JDK-8247965 : Two JFR tests failing in Loom repo
  • Type: Bug
  • Component: hotspot
  • Sub-Component: jfr
  • Affected Version: repo-loom
  • Priority: P4
  • Status: Resolved
  • Resolution: Not an Issue
  • Submitted: 2020-06-20
  • Updated: 2022-04-28
  • Resolved: 2022-04-28
The Version table provides details related to the release that this issue/RFE will be addressed.

Unresolved : Release in which this issue/RFE will be addressed.
Resolved: Release in which this issue/RFE has been resolved.
Fixed : Release in which this issue/RFE has been fixed. The release containing this fix may be available for download as an Early Access Release or a General Availability Release.

To download the current JDK release, click here.
Other
repo-loomResolved
Related Reports
Relates :  
JDK-8248485 - Poor scalability in JfrCheckpointManager when using many threads after JDK-8242088
Description
test/jdk/jdk/jfr/event/runtime/TestClassLoaderStatsEvent.java has been failing for some time in the loom repo with:
java.lang.RuntimeException: Value not equal to 4, field='hiddenClassCount', value='0': expected 0 to equal 4

test/jdk/jdk/jfr/threading/TestManyVirtualThreads.java has been failing with a timeout since the repo was sync'ed up to jdk-16+1 (from jdk-15+26).
Comments
Both tests have been running successfully for some time. Closing as Not an Issue.
28-04-2022
[~mgronlun] Is it still an issue or could bug be closed?
12-04-2022
test/jdk/jdk/jfr/event/runtime/TestClassLoaderStatsEvent.java has been failing for some time in the loom repo with: java.lang.RuntimeException: Value not equal to 4, field='hiddenClassCount', value='0': expected 0 to equal 4 Seems to be an effect of changes done as part of Hidden Classes JEP, https://openjdk.java.net/jeps/371, and triggers in the Loom repo. Appears sensitive to what GC is being used, if using -XX:+UseParallelGC in Loom repo, the test passes. On success, there are two ClassLoaderData with property mirrorHolder = true assigned to load a few anonymous / hidden classes, and they get attributed to the ClassLoaderStats used by JFR to report the statistics. On failure, there are no ClassLoaderData with property mirrorHolder = true that load any anonymous / hidden classes.
06-07-2020
JFR thread local checkpoints now in place: https://github.com/openjdk/loom/commit/6dbc3cbf2a52dc19fe0feff74bccc14175819af9
05-07-2020
[~alanb] Thanks Alan, I will complete the full solution for Loom once merged.
03-07-2020
I have an initial merge with jdk-16+4 (which includes JDK-8248485) and I can confirm that threading/TestManyVirtualThreads.java is passing.
03-07-2020
After JDK-8248485 is fixed upstream and synched back into the Loom repo, I will complement the fix by increasing the size of the thread local buffers and by adding a thread-local cache mechanism.
29-06-2020
TestManyVirtualThreds.java is affected by JDK-8248485.
29-06-2020