Andi Kleen wrote:
> It's hidden now so it causes any obvious failures any more. Just
> subtle ones which is much worse.
>
I think anything detected by Xen is still classed as "obscure" ;)
> But why not just disable it? It's not critical functionality,
> just a optimization that unfortunately turned out to be incorrect.
>
> It could be made correct short term by not freeing the pages until
> vunmap for example.
>
I think it only ends up holding 64 pages (or is it 64 mappings?), so its
not a horrible use of memory. Particularly since it responds to memory
pressure callbacks.
>> does his grand unified vmap manager. I guess a clean workaround would
>> be to add a CONFIG_XFS_LAZY_UNMAP, and do it at the Kconfig level...
>> I'll cook up a patch.
>>
>
> This option could only be safely set in architectures that don't
> care about caching attribute aliases or never remap kernel pages
> uncached. That's not x86, powerpc, ia64 at a minimum.
>
> I think the only good short term fix is to turn your ifdef CONFIG_XEN
> into a #if 0
>
#if 1, but yes, I see your point.
J
|