Re: [PATCH RFC v2] mm: use per_vma lock for MADV_DONTNEED

[Lorenzo Stoakes <lorenzo.stoakes@oracle.com>]

On Tue, Jun 03, 2025 at 01:06:15PM +1200, Barry Song wrote:
> Hi Lorenzo,
>
> On Mon, Jun 2, 2025 at 11:50 PM Lorenzo Stoakes
> <lorenzo.stoakes@oracle.com> wrote:
> >
> > Barry - would you mind if I went off and wrote a quick patch to separate
> > walk_page_range_novma() into walk_page_range_kernel() and
> > walk_page_range_user_novma()?
> >
> > I realise this is a pain, but I feel strongly that having them together is
> > a source of confusion and having us special case a -only used in ptdump-
> > case like this is not great.
> >
> > It wouldn't include any VMA locking stuff from this patch, which you could
> > then rebase on that.
> >
> > I think it'd make sense as a separate series and I can throw that out
> > fairly quickly...
> >
> > But I don't want to step on any toes so just let me know!
>
> Feel free to proceed with the patch. I can rebase on top of your
> walk_page_range_kernel() changes.

Thanks! Will send that out hopefully today.

I will also come back about how I feel the abstraction here should work, sorry
getting tied up with other stuff so got delayed on that...!

>
> Thanks
> Barry