Re: [mm/page_alloc] f26b3fa046: netperf.Throughput_Mbps -18.0% regression

[Mel Gorman]

On Thu, May 05, 2022 at 04:27:04PM +0800, Aaron Lu wrote:
> On Fri, Apr 29, 2022 at 02:39:18PM +0100, Mel Gorman wrote:
> > On Fri, Apr 29, 2022 at 07:29:19PM +0800, Aaron Lu wrote:
> 
> ... ...
> 
> > > The said change looks like this:
> > > (relevant comment will have to be adjusted)
> > > 
> > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > > index 505d59f7d4fa..130a02af8321 100644
> > > --- a/mm/page_alloc.c
> > > +++ b/mm/page_alloc.c
> > > @@ -3332,18 +3332,19 @@ static int nr_pcp_high(struct per_cpu_pages *pcp, struct zone *zone,
> > >  		       bool free_high)
> > >  {
> > >  	int high = READ_ONCE(pcp->high);
> > > +	int batch = READ_ONCE(pcp->batch);
> > >  
> > > -	if (unlikely(!high || free_high))
> > > +	if (unlikely(!high))
> > >  		return 0;
> > >  
> > > -	if (!test_bit(ZONE_RECLAIM_ACTIVE, &zone->flags))
> > > -		return high;
> > > -
> > >  	/*
> > >  	 * If reclaim is active, limit the number of pages that can be
> > >  	 * stored on pcp lists
> > >  	 */
> > > -	return min(READ_ONCE(pcp->batch) << 2, high);
> > > +	if (test_bit(ZONE_RECLAIM_ACTIVE, &zone->flags) || free_high)
> > > +		return min(batch << 2, high);
> > > +
> > > +	return high;
> > >  }
> > >  
> > >  static void free_unref_page_commit(struct page *page, int migratetype,
> > > 
> > > Does this look sane? If so, I can prepare a formal patch with proper
> > > comment and changelog, thanks.
> > 
> > I think it looks reasonable sane. The corner case is that if
> > ((high - (batch >> 2)) > cachesize) that the pages will not get recycled
> 
> When free_high is true, the above diff changed the return value of
> nr_pcp_high() from 0 to min(batch << 2, pcp->high) so the corner case is
> when (min(batch << 2, pcp->high) > cachesize)?
> 

Yes. It's not perfect due to cache aliasing so the actual point where it
matters will be variable. Whatever the value is, there a value where the
corner case applies that pages do not get recycled quickly enough and
are no longer cache-hot.

-- 
Mel Gorman
SUSE Labs