Integrated: 8309296: jdk/jfr/event/runtime/TestAgentEvent.java fails due to "missing" dynamic JavaAgent

Erik Gahlin egahlin at openjdk.org
Wed Jun 21 11:50:10 UTC 2023


On Tue, 20 Jun 2023 23:38:04 GMT, Erik Gahlin <egahlin at openjdk.org> wrote:

> Could I have a review of a test fix.
> 
> Problem is that events use another clock source than Recording::getStartTime(). 
> 
> I added -XX:+UnlockExperimentalVMOptions -XX:-UseFastUnorderedTimeStamps in case the Attach thread and the periodic task thread executes on different cores and there is drift. The flags may, or may not, also fix the time source issue. Regardless, I think it's prudent to do time comparison against an event, not the recording timestamps, because they are different things conceptually and there are tweaks to the recording timestamps (to make them work well with Instant.now(), not the event clock source)
> 
> Testing: 100 times jdk/jdk/jfr/event/runtime/TestAgentEvent.java on Mac, Windows and Linux.
> 
> Thanks
> Erik

This pull request has now been integrated.

Changeset: 658c3374
Author:    Erik Gahlin <egahlin at openjdk.org>
URL:       https://git.openjdk.org/jdk/commit/658c3374d88836f5c8745f777e836d502f0ec2ca
Stats:     38 lines in 1 file changed: 21 ins; 0 del; 17 mod

8309296: jdk/jfr/event/runtime/TestAgentEvent.java fails due to "missing" dynamic JavaAgent

Reviewed-by: mgronlun

-------------

PR: https://git.openjdk.org/jdk/pull/14572


More information about the hotspot-jfr-dev mailing list