RFR: 8274563: jfr/event/oldobject/TestClassLoaderLeak.java fails when GC cycles are not happening

[Erik Gahlin]

On Thu, 30 Sep 2021 11:35:52 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

> This reliably fails with Serial (ignored by @requires), Shenandoah and Z:
> 
> 
> $ CONF=linux-x86_64-server-fastdebug make run-test TEST=jdk/jfr/event/oldobject/TestClassLoaderLeak.java TEST_VM_OPTS="-XX:+UseShenandoahGC"
> 
> ...
> 
> STDERR:
> java.lang.RuntimeException: Could not find class leak
> at jdk.test.lib.Asserts.fail(Asserts.java:594)
> at jdk.jfr.event.oldobject.TestClassLoaderLeak.main(TestClassLoaderLeak.java:80)
> at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
> at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:77)
> at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> at java.base/java.lang.reflect.Method.invoke(Method.java:568)
> at com.sun.javatest.regtest.agent.MainWrapper$MainThread.run(MainWrapper.java:127)
> at java.base/java.lang.Thread.run(Thread.java:833)
> 
> 
> I believe this test implicitly wants at least one GC cycle to happen so that `ObjectSampler::oop_storage_gc_notification` turns the sampled objects "old". An easy way to do this is to trim the Java heap size. If that does not work, we would need to go to MXBeans and watch GC cycles happening. 
> 
> Additional testing:
>  - [x] Affected test with `-XX:+UseShenandoahGC`, 10 times
>  - [x] Affected test with `-XX:+UseZGC`, 10 times
>  - [x] Affected test with `-XX:+UseSerialGC`, 10 times
>  - [x] Affected test with `-XX:+UseParallelGC`, 10 times
>  - [x] Affected test with `-XX:+UseG1GC`, 10 times

Marked as reviewed by egahlin (Reviewer).

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

PR: https://git.openjdk.java.net/jdk/pull/5775