Re: [RFC][PATCH 1/3] mm/debug-pagealloc.c: Split out page poisoning from debug page_alloc

[Laura Abbott]

On 01/25/2016 10:26 PM, Jianyu Zhan wrote:
> On Tue, Jan 26, 2016 at 12:55 AM, Laura Abbott
> <labbott@xxxxxxxxxxxxxxxxx> wrote:
> > +static bool __page_poisoning_enabled __read_mostly;
> > +static bool want_page_poisoning __read_mostly =
> > +       !IS_ENABLED(CONFIG_ARCH_SUPPORTS_DEBUG_PAGEALLOC);
> > +
>
>
> I would say this patch is nice with regard to decoupling
> CONFIG_DEBUG_PAGEALLOC and CONFIG_PAGE_POISONING.
>
> But  since when we enable CONFIG_DEBUG_PAGEALLOC,
> CONFIG_PAGE_POISONING will be selected.
>
> So it would be better to make page_poison.c totally
> CONFIG_DEBUG_PAGEALLOC agnostic,  in case we latter have
> more PAGE_POISONING users(currently only DEBUG_PAGEALLOC ). How about like this:
>
> +static bool want_page_poisoning __read_mostly =
> +       !IS_ENABLED(CONFIG_PAGE_POISONING );
>
> Or just let it default to 'true',  since we only compile this
> page_poison.c when we enable CONFIG_PAGE_POISONING.

This patch was just supposed to be the refactor and keep the existing
behavior. There are no Kconfig changes here and the existing behavior
is to poison if !ARCH_SUPPORTS_DEBUG_PAGEALLOC so I think keeping
what I have is appropriate for this particular patch. This can be
updated in another series if appropriate.

Thanks,
Laura
 
> Thanks,
> Jianyu Zhan