RFR: 8135078: Refactor InCSetState::is_in_cset_or_humongous

Thomas Schatzl thomas.schatzl at oracle.com
Thu Oct 15 09:08:46 UTC 2015


Hi,

On Tue, 2015-10-06 at 17:43 +0200, Erik Helin wrote:
> Hi all,
> 
> this tiny patch slightly refactors the function
> InCSetState::is_in_cset_or_humongous. The new version decouples the
> function from the enum representation, this in turn enables future
> extensions of the enum wihout affecting the logic. The reason for
> InCSetState::is_in_cset_or_humongous was implemented as
> `_value != NotInCSet` was for performance reasons, but careful
> performance measurements shows that no regressions are introduced with
> this patch.
> 
> Webrev:
> http://cr.openjdk.java.net/~ehelin/8135078/webrev.00/
> 
> Testing:
> - JPRT
> 
> Performance testing:
> - SPECjbb2005 with a fixed load to get as stable runs as possible. No
>   regression in total gc time nor in object copy time.
> - Perf Aurora test run, no regressions

  looks good.

Thomas




More information about the hotspot-gc-dev mailing list