On Thu, Jun 7, 2012 at 4:47 PM, Paolo Bonzini <pbonzini@xxxxxxxxxx> wrote:
> Il 07/06/2012 03:38, Asias He ha scritto:
>>> > I never looked at the kvmtool code, but I think Asias has a point. If
>>> > the guest submits I/O to lots of devices, they would contend on the
>>> > single thread. There was a similar proof of concept patch for QEMU that
>>> > provided substantial benefit.
>>> >
>>> > However, it sounds a bit wasteful to have the dedicated thread run with
>>> > a second eventfd. Would it be hard to just use the virtio-blk ioeventfd
>>> > directly?
>> Without this patch, we are already using the virtio-blk ioeventfd
>> which is being monitored in the shared ioeventfd__thread() thead.
>
> Yes, I understood that. :)
;-)
> I'm wondering why it is still necessary to
> monitor that eventfd in the shared thread. Perhaps you could optionally
> skip the epoll registration/deregistration in ioeventfd__{add,del}_event.
Yes, we can monitor the eventfd only in block thread. But that
requires changing our virtio core code and doing ioeventfd setup in
block specific code. This still does not save a thread for us. And the
overhead of using shared eventfd + notify to block thead should be
small. So I'd prefer the current model.
--
Asias He
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
