Re: [PATCH 1/3] Reintroduce zone_reclaim_interval for whenzone_reclaim() scans and fails to avoid CPU spinning at 100% on NUMA

[Mel Gorman]

On Tue, Jun 09, 2009 at 06:59:03PM +0900, KOSAKI Motohiro wrote:
> > > > > Here it is just recording the jiffies value. The real smarts with the counter
> > > > > use time_before() which I assumed could handle jiffie wrap-arounds. Even
> > > > > if it doesn't, the consequence is that one scan will occur that could have
> > > > > been avoided around the time of the jiffie wraparound. The value will then
> > > > > be reset and it will be fine.
> > > > 
> > > > time_before() assume two argument are enough nearly time.
> > > > if we use 32bit cpu and HZ=1000, about jiffies wraparound about one month.
> > > > 
> > > > Then, 
> > > > 
> > > > 1. zone reclaim failure occur
> > > > 2. system works fine for one month
> > > > 3. jiffies wrap and time_before() makes mis-calculation.
> > > > 
> > > 
> > > And the scan occurs uselessly and zone_reclaim_failure gets set again.
> > > I believe the one useless scan is not significant enough to warrent dealing
> > > with jiffie wraparound.
> > 
> > Thank you for kindful explanation.
> > I fully agreed.
> 
> Bah, no, not agreed.
> simple last failure recording makes following scenario.
> 
> 
> 1. zone reclaim failure occur. update zone_reclaim_failure.
>       ^
>       |  time_before() return 1, and zone_reclaim() return immediately.
>       v
> 2. after 32 second.
>       ^
>       |  time_before() return 0, and zone_reclaim() works normally
>       v
> 3. after one month
>       ^
>       |  time_before() return 1, and zone_reclaim() return immediately.
>       |  although recent zone_reclaim() never failed.
>       v
> 4. after more one month
>       

Pants.

/me slaps self

+       /* Watch for jiffie wraparound */
+       if (unlikely(jiffies < zone->zone_reclaim_failure))
+               zone->zone_reclaim_failure = jiffies;
+
+       /* Do not attempt a scan if scanning failed recently */
+       if (time_before(jiffies,
+                       zone->zone_reclaim_failure + zone_reclaim_interval))
+               return 0;
+

?

-- 
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/