RFR (S): 8078669: G1 applies SurvivorAlignmentInBytes to both survivor and old gen

Thomas Schatzl thomas.schatzl at oracle.com
Tue Jun 2 12:54:40 UTC 2015


Hi all,

On Mon, 2015-05-18 at 10:58 +0200, Thomas Schatzl wrote: 
> Hi,
> 
> On Wed, 2015-05-13 at 21:57 +0300, Filipp Zhinkin wrote:
> > Hi Thomas,
> > 
> > On Wed, May 13, 2015 at 5:33 PM, Thomas Schatzl
> > <thomas.schatzl at oracle.com> wrote:
> > > Hi all,
> > >
> > >   can I have reviews for the following small fix: since JDK-8060025
> > > survivor alignment has been applied to both survivor and old gen regions
> > > (if enabled of course).
> > >
> > > This has gone unnoticed, because the tests somehow detected that in this
> > > situation everyhing is fine due to "someone else that allocated
> > > objects".
> > 
> > Without such check tests may fail when executed with JFR or some agent.
> > If it's an issue, then I guess it will be better to improve it instead
> > of eliminating at all.
> 
>   do you have any suggestion to fix this issue? I see no way how the
> java application or the VM would be able to distinguish between objects
> allocated by the application or any java agent.
> 
> This check seems to make the test useless as at the moment it will
> always "detect" that someone else allocated objects.

  since there were no particular comments on how to fix this, and I do
not have any good idea about this either, I reverted the change that
removed that check.

Also because while testing some unrelated change, this test failed once.

Webrev:
http://cr.openjdk.java.net/~tschatzl/8078669/webrev.1/
Incremental webrev:
http://cr.openjdk.java.net/~tschatzl/8078669/webrev.0_to_1/

I created JDK-8081688 to investigate alternatives for the test.

Thanks,
  Thomas




More information about the hotspot-gc-dev mailing list