jsr292Test_JitCount0 memoryCorruptionDetected

[pshipton]

https://openj9-jenkins.osuosl.org/job/Test_openjdk21_j9_extended.functional_s390x_linux_Nightly_testList_1/358 - rh7-390-2
jsr292Test_JitCount0_0

https://openj9-artifactory.osuosl.org/artifactory/ci-openj9/Test/Test_openjdk21_j9_extended.functional_s390x_linux_Nightly_testList_1/358/functional_test_output.tar.gz

19:24:01  testIdentity.1
19:24:01  testIdentity.2
19:24:01  testIdentity.3
19:26:01  00:25:57.422 0x1763800 omrport.359    *   ** ASSERTION FAILED ** at /home/jenkins/workspace/Build_JDK21_s390x_linux_Nightly/omr/port/common/omrmemtag.c:145: ((memoryCorruptionDetected))

[pshipton]

@hzongaro fyi

[github-actions]

Issue Number: 20617
Status: Open
Recommended Components: comp:gc, comp:vm, comp:test

[hzongaro]

I wasn't able to do anything useful with the core file that was produced in gdb. I was able to reproduce the crash in a grinder run, but still couldn't manage to debug in gdb.

(gdb) where
#0  0x000003ffafc8d484 in ?? ()
Backtrace stopped: frame did not save the PC

I spent some time looking at the core file in jdmpview, but nothing obvious jumped out at me.

If failures show up frequently, I'll try to capture further information in a grinder run.

[pshipton]

I assume you are either running gdb on the machine that produced the core or running jpackcore and capture the libraries to use on another machine.

It's not obvious if you know how to get the corrupted address, fyi if you don't.

!omrportlibraryglobaldata 0x000003FFA800C310
OMRPortLibraryGlobalData at 0x3ffa800c310 {
  Fields for OMRPortLibraryGlobalData:
        0x0: void* corruptedMemoryBlock = !j9x 0x000003FFA820A410

[hzongaro]

Thanks, @pshipton! I had missed your comments.

I assume you are either running gdb on the machine that produced the core or running jpackcore and capture the libraries to use on another machine.

I was running gdb on a different machine, but I had downloaded the jdk and debug images, and pointed gdb to the directories containing them. That usually works for me, but if there are reasons why jpackcore would be more likely to be successful, I'll try using it in future.

It's not obvious if you know how to get the corrupted address, fyi if you don't.

Thanks for that tip on getting a corrupted address! I did not know about it.

[pshipton]

I was running gdb on a different machine

That only works if it's the same OS and versions of all the system libraries. Otherwise you need jpackcore to collect the libraries from the crashing machine.