Re: [PATCH RFC V2 3/5] kvm hypervisor : Add two hypercalls to support pv-ticketlock

[Avi Kivity <avi@redhat.com>]

On 10/26/2011 09:08 PM, Raghavendra K T wrote:
> On 10/26/2011 04:04 PM, Avi Kivity wrote:
>> On 10/25/2011 08:24 PM, Raghavendra K T wrote:
> CCing Ryan also
>>>
>>> So then do also you foresee the need for directed yield at some point,
>>> to address LHP? provided we have good improvements to prove.
>>
>> Doesn't this patchset completely eliminate lock holder preemption?
>>
> Basically I was curious whether we can do more better with your
> directed yield discussions in https://lkml.org/lkml/2010/8/2/106 .
>
> I felt we can get little more improvement with doing directed yield to
> lock-holder in case of LHP than sleeping. But I may be wrong.
>
> So wanted to get the feedback, on whether I am thinking in right
> direction.

i guess donating some time to the lock holder could help, but not by
much. The problem with non-pv spinlocks is that you can't just sleep,
since no one will wake you up, so you have to actively boost the lock
holder.

-- 
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.