Re: [PATCH] mm: memcontrol: fix memcg accounting during cpu hotplug

[Andrew Guerrero]

On 2025-09-12 12:45 UTC, Greg KH wrote:
> On Mon, Sep 08, 2025 at 09:09:00PM +0000, Andrew Guerrero wrote:
> > On 2025-09-07 13:10 UTC, Greg KH wrote:
> > > On Sat, Sep 06, 2025 at 03:21:08AM +0000, Andrew Guerrero wrote:
> > > > This patch is intended for the 5.10 longterm release branch. It will not apply
> > > > cleanly to mainline and is inadvertantly fixed by a larger series of changes in 
> > > > later release branches:
> > > > a3d4c05a4474 ("mm: memcontrol: fix cpuhotplug statistics flushing").
> > > 
> > > Why can't we take those instead?
> > > 
> > > > In 5.15, the counter flushing code is completely removed. This may be another
> > > > viable option here too, though it's a larger change.
> > > 
> > > If it's not needed anymore, why not just remove it with the upstream
> > > commits as well?
> > 
> > Yeah, my understanding is the typical flow is to pull commits from upstream into
> > stable branches. However, I'm not confident I know the the answer to "which
> > upstream commits?" To get started,
> > 
> > `git log -L :memcg_hotplug_cpu_dead:mm/memcontrol.c linux-5.10.y..linux-5.15.y`
> > 
> > tells me that the upstream changes to pull are:
> > 
> > - https://lore.kernel.org/all/20210209163304.77088-1-hannes@xxxxxxxxxxx/T/#u
> > - https://lore.kernel.org/all/20210716212137.1391164-1-shakeelb@xxxxxxxxxx/T/#u
> > 
> > However, these are substantial features that "fix" the issue indirectly by
> > transitioning the memcg accounting system over to rstats. I can pick these 10
> > upstream commits, but I'm worried I may overlook some additional patches from
> > 5.15.y that need to go along with them. I may need some guidance if we go this
> > route.
> 
> Testing is key :)
> 
> > Another reasonable option is to take neither route. We can maintain this patch
> > internally and then drop it once we upgrade to a new kernel version.
> 
> Perhaps just do that for now if you all are hitting this issue?  It
> seems to be the only report I've seen so far.

We are hitting this issue only in a stress test, and I think we got lucky with
experiencing it, so I wouldn't be too surprised if this is the first and only
report.

Thanks for taking a look!

Andrew