RFR (S): 8136398: Create test that checks that G1 MixedGC produces correct output to logfile

Thomas Schatzl thomas.schatzl at oracle.com
Wed Sep 23 08:05:02 UTC 2015


Hi Kirill,

On Fri, 2015-09-18 at 14:35 +0300, Kirill Zhaldybin wrote:
> Hi!
> 
> Could you please review a patch for JDK-8136398?
> 
> I added very simple test that starts JVM  with "-XX:+PrintGC" or 
> "-XX:+PrintGCDetails" flags, provokes mixed gc and checks that 
> corresponding log entry appears in gc log.
> 
> CR: https://bugs.openjdk.java.net/browse/JDK-8136398
> 
> Webrev:
> 
> hotspot: http://cr.openjdk.java.net/~kzhaldyb/webrevs/JDK-8136398/webrev.00/

  - the @requires tag to enforce use of G1 is missing.

  - I do not really like the busy wait for completion of the concurrent
marking in lines 140 and 143 of the test. This will (for a short while
at least) max out one cpu. Maybe a sleep(0) would "help" here?

  - I am not sure what "MixedGC" class is referenced in the comment of
the MixedGCProvoker method. There does not seem to be a WhiteBox method
to provoke a mixed gc either.

  - there is a small typo in that same comment: I think the "promote"
verb in the second sentence requires a third person singular "s" here.

Thanks,
  Thomas




More information about the hotspot-gc-dev mailing list