Re: [RFC PATCH v1 1/4] mm: Introduce ptep_get_lockless_norecency()

[David Hildenbrand]

> > > > > diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> > > > > index 68283e54c899..41dc44eb8454 100644
> > > > > --- a/mm/hugetlb.c
> > > > > +++ b/mm/hugetlb.c
> > > > > @@ -7517,7 +7517,7 @@ pte_t *huge_pte_alloc(struct mm_struct *mm, struct
> > > > > vm_area_struct *vma,
> > > > >         }
> > > > >
> > > > >         if (pte) {
> > > > > -        pte_t pteval = ptep_get_lockless(pte);
> > > > > +        pte_t pteval = ptep_get_lockless_norecency(pte);
> > > > >
> > > > >             BUG_ON(pte_present(pteval) && !pte_huge(pteval));
> > > > >         }
> > > > > diff --git a/mm/khugepaged.c b/mm/khugepaged.c
> > > > > index 2771fc043b3b..1a6c9ed8237a 100644
> > > > > --- a/mm/khugepaged.c
> > > > > +++ b/mm/khugepaged.c
> > > > > @@ -1019,7 +1019,7 @@ static int __collapse_huge_page_swapin(struct mm_struct
> > > > > *mm,
> > > > >                 }
> > > > >             }
> > > > >
> > > > > -        vmf.orig_pte = ptep_get_lockless(pte);
> > > > > +        vmf.orig_pte = ptep_get_lockless_norecency(pte);
> > > > >             if (!is_swap_pte(vmf.orig_pte))
> > > > >                 continue;
> > > >
> > > >
> > > > Hm, I think you mentioned that we want to be careful with vmf.orig_pte.
> > >
> > > Yeah good point. So I guess this should move to patch 3 (which may be dropped -
> > > tbd)?
> >
> > Yes. Or a separate one where you explain in detail why do_swap_page() can handle
> > it just fine.
>
> Ahh no wait - I remember now; the reason I believe this is a "trivial" case is
> because we only leak vmf.orig_pte to the rest of the world if its a swap entry.

Ugh, yes!

--
Cheers,

David / dhildenb