Re: [RFC PATCH RESEND 08/28] mm/khugepaged: mark VMA as locked while collapsing a hugepage

From: Suren Baghdasaryan
Date: Thu Sep 08 2022 - 20:16:01 EST


On Tue, Sep 6, 2022 at 7:43 AM Laurent Dufour <ldufour@xxxxxxxxxxxxx> wrote:
>
> Le 01/09/2022 à 19:34, Suren Baghdasaryan a écrit :
> > Protect VMA from concurrent page fault handler while modifying it in
> > collapse_huge_page.
>
> Is the goal to protect changes in the anon_vma structure?
>
> AFAICS, the vma it self is not impacted here, only the anon_vma and the
> PMD/PTE are touched, and they have their own protection mechanism, isn't it?

Yes, I think you are right about not needing to lock VMA here as all
modified components are already protected. Thanks!

>
> >
> > Signed-off-by: Suren Baghdasaryan <surenb@xxxxxxxxxx>
> > ---
> > mm/khugepaged.c | 1 +
> > 1 file changed, 1 insertion(+)
> >
> > diff --git a/mm/khugepaged.c b/mm/khugepaged.c
> > index 01f71786d530..030680633989 100644
> > --- a/mm/khugepaged.c
> > +++ b/mm/khugepaged.c
> > @@ -1072,6 +1072,7 @@ static void collapse_huge_page(struct mm_struct *mm,
> > if (mm_find_pmd(mm, address) != pmd)
> > goto out_up_write;
> >
> > + vma_mark_locked(vma);
> > anon_vma_lock_write(vma->anon_vma);
> >
> > mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, NULL, mm,
>