Re: [PATCH 2/2] smp_call_function: use rwlocks on queues rather than rcu

[Andi Kleen <andi@firstfloor.org>]

> > Ah so it was already 25% slower even without kmalloc? I thought
> > that was with already. That doesn't sound good. Any idea where that slowdown 
> > comes from?
> 
> Just longer code path, I think.  It calls the generic

I did IPI measurements quite some time ago and what I remember
from them is that IPI latencies were in the low multiple thousands cycle
ballpark.

> smp_call_function_mask(), which then does a popcount on the cpu mask
> (which it needs to do anyway), sees only one bit set, and then punts to
> the smp_call_function_single() path.  

But that is more in the a few tens of cycles (or maybe 1-2 hundreds
if you have a NR_CPU==4096 kernel with really large cpumask)

Doesn't really explain a 25% slowdown I would say. 

Are you sure there isn't a new cache miss in there or something? Actually
it must be even multiple ones to account for such a slow down.

-Andi

-- 
ak@linux.intel.com