Re: [RFC][PATCH 1/2] memcg: Ensure every task that uses an mm is in the same memory cgroup

[Eric W. Biederman]

Tejun Heo <tj@xxxxxxxxxx> writes:

>> This effectively forbids process with mm's shared processes being migrated.
>> Although enabling the control file might work.
>
> So, I don't think we need to support putting tasks which share a mm in
> different cgroups.  That said, if we're gonna put in that restriction,
> I think it should be in cgroup core rather than memcg can_attach.  The
> only thing we'd need to do is widening what cgroup migration considers
> to be a process.

Widening the definition of a process sounds good.  The memory control
group code would still need a way to forbid these in cgroup v1 mode,
when someone uses the task file.

Using widening instead of denying should reduce the risk of introducing
a regression.

The only reason I support the crazy case in my earlier patch is so that
we can have this discussion and in case we do cause a regression with
this change the previous algorithmic cleanup won't have to be reverted
as well.

Eric