Re: [RFC] Generalized prio_tree, revisited

[Con Kolivas]

Werner Almesberger wrote:
> did you have a chance to look at the prio_tree generalization ?
>
> I've attached the patch I posted a month ago (plus the trivial
> "const" change). It's for 2.6.9, but also applies to 2.6.10-rc2,
> which I've been using for a good while now.
>
> The patch splits the radix priority search trees into two types:
> the "raw" one with implicit keys, as it's currently used, and a
> new, generalized one with explicit keys, which should be used by
> new code.

>  struct prio_tree_root {
>  	struct prio_tree_node	*prio_tree_node;
> -	unsigned int 		index_bits;
> +	unsigned short 		index_bits;
> +	unsigned short		raw;
> +		/*
> +		 * 0: nodes are of type struct prio_tree_node
> +		 * 1: nodes are of type raw_prio_tree_node
> +		 */
>  };

> -	INIT_PRIO_TREE_ROOT(&inode->i_data.i_mmap);
> +	INIT_PRIO_TREE_ROOT(&inode->i_data.i_mmap, 1);

While not being able to comment on the actual patch I think having a 1 
or 0 for different types is not clear. Naming them different struct 
names would seem to me much more readable.

Cheers,
Con

Attachment: signature.asc
Description: OpenPGP digital signature