Re: [Xen-devel] Xen spinlock questions

["Jan Beulich" <jbeulich@xxxxxxxxxx>]

>>> Jeremy Fitzhardinge <jeremy@xxxxxxxx> 11.08.08 20:10 >>>
>Jan Beulich wrote:
>> Running kernel builds on 8 vCPU-s competing for 4 pCPU-s shows a 10%
>> improvement in performance with the individual wakeup (patch attached
>> - probably sub-optimal, but I didn't seem to be able to think of a lock-less
>> mechanism to achieve the desired behavior), using ticket locks in the
>> kernel.
>
>What exactly are you measuring here?  Are you testing your 2.6.2x 
>forward-port with an adaptation of pv spinlocks using ticket locks as 
>the underlying locking mechanism?  Does the 10% improvement come 

Yes.

>compared to straightforward use of sched_poll, or compared to doing 
>spurious wakeup detection?

Compared to spurious wakeup detection done in the kernel, similar to
how you had done it in the latest version of your respective patch that
I have seen.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel