Re: [PATCH mm-unstable v1 2/3] mm/migrate.c: Prevent folio splitting from interacting with KSM

[Usama Arif <usama.arif@linux.dev>]

On Tue, 14 Jul 2026 22:05:31 -0600 Nico Pache <npache@redhat.com> wrote:

> On Tue, Jun 9, 2026 at 11:27 AM Usama Arif <usama.arif@linux.dev> wrote:
> >
> > On Tue, 09 Jun 2026 10:07:20 -0400 Zi Yan <ziy@nvidia.com> wrote:
> >
> > > On 9 Jun 2026, at 9:47, xu.xin16@zte.com.cn wrote:
> > >
> > > >>>> Since commit b1f202060afe ("mm: remap unused subpages to shared zeropage
> > > >>>> when splitting isolated thp"), splitting an anonymous THP remaps all
> > > >>>> zero-filled subpages to the shared zeropage via TTU_USE_SHARED_ZEROPAGE.
> > > >>>> This flag is set unconditionally for every anonymous folio split,
> > > >>>> including splits triggered by KSM.
> > > >>>>
> > > >>>> When KSM is enabled with THP=always, this causes two regressions:
> > > >>>>
> > > >>>> 1. use_zero_pages=1: KSM calls try_to_merge_one_page() which triggers
> > > >>>>   split_huge_page(). The split remaps all 512 zero-filled subpages to
> > > >>>>   the shared zeropage at once, freeing the entire 2MB THP when KSM only
> > > >>>>   intended to process a single 4KB page. This bypasses KSM's
> > > >>>>   pages_to_scan rate limiting, causing ~1GB to be freed almost
> > > >>>>   instantly.
> > > >>>>
> > > >>>
> > > >>> Why do you see it as regressions?
> > > >>
> > > >> Since the zero-page remapping was introduced our test has shown the
> > > >> following behavior changes:
> > > >>
> > > >> With use_zero_pages=0, the merge rate drops from 60MB/s to ~6 MB/s
> > > >> even after raising pages_to_scan. The KSM merging is now much slower
> > > >> and CPU utilization has increased.
> > > >>
> > > >> With use_zero_pages=1, ~1 GB is freed almost instantly, and it no
> > > >> longer respects the pages_to_scan behavior.
> > > >>
> > > >> Even with just this patch (1 & 2) or the RFC linked in the cover
> > > >> letter, the issue no longer occurs.
> > > >
> > > > Understood. You're saying that the additional processing action in split_huge_page
> > > > (remap unused subpages to shared zeropage) increases the scanning cost of ksmd.
> > > >
> > > > However, I still wouldn't simply classify this as a performance regression,
> > > > because commit b1f202060afe increases memory savings through this action — so
> > > > it saves memory at the cost of additional CPU overhead.
> > > >
> > > > If you want to address the increased overhead on ksmd, I think we could add a
> > > > check for the shared zeropage in cmp_and_merge_page, and skip merging when a
> > > > shared zeropage is detected.
> > > >
> > > >>>
> > > >>> AFAIU, KSM and THP do often conflict with each other. THP tries hard to collapse
> > > >>> a huge page (which may contain many zero pages). If KSM is enabled and part of
> > > >>> that huge page is mergeable, it can easily be split by KSM, rendering THP's
> > > >>> efforts futile.
> > > >>>
> > > >>> Therefore, in our actual production environment, we typically avoid making the
> > > >>> same region both KSM mergeable and THP always.
> > > >>
> > > >> THP=always is a global setting used in many production environments,
> > > >> so these features now interact very poorly together.
> > > >>
> > > >
> > > > Actually, I have long thought about submitting a patch: add a new interface
> > > > 'skip_huge_page' under KSM's sysfs, allowing users to choose not to split huge pages.
> 
> This would be a good intermediate solution while allowing backward
> compatibility.
> 
> > >
> > > Just think out loud. Or just skip huge pages all the time unless memory pressure
> > > is present. Basically treat KSM as a way of reducing memory pressure by merging
> > > pages.
> 
> Hmm interesting. We should decide on future changes before committing
> to one direction; however, i think the immediate solution would be
> what I currently have prepped for v2.
> 
> My v2 follows my RFC alternative approach: skip zeropage remapping if
> the split comes from KSM. It also contains the patch (3) from this
> version, which skips zeropage remapping if the shrinker is disabled.
> 
> >
> > I do agree with this. I questioned about if it even makes sense to split THP
> > with KSM in RFC (https://lore.kernel.org/all/20260510114001.600681-1-usama.arif@linux.dev/)
> 
> This is indeed a good question but I don't know the right solution.
> 
> Perhaps we can discuss it here.
> 
> So far here are the potential solutions:
> a) skip zeropage remapping if the split comes from KSM
> b) skip zeropage remapping completely if the shrinker is disabled
> c) Disable KSM THP splitting completely
> d) add a skip_huge_page sysfs toggle
> e) only scan THPs when memory pressure is present
> 
> (a)+(b) would be my V2 which I can send out whenever.
> 
> (c) is the nuclear option, but perhaps KSM for THP is rather
> inefficient and pointless

I really like and want (c), but I know David has already pointed out
that there might be someone already deploying and expecting this.

> 
> (d) could be used with (a) and (b) and allows more flexibility than (c)
> 
> (e) is a more dynamic approach, and could be combined with (b) and (d)
> 

(e) sounds like a good idea, but then KSM effectively becomes a THP shrinker?

> Let me know what you think :)
> -- Nico
> 
> >
> >
> > >
> > > Best Regards,
> > > Yan, Zi
> > >
> >
> 
>