On 08/15/2012 07:34 PM, Christoph Lameter wrote:

On Wed, 15 Aug 2012, Glauber Costa wrote:

quoted

On 08/15/2012 06:47 PM, Christoph Lameter wrote:

quoted

On Wed, 15 Aug 2012, Michal Hocko wrote:

quoted

quoted

That is not what the kernel does, in general. We assume that if he wants
that memory and we can serve it, we should. Also, not all kernel memory
is unreclaimable. We can shrink the slabs, for instance. Ying Han
claims she has patches for that already...

Are those patches somewhere around?

You can already shrink the reclaimable slabs (dentries / inodes) via
calls to the subsystem specific shrinkers. Did Ying Han do anything to
go beyond that?

That is not enough for us.
We would like to make sure that the objects being discarded belong to
the memcg which is under pressure. We don't need to be perfect here, and
an occasional slip is totally fine. But if in general, shrinking from
memcg A will mostly wipe out objects from memcg B, we harmed the system
in return for nothing good.

How can you figure out which objects belong to which memcg? The ownerships
of dentries and inodes is a dubious concept already.

Remember we copy over the metadata and create copies of the caches
per-memcg. Therefore, a dentry belongs to a memcg if it was allocated
from the slab pertaining to that memcg.

It is not 100 % accurate, but it is good enough.