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-5484) [drlvm] Handles leakage in StackTraceTest
Date Wed, 27 Feb 2008 16:37:51 GMT

    [ https://issues.apache.org/jira/browse/HARMONY-5484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12572963#action_12572963

Pavel Rebriy commented on HARMONY-5484:

I've recheck the test on release and found handle leak as was reported. Working with the reliability
test I've created a small test "Handle.java" to investigate a handle leaks and to reproduce
the problem.

The result a the following:

1. $ java Handle

In this case test uses about 6000 handles by creating a million threads.

2. $ java Handle 10

In this case test uses about 200000 handles by creating a million threads.

3. $ java Handle local

The same result as in the 2nd case - 200000 handles on million threads.

It looks like the problem in GC working with thread. Somehow thread blocks don't reuse and
old handles doesn't released if we store reference to thread on array or many local variables.

> [drlvm] Handles leakage in StackTraceTest
> -----------------------------------------
>                 Key: HARMONY-5484
>                 URL: https://issues.apache.org/jira/browse/HARMONY-5484
>             Project: Harmony
>          Issue Type: Bug
>          Components: DRLVM
>         Environment: Windows x86, could be on Linux too
> Revision 617144
>            Reporter: Andrey Yakushev
>         Attachments: exclude_StackTraceTest.patch, Handle.java
>  api.kernel.thread.StackTraceTest.StackTraceTest from the reliability suite have handle
leakage ~40000 for 20 minutes run on Windows x86. It causes hang in CC test running.

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

View raw message