Re: [PATCH] sparse: Track the boundaries of memory sections for accurate checks

[Dave Hansen <dave.hansen@intel.com>]

On 06/18/2016 03:11 AM, KarimAllah Ahmed wrote:
> @@ -1067,8 +1067,12 @@ struct mem_section {
>  	 * section. (see page_ext.h about this.)
>  	 */
>  	struct page_ext *page_ext;
> -	unsigned long pad;
> +	unsigned long pad[3];
>  #endif
> +
> +	unsigned long first_pfn;
> +	unsigned long last_pfn;

mem_section started out as a single pointer, and it's getting a bit, um,
rotund.  Remember, some architectures have a lot of physical address
space, and thus a ton of mem_sections, but very little actual memory.
This eats valuable RAM if we bloat mem_section too much.

Oh, and with this:

>         /*
>          * WARNING: mem_section must be a power-of-2 in size for the
>          * calculation and use of SECTION_ROOT_MASK to make sense.
>          */

Aren't you making a non-power-of-2 sized 'mem_section" in some cases?

This also doesn't handle if there is a hole in the _middle_ of a section.

What's wrong with using some bits in the existing usemap?

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>