RFR: 8282704: runtime/Thread/StopAtExit.java may leak memory

[Daniel D.Daugherty]

On Thu, 17 Mar 2022 05:43:52 GMT, David Holmes <dholmes at openjdk.org> wrote:

>> Fix memory leak/pinning in runtime/Thread/StopAtExit.java stress test.
>> Change the test to:
>> - create and use its own ThreadGroup so it knows the expected thread count
>> - detect failures to cleanup the ThreadGroup
>> - these failures to cleanup can occur due to ThreadStop landing during Thread.exit()
>> 
>> The test uses reflection to call the package private ThreadGroup.threadTerminated()
>> or ThreadGroup.destroy() as needed. Sometimes the manual cleanup is not needed;
>> in a 30 second run, I typically see between 0-5 manual cleanups on my MBP13.
>
> Hi Dan,
> 
> This problem will go away when I fix:
> 
> https://bugs.openjdk.java.net/browse/JDK-8282952
> 
> Cheers,
> David

@dholmes-ora - I would still prefer to fix this test. I maintain my stress kit for
many releases so having an approved and fixed version of this test would be
good to have. Not just until you fix this bug in JDK19:

https://bugs.openjdk.java.net/browse/JDK-8282952

but for potential backport to earlier releases also. And for earlier versions
of my stress kit, the fixed test will be included in the stress kit itself until
(and if) the fix is backported. Thanks!

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

PR: https://git.openjdk.java.net/jdk/pull/7819