RFR (S): 8136854: G1 ConcurrentG1RefineThread::stop delays JVM shutdown for >150ms

Tom Benson tom.benson at oracle.com
Wed Jan 27 14:37:29 UTC 2016

Was the change tried on a small system to see if startup is actually 
slowed down there, by the much tighter loop consuming CPU?  I guess one 
possible answer is "we don't care that much about systems that small for 
G1. "  But Derek suggested an increasing delay scheme in the CR, which I 
don't think would be much more difficult than the current change, and 
could address that.

On 1/27/2016 7:09 AM, Thomas Schatzl wrote:
> Hi all,
>    can I have reviews for this small fix to decrease startup time of the
> G1 collector by increasing the polling frequency of the main thread to
> look whether all concurrent mark threads already completed their
> initialization.
> There are some additional volatile declarations of variables that are
> set by multiple threads.
> We are aware that this might not be an ideal solution, some kind of
> wait/notify scheme for the mark threads might be better. That would be
> much more work though. I created JDK-8148359 for this work.
> CR:
> https://bugs.openjdk.java.net/browse/JDK-8136854
> Webrev:
> http://cr.openjdk.java.net/~tschatzl/8136854/webrev/
> Testing:
> jprt, local testing, dev-submit perf testing
> Thanks,
>    Thomas

More information about the hotspot-gc-dev mailing list