RFR: 8138975: G1CollectorPolicy::calculate_young_list_target_length should be const
Thomas Schatzl
thomas.schatzl at oracle.com
Thu Oct 22 12:52:31 UTC 2015
Hi Erik,
On Thu, 2015-10-22 at 13:06 +0200, Erik Helin wrote:
> On 2015-10-19, Thomas Schatzl wrote:
> > Hi Erik,
> [...]
> > At the start we know that rs_length is zero, there is no need to get a
> > prediction from an empty sequence.
> >
> > Feel free to ignore, as that and related issues have already been filed
> > under JDK-8139594.
> >
> > I mention this, because adding this would remove the need for the
> > overload.
> >
> > Looks good otherwise. I do not need a re-review if you decide to change
> > this as suggested.
>
> Thanks for the review Thomas!
>
> When running some additional testing I discovered that I forgot to
> update a call site from using update_young_list_target_length to
> update_young_list_max_and_target_length. To make the patch behave
> exactly like the old code, I also added the method
> update_rs_lengths_prediction, because the old code only updated
> _rs_lengths_prediction if gcs_are_young() &&
> adaptive_young_list_length().
>
> Please see new webrevs:
> - full:
> http://cr.openjdk.java.net/~ehelin/8138975/webrev.03/
> - incremental:
> http://cr.openjdk.java.net/~ehelin/8138975/webrev.02-03/
looks good.
Thanks,
Thomas
More information about the hotspot-gc-dev
mailing list