harmony-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Pavel Rebriy (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HARMONY-4389) [drlvm][jvmti] Thread.isAlive() is false in THREAD_STRAT event callback
Date Fri, 14 Sep 2007 09:54:32 GMT

    [ https://issues.apache.org/jira/browse/HARMONY-4389?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12527430

Pavel Rebriy commented on HARMONY-4389:


Thread.isAlive field is changed in synchronized block, that is your patch breaks access serializability.

> [drlvm][jvmti] Thread.isAlive() is false in THREAD_STRAT event callback
> -----------------------------------------------------------------------
>                 Key: HARMONY-4389
>                 URL: https://issues.apache.org/jira/browse/HARMONY-4389
>             Project: Harmony
>          Issue Type: Bug
>          Components: DRLVM
>         Environment: Windows/x86, harmony-jdk-r553376
>            Reporter: Ivan Popov
>         Attachments: 0001-Fix-of-isAlive-during-JVM-TI-THREAD_START-event-proc.patch,
> After applying patch HARMONY-2889 for JDWP agent the following JDWP stress tests:
>     org.apache.harmony.test.stress.jpda.jdwp.scenario.EVENT014.EventTest015
>     org.apache.harmony.test.stress.jpda.jdwp.scenario.EVENT014.EventTest016
> still hangs on Harmony and pass on RI. The problem is in different behavior of j.l.Thread.isAlive()
while thread is suspended in THREAD_START callback. In RI isAlive()==true, in Harmony isAlive()==false.
> In these tests debuggee application starts tested thread, which generates THREAD_START
events, and waits until isAlive()==true for this thread. Then it sends message to debugger
and debugger handles THREAD_START events and resumes tested thread. With RI this works fine,
but with Harmony isAlive() never returns true until thread is resumed in THREAD_START event,
and test hangs.
> I doubt if there is any requirement in classlib or JVMTI specifications which would clarify
this situation. However, it makes sense to follow RI's behavior. Otherwise, affected tests
should be modified. 

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message