On Wed, May 07, 2008 at 08:18:45PM +0100, Hugh Dickins wrote:
Hugh: Thanks for looking into this. Yes. I like your modified patch. Simpler
Yes. we can call it a loophole or defect.
Yes. That PAGE_NX part indeed looks bogus.
They cannot get out of sync. So, we can do without the change here. But, I felt
it is safer to take things from old pte.
Yes. We can mask out these bits from vm_page_prot, that way we will not
'or' them if they get out of sync.
If your below modified patch is OK, I can send an incremental patch to
change pte_modify() to inherit PAT bits only from oldpte.