RFR (S/M): 8164936: G1 age table printout contains contents from previous GC

Thomas Schatzl thomas.schatzl at oracle.com
Wed Aug 31 09:54:48 UTC 2016


Hi Sangheon,

  thanks for your review.

On Tue, 2016-08-30 at 12:00 -0700, sangheon wrote:
> Hi Thomas,
> 
> On 08/30/2016 07:48 AM, Thomas Schatzl wrote:
> > 
> > Hi all,
> > 
> >    can I have reviews for this change that fixes the age table
> > output for the g1 collector, where the data printed is always the
> > age data for the previous gc?
> > 
> > The fix is as suggested in the CR, splitting computation of the new
> > tenuring threshold and printing the age table information.
> > 
> > (The majority of the change is the new test which is very
> > straightforward)
> > 
> > CR:
> > https://bugs.openjdk.java.net/browse/JDK-8164936
> > Webrev:
> > http://cr.openjdk.java.net/~tschatzl/8164936/webrev/
> > Testing:
> > jprt, test case, local testing with some applications
> Looks good.
> 
> I think you intended to extract 'gc counter' to focus the work as the
> method named.If it is the case, I think
> 
[...]
> should be called after line 573 when UsePerfData is enabled.
> 
> And the include of 'gcPolicyCounters.hpp' at ageTable.cpp and forward
> declaration of 'GCPolicyCounters' at ageTable.hpp can be removed.

Fixed in http://cr.openjdk.java.net/~tschatzl/8164936/webrev.1 (full)
and http://cr.openjdk.java.net/~tschatzl/8164936/webrev.0_to_1
(incremental).

Thanks,
  Thomas



More information about the hotspot-gc-dev mailing list