On Tue, 14 May 2019, Roman Gushchin wrote:
> To make this possible we need to introduce a new percpu refcounter
> for non-root kmem_caches. The counter is initialized to the percpu
> mode, and is switched to atomic mode after deactivation, so we never
> shutdown an active cache. The counter is bumped for every charged page
> and also for every running allocation. So the kmem_cache can't
> be released unless all allocations complete.
Increase refcounts during each allocation? Looks to be quite heavy
processing.
From: Christopher Lameter <[email protected]>
Date: Wed, May 15, 2019 at 7:00 AM
To: Roman Gushchin
Cc: Andrew Morton, Shakeel Butt, <[email protected]>,
<[email protected]>, <[email protected]>, Johannes Weiner,
Michal Hocko, Rik van Riel, Vladimir Davydov,
<[email protected]>
> On Tue, 14 May 2019, Roman Gushchin wrote:
>
> > To make this possible we need to introduce a new percpu refcounter
> > for non-root kmem_caches. The counter is initialized to the percpu
> > mode, and is switched to atomic mode after deactivation, so we never
> > shutdown an active cache. The counter is bumped for every charged page
> > and also for every running allocation. So the kmem_cache can't
> > be released unless all allocations complete.
>
> Increase refcounts during each allocation? Looks to be quite heavy
> processing.
Not really, it's a percpu refcnt. Basically the memcg's
percpu_ref_tryget* is replaced with kmem_cache's percpu_ref_tryget,
so, no additional processing.