Re: [PATCH] drivers/mtd/ubi/wl.c: optimization for in_wl_tree()

[Daniel Santos <danielfsantos@att.net>]

Oh, and I forgot to mention that I haven't tested it (don't have this
device), but it does compile.

On 06/09/2012 08:57 PM, Daniel Santos wrote:
> Essentially, instead of performing a search through the tree from the
> top down, like we would when doing a lookup, we can just search from the
> node up and then check to see if the root of the tree that the node is
> in is the same as the one supplied.  Note that if you call this passing
> a ubi_wl_entry who's u.rb member has not been inited with rb_init_node()
> or added to a tree, the results are undefined.
>
> This should be a little faster, but mostly, it'll be smaller.
> ---
>  drivers/mtd/ubi/wl.c |   32 ++++++++------------------------
>  1 files changed, 8 insertions(+), 24 deletions(-)
>
> diff --git a/drivers/mtd/ubi/wl.c b/drivers/mtd/ubi/wl.c
> index 9df100a..d415011 100644
> --- a/drivers/mtd/ubi/wl.c
> +++ b/drivers/mtd/ubi/wl.c
> @@ -287,33 +287,17 @@ static int produce_free_peb(struct ubi_device *ubi)
>   */
>  static int in_wl_tree(struct ubi_wl_entry *e, struct rb_root *root)
>  {
> -	struct rb_node *p;
> -
> -	p = root->rb_node;
> -	while (p) {
> -		struct ubi_wl_entry *e1;
> +	struct rb_node *p = &e->u.rb, *parent = rb_parent(&e->u.rb);
>  
> -		e1 = rb_entry(p, struct ubi_wl_entry, u.rb);
> -
> -		if (e->pnum == e1->pnum) {
> -			ubi_assert(e == e1);
> -			return 1;
> -		}
> +	if (parent == p)
> +		return 0;
>  
> -		if (e->ec < e1->ec)
> -			p = p->rb_left;
> -		else if (e->ec > e1->ec)
> -			p = p->rb_right;
> -		else {
> -			ubi_assert(e->pnum != e1->pnum);
> -			if (e->pnum < e1->pnum)
> -				p = p->rb_left;
> -			else
> -				p = p->rb_right;
> -		}
> -	}
> +	do {
> +		p = parent;
> +		parent = rb_parent(parent);
> +	} while (parent);
>  
> -	return 0;
> +	return p == root->rb_node;
>  }
>  
>  /**

-- 
Daniel

"Just because you go to church on Sunday, it doesn't absolve you from
where you put your dick during the week" -- Michele Q.