lkml.org 
[lkml]   [2015]   [May]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [RFC 3/3] memcg: get rid of mm_struct::owner
On Thu 28-05-15 17:07:42, Tejun Heo wrote:
> Hello, Johannes, Michal.
>
> On Tue, May 26, 2015 at 10:10:11AM -0400, Johannes Weiner wrote:
> > On Tue, May 26, 2015 at 01:50:06PM +0200, Michal Hocko wrote:
> > > Please note that this patch introduces a USER VISIBLE CHANGE OF BEHAVIOR.
> > > Without mm->owner _all_ tasks associated with the mm_struct would
> > > initiate memcg migration while previously only owner of the mm_struct
> > > could do that. The original behavior was awkward though because the user
> > > task didn't have any means to find out the current owner (esp. after
> > > mm_update_next_owner) so the migration behavior was not well defined
> > > in general.
> > > New cgroup API (unified hierarchy) will discontinue tasks file which
> > > means that migrating threads will no longer be possible. In such a case
> > > having CLONE_VM without CLONE_THREAD could emulate the thread behavior
> > > but this patch prevents from isolating memcg controllers from others.
> > > Nevertheless I am not convinced such a use case would really deserve
> > > complications on the memcg code side.
> >
> > I think such a change is okay. The memcg semantics of moving threads
> > with the same mm into separate groups have always been arbitrary. No
> > reasonable behavior can be expected of this, so what sane real life
> > usecase would rely on it?
>
> I suppose that making mm always follow the threadgroup leader should
> be fine, right?

That is the plan.

> While this wouldn't make any difference in the unified hierarchy,

Just to make sure I understand. "wouldn't make any difference" because
the API is not backward compatible right?

> I think this would make more sense for traditional hierarchies.

Yes I believe so.
--
Michal Hocko
SUSE Labs


\
 
 \ /
  Last update: 2015-05-29 14:41    [W:0.135 / U:1.800 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site