Messages in this thread | | | Date | Thu, 22 May 2014 14:25:30 -0500 (CDT) | From | Christoph Lameter <> | Subject | Re: [PATCH RFC 3/3] slub: reparent memcg caches' slabs on memcg offline |
| |
On Thu, 22 May 2014, Vladimir Davydov wrote:
> On Wed, May 21, 2014 at 07:13:21PM -0500, Christoph Lameter wrote: > > On Wed, 21 May 2014, Vladimir Davydov wrote: > > > > > Do I understand you correctly that the following change looks OK to you? > > > > Almost. Preemption needs to be enabled before functions that invoke the > > page allocator etc etc. > > I need to disable preemption only in slab_free, which never blocks > according to its semantics, so everything should be fine just like that.
slab_free calls __slab_free which can release slabs via put_cpu_partial()/unfreeze_partials()/discard_slab() to the page allocator. I'd rather have preemption enabled there.
| |