Re: [RFC PATCH] KVM: x86: Skip request checking branches in vcpu_enter_guest() more effectively

[Avi Kivity <avi@redhat.com>]

On 09/24/2012 08:59 AM, Xiao Guangrong wrote:
> On 09/24/2012 02:24 PM, Takuya Yoshikawa wrote:
>> This is an RFC since I have not done any comparison with the approach
>> using for_each_set_bit() which can be seen in Avi's work.
>> 
> 
> Why not compare it? I think for_each_set_bit is better and it can
> improve for all cases (in your patch, you did not consider all case,
> for example, if the guest under mm overcommit, it should generate
> lots of TLB flush/RELOAD request).
> 
> Actually, i think Avi's way can be improved in the further, we can
> just use one atomic operation to avoid cache-miss. May be like this:
> 
> while (vcpu->request) {
> 	xchg(vcpu->request, request);
> 
> 	for_each_set_bit(request) {
> 		clear_bit(X);
> 
> 		......
> 	}
> 	
> }

In fact I had something like that in one of the earlier versions, but it
was problematic.  I'll try to see what the issue was.


-- 
error compiling committee.c: too many arguments to function