Re: [PATCH mm-unstable v1] mm: add a total mapcount for large folios

[Peter Xu <peterx@redhat.com>]

On Thu, Aug 10, 2023 at 10:37:04AM +0200, David Hildenbrand wrote:
> On 10.08.23 05:25, Matthew Wilcox wrote:
> > On Wed, Aug 09, 2023 at 05:23:46PM -0400, Peter Xu wrote:
> > > Hi, David,
> > > 
> > > Some pure questions below..
> > > 
> > > On Wed, Aug 09, 2023 at 10:32:56AM +0200, David Hildenbrand wrote:
> > > > Let's track the total mapcount for all large folios in the first subpage.
> > > > 
> > > > The total mapcount is what we actually want to know in folio_mapcount()
> > > > and it is also sufficient for implementing folio_mapped(). This also
> > > > gets rid of any "raceiness" concerns as expressed in
> > > > folio_total_mapcount().
> > > 
> > > Any more information for that "raciness" described here?
> > 
> > UTSL.
> > 
> >          /*
> >           * Add all the PTE mappings of those pages mapped by PTE.
> >           * Limit the loop to folio_nr_pages_mapped()?
> >           * Perhaps: given all the raciness, that may be a good or a bad idea.
> >           */
> > 
> 
> Yes, that comment from Hugh primarily discusses how we could possibly
> optimize the loop, and if relying on folio_nr_pages_mapped() to reduce the
> iterations would be racy. As far as I can see, there are cases where "it
> would be certainly a bad idea" :)

Is the race described about mapcount being changed right after it's read?
Are you aware of anything specific that will be broken, and will be fixed
with this patch?

I assume mapcount==1 will be very special in this case when e.g. holding a
pgtable lock, other than that I won't be surprised if mapcount changes in
parallel.  But I must confess I don't really have any thorough digests on
this whole matter.

> 
> 
> In the other comment in that function, it's also made clear what the
> traditional behavior with PMD-mappable THP was "In the common case, avoid
> the loop when no pages mapped by PTE", which will no longer hold with
> sub-PMD THP.

Having a total mapcount does sound helpful if partial folio is common
indeed.

I'm curious whether that'll be so common after the large anon folio work -
isn't it be sad if partial folio will be a norm?  It sounds to me that's
the case when small page sizes should be used.. and it's prone to waste?

-- 
Peter Xu