@ static __always_inline unsigned char ma_data_end(struct maple_node *node,
>   */
>  static inline unsigned char mas_data_end(struct ma_state *mas)
>  {
> -	enum maple_type type;
> -	struct maple_node *node;
> -	unsigned char offset;
> -	unsigned long *pivots;
> -
> -	type = mte_node_type(mas->node);
> -	node = mas_mn(mas);
> -	if (type == maple_arange_64)
> -		return ma_meta_end(node, type);
> +	enum maple_type type = mte_node_type(mas->node);
> +	struct maple_node *node = mas_mn(mas);
> +	unsigned long *pivots = ma_pivots(node, type);
>  
> -	pivots = ma_pivots(node, type);
>  	if (unlikely(ma_dead_node(node)))
>  		return 0;
>  
> -	offset = mt_pivots[type] - 1;
> -	if (likely(!pivots[offset]))
> -		return ma_meta_end(node, type);
> -
> -	if (likely(pivots[offset] == mas->max))
> -		return offset;
> -
> -	return mt_pivots[type];
> +	return ma_data_end(node, type, pivots, mas->max);
>  }
>  
>  /*
> -- 
> 2.34.1
> 
>