Re: [PATCH v8] mm: make lru_add_drain_all() selective

[Tejun Heo <tj@kernel.org>]

Hello,

On Wed, Aug 14, 2013 at 01:44:30PM -0700, Andrew Morton wrote:
> > +static bool need_activate_page_drain(int cpu)
> > +{
> > +	return pagevec_count(&per_cpu(activate_page_pvecs, cpu)) != 0;
> > +}
> 
> static int need_activate_page_drain(int cpu)
> {
> 	return pagevec_count(&per_cpu(activate_page_pvecs, cpu));
> }
> 
> would be shorter and faster.  bool rather sucks that way.  It's a
> performance-vs-niceness thing.  I guess one has to look at the call
> frequency when deciding.

"!= 0" can be dropped but I'm fairly sure the compiler would be able
to figure out that the type conversion can be skipped.  It's a trivial
optimization.

> > +			schedule_work_on(cpu, work);
> > +			cpumask_set_cpu(cpu, &has_work);
> > +		}
> > +	}
> > +
> > +	for_each_cpu(cpu, &has_work)
> 
> for_each_online_cpu()?

That would lead to flushing work items which aren't used and may not
have been initialized yet, no?

> > +		flush_work(&per_cpu(lru_add_drain_work, cpu));
> > +
> > +	put_online_cpus();
> > +	mutex_unlock(&lock);
> >  }
> 

Thanks.

-- 
tejun