Closed ChengJin01 closed 1 year ago
All the elements in the above stack trace are implemented in the extension repo (OpenJDK). The test fails because msysc
fails on AIX:
@backwaterred Are there plans to backport the above fix to JDK21?
I'm happy to backport this fix when it's merged :-)
I'm happy to backport this fix when it's merged :-)
Thanks, let us know once the backport is delivered. We will use it to close this issue.
@backwaterred any update on this?
Yeah, sorry. The backport merged in the last week or so. These changes are now in the headstream for jdk 22, & 21.
Thanks @backwaterred. Closing as per https://github.com/eclipse-openj9/openj9/issues/17870#issuecomment-1716185910.
We should keep this open until we have the fix. Not sure if it will be in 21.0.1, we might be waiting for 21.0.2. Or we can bring it manually for now.
Added the exclude label since the test is currently excluded.
Not sure if it will be in 20.0.1, we might be waiting for 20.0.2. Or we can bring it manually for now.
For Java 21? Since Java 20 is no longer supported.
Yes, I meant 21.0.1 and 21.0.2.
@backwaterred https://github.com/openjdk/jdk/pull/14963 seems RI specific. Will this fix also apply to OpenJ9?
Ah, I see you have linked part 1 of the fix above, which is Hotspot specific.
The main part of the issue is here: https://github.com/openjdk/jdk/pull/14964
@backwaterred Has https://github.com/openjdk/jdk/pull/14964 also been merged into the headstream for jdk 22, & 21?
Yes it has. Both the PR against 22 and the backport to 21 are merged.
We should keep this open until we have the fix. Not sure if it will be in 21.0.1, we might be waiting for 21.0.2. Or we can bring it manually for now.
The fix already exists in JDK-next: https://github.com/ibmruntimes/openj9-openjdk-jdk/commit/13fa40bdfeb6083a1114fbce24abc51e1ea8282f.
We should manually bring the fix to JDK21. It will allow us to enable the test and identify other OpenJ9 specific issues (if any) with the test earlier. @pshipton https://github.com/ibmruntimes/openj9-openjdk-jdk21/pull/44 backports the fix to JDK21.
@babsingh I think the test can be unexcluded now and this closed?
yes, i was looking for a nightly build to run a grinder. but, i didn't find a build with the fix yet. i will run a grinder from a personal build to verify the fix and re-enable the test.
The latest nightly jdk21 build contains the fix. We have builds for all platforms but AIX. In the grinder you can use the default, which is the latest.
Right, duh, AIX is the platform we need. I'll re-run the build. https://openj9-jenkins.osuosl.org/job/Pipeline-Build-Test-JDK21/114/
https://openj9-jenkins.osuosl.org/job/Pipeline-Build-Test-JDK21/114/
Thanks, the build successfully completed. Started a grinder to verify the fix: https://openj9-jenkins.osuosl.org/job/Grinder/2929.
TestByteBuffer
passes with JDK21 on AIX.
There is a subtest failure in TestByteBuffer.java at https://openj9-jenkins.osuosl.org/job/Grinder/2619/consoleText
which is most likely related to
ScopedMemoryAccess
.Note: the failure was only detected on AIX for the moment.
FYI: @babsingh, @tajila, @pshipton