Re: [RFC][PATCH 1/3] radix priority search tree - objrmap complexity fix
From: Andrea Arcangeli
Date: Fri Mar 26 2004 - 14:18:58 EST
On Fri, Mar 26, 2004 at 04:26:36AM -0800, William Lee Irwin III wrote:
> On Thu, Mar 25, 2004 at 11:59:19PM +0100, Andrea Arcangeli wrote:
> > btw, the truncate of hugetlbfs didn't serialize correctly against the
> > do_no_page page faults, that's fixed too.
>
> If a fault on hugetlb ever got as far as do_no_page() on ia32, the
> kernel would oops on the bogus struct page it gets out of the bogus
> pte. I believe the way faults are handled in out-of-tree patches if by
> calling hugetlb-specific fault handling stacks instead of
> handle_mm_fault() if hugetlb vmas are found by arch code.
>
this is certainly true, but still the pmd fault handling should have the
same locking of do_no_page, the race sounds the same, no matter if it's
a pmd or pte fill, no?
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/