RFR(S): 8203943: eventThreadGroup was null in TestJavaBlockedEvent.java

Erik Gahlin erik.gahlin at oracle.com
Sat Jul 7 12:09:08 UTC 2018


The purpose of the dead flag is to prevent incorrect information, such as a null thread group, being written into a recording after the correct information has been serialized (in a previous checkpoint). The counter is needed now that not necessarily all threads are written at a check point, only those that are not dead. And all this happens during a safepoint, so no new threads can be started.

Ok, looks good.

Erik


 about thread being written into a chekcpoint

> On 7 Jul 2018, at 13:21, Markus Gronlund <markus.gronlund at oracle.com> wrote:
> 
> Greetings,
> 
> Please review this solution to address the issue described in the bug.
> 
> Bug: https://bugs.openjdk.java.net/browse/JDK-8203943 
> Webrev: http://cr.openjdk.java.net/~mgronlun/8203943/webrev00/ 
> 
> In addition, the JfrCheckpointThreadClosure now only needs a single iteration instead of the previous two.
> 
> Thanks
> Markus



More information about the hotspot-jfr-dev mailing list