Re: [PATCH] cpuacct: VIRT_CPU_ACCOUNTING don't prevent percpu cputime count

[Bharata B Rao <bharata@linux.vnet.ibm.com>]

On Wed, Apr 29, 2009 at 11:34:15AM +0530, Balbir Singh wrote:
> * Bharata B Rao <bharata@linux.vnet.ibm.com> [2009-04-29 08:51:46]:
> 
> > On Wed, Apr 29, 2009 at 03:38:54AM +0530, Balbir Singh wrote:
> > > * KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> [2009-04-28 15:53:32]:
> > > > 
> > > >  	do {
> > > > -		percpu_counter_add(&ca->cpustat[idx], val);
> > > > +		__percpu_counter_add(&ca->cpustat[idx], val, cpuacct_batch);
> > > 
> > > This will make the end result very off the real value due to large
> > > batch value per cpu. If we are going to go this route, we should
> > > probably consider using __percpu_counter_sum so that the batch value
> > > does not show data that is way off.
> > 
> > But __percpu_counter_sum takes fbc->lock spinlock and that is the cause of
> > the problem Kosaki mentions I believe.
> 
> But this is on the user space read side (show stats), I thought
> Kosaki's problem was with per_cpu_counter_add called from update
> stats.

Right. Agreed.

Regards,
Bharata.