[Devel] Re: [RFC] memory controller : backgorund reclaim and avoid excessive locking [1/5] high-low watermark

[KAMEZAWA Hiroyuki]

On Thu, 14 Feb 2008 14:18:33 +0530
Balbir Singh <balbir at linux.vnet.ibm.com> wrote:
> > If I have to check under lock, please teach me.
> > 
> 
> If there are several processes running in parallel in the same cgroup, the end
> result might not be so nice, specially if the usage is close to the watermarks.
> I suspect that we should  be OK for now, but might be worth keeping in mind.
> 
I'll add text somewhere.

> > -	counter->usage += val;
> > +        if (newval > counter->hwmark) {
> > +		counter->wmark_state = RES_WMARK_ABOVE_HIGH;
> > +		smp_wmb();
> 
> Do we need a barrier here? I suspect not, could you please document as to why a
> barrier is needed?
> 

just chainging value with smp_wmb() and read value after smp_rmb().
By this, I think we can expect we can read snapshot value of wmark_state at
smp_rmb().
......I misunderstand that spin_unlock() has no barrier().
ok, I'll remove smp_wmb() here.

Thanks,
-Kame



_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers