Messages in this thread | | | Date | Tue, 17 May 2011 09:38:02 -0500 (CDT) | From | Christoph Lameter <> | Subject | Re: [patch V3] percpu_counter: scalability works |
| |
On Tue, 17 May 2011, Tejun Heo wrote:
> We can implement force-flush for dire situations, similar to expedited > RCU flush thing.
I would suggest that if the user wants accurate results then another external form of synchronization needs to be provided. Maybe the counter is always incremented when another lock has been taken.
| |