powerpc/mm: Move around testing of _PAGE_PRESENT in hash code
authorBenjamin Herrenschmidt <benh@kernel.crashing.org>
Thu, 22 Jul 2010 22:53:23 +0000 (08:53 +1000)
committerBenjamin Herrenschmidt <benh@kernel.crashing.org>
Thu, 22 Jul 2010 22:53:23 +0000 (08:53 +1000)
Instead of adding _PAGE_PRESENT to the access permission mask
in each low level routine independently, we add it once from
hash_page().

We also move the preliminary access check (the racy one before
the PTE is locked) up so it applies to the huge page case. This
duplicates code in __hash_page_huge() which we'll remove in a
subsequent patch to fix a race in there.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

No differences found