ppc64: Make hash_preload() and update_mmu_cache() cope with hugepages

[David Gibson]

On Tue, Nov 08, 2005 at 02:41:47PM +1100, David Gibson wrote:
> Paulus, please apply.
> 
> At present, hash_preload() (and hence update_mmu_cache()) will not
> work correctly if called on a hugepage address, thus relying on the
> fact that the hugepage fault paths never call update_mmu_cache().  I'm
> not 100% sure that's safe now (although I think it is), and it
> certainly won't be safe for some of the places we want to go with
> hugepage.
> 
> Thus, this patch extends hash_preload() to work correctly on hugepage
> addresses.

Bleh, sorry Paul, don't apply.  The patch is slightly wrong and the
description is largely wrong.  In fact the current version of
update_mmu_cache() will not blow up on hugepages, and making it do
more is merely an optimization which is probably only important in the
case of hugepage COW.  I'll resend something which will actually work
to implement that optimization if COW is looking promising for
merging.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson