[ https://issues.apache.org/jira/browse/HARMONY-3861?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12495957
]
Gregory Shimansky commented on HARMONY-3861:
--------------------------------------------
Running the test 1000 times in a loop with -Xem:jet (in fact it appears that -Xem:jet is equivalent
to running JVMTI tests in default mode since ti.emconf doesn't contain any differences from
jet.emconf) produced no crash. Command line that I used is the following:
./lnx_ia32_gcc_debug/deploy/jdk/jre/bin/java -Xem:jet -cp ./lnx_ia32_gcc_debug/semis/jvmti.tests/classes:./make/tmp/junit.jar
-agentpath:./lnx_ia32_gcc_debug/semis/jvmti.tests/native/PopFrame1/libPopFrame1.so junit.textui.TestRunner
PopFrame1.PopFrame1
I wonder if some difference in CC/CI environment causes the crash. Is it possible to enable
core dumps when running CC/CI (using "ulimit -c unlimited") and use core dump to at least
get stack traces in the threads when crash happened?
> [drlvm] DRLVM jvmti test PopFrame1 intermittently failed on Linux x86
> ---------------------------------------------------------------------
>
> Key: HARMONY-3861
> URL: https://issues.apache.org/jira/browse/HARMONY-3861
> Project: Harmony
> Issue Type: Bug
> Components: DRLVM
> Reporter: Vladimir Ivanov
>
> The CC/CI report issue with PopFrame1 DRLVM test.
> Note this is intermittent failure and it hard to reproduce.
> ---------------- Execution log ---------------------------
> ==================================
> Run JVMTI tests using jitrino.JET
> ==================================
>
> Running Breakpoint1
> Running PopFrame1
> SIGTRAP in VM code.
> Stack trace:
> <end of stack trace>
> Running SingleStep1
> Running VMInit1
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
|