Hi,
thanks do David and Staffan it was possible to pinpoint the contention
to the static initializaton and classloading.
David suggested running the test routine twice to exercise any
initialization during the first run and performs the checks only for the
results of the second run.
The test is using Phaser for synchronization. All the potentially
blocking code (System.out.println(), Thread.currentThread(), etc.) was
removed from the tested thread.
The webrev is available at
http://cr.openjdk.java.net/~jbachorik/8030847/webrev.01
Thanks,
-JB-
