On Thu, Nov 28, 2013 at 12:10:40PM +0100, Paolo Bonzini wrote:
> Il 28/11/2013 12:09, Gleb Natapov ha scritto:
> > > - if there are no callbacks, but there are readers, synchronize_srcu
> > > busy-loops for some time checking if the readers complete. After a
> > > while (20 us for synchronize_srcu, 120 us for
> > > synchronize_srcu_expedited) it gives up and starts using a workqueue to
> > > poll every millisecond. This should never happen unless
> >
> > Unless what ? :) Unless reader is scheduled out?
>
> Yes. Or unless my brain is scheduled out in the middle of a sentence.
>
So we will have to disable preemption in a reader to prevent big latencies for
a writer, no?
--
Gleb.
