On Fri 16-12-11 17:02:51, Glauber Costa wrote:

On 12/16/2011 04:32 PM, Michal Hocko wrote:

[...]

quoted

So why do we need kmem accounting when tcp (the only user at the moment)
doesn't use it?

Well, a bit historical. I needed a basic placeholder for it, since
it tcp is officially kmem. As the time passed, I took most of the
stuff out of this patch to leave just the basics I would need for
tcp.
Turns out I ended up focusing on the rest, and some of the stuff was
left here.

At one point I merged tcp data into kmem, but then reverted this
behavior. the kmem counter stayed.

I agree deferring the whole behavior would be better.

quoted

quoted

In summary, we still never do non-independent accounting. When we
start doing it for the other caches, We will have to add a test at
charge time as well.

So we shouldn't do it as a part of this patchset because the further
usage is not clear and I think there will be some real issues with
user+kmem accounting (e.g. a proper memcg-oom implementation).
Can you just drop this patch?

Yes, but the whole set is in the net tree already. 

Isn't it only in some for-next branch? Can that one be updated?

(All other patches are tcp-related but this) Would you mind if I'd
send a follow up patch removing the kmem files, and leaving just the
registration functions and basic documentation? (And sorry for that as
well in advance)

Yes a followup patch would work as well.

-- 
Michal Hocko
SUSE Labs
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9    
Czech Republic
--
To unsubscribe from this list: send the line "unsubscribe cgroups" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html