[jira] [Comment Edited] (POOL-279) Thread concurrency issue in DefaultPooledObject.getIdleTimeMillis()

["Sebb (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/POOL-279?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14159512#comment-14159512 ] 

Sebb edited comment on POOL-279 at 10/9/14 1:07 AM:
----------------------------------------------------

Thanks, but it does not really explain why the time can be negative.

AFAICT the problem is that the calculation:

bq. return System.currentTimeMillis() - lastReturnTime;

will first fetch the current time, and then fetch lastReturnTime.
However this may have been updated in the meantime to a later time, hence the negative value.

-Also, as has been pointed out on the dev list, there is a multi-threading issue - the field is not safely published, so it needs to be made volatile- [The field is already volatile]


was (Author: sebb@apache.org):
Thanks, but it does not really explain why the time can be negative.

AFAICT the problem is that the calculation:

bq. return System.currentTimeMillis() - lastReturnTime;

will first fetch the current time, and then fetch lastReturnTime.
However this may have been updated in the meantime to a later time, hence the negative value.

Also, as has been pointed out on the dev list, there is a multi-threading issue - the field is not safely published, so it needs to be made volatile

> Thread concurrency issue in DefaultPooledObject.getIdleTimeMillis()
> -------------------------------------------------------------------
>
>                 Key: POOL-279
>                 URL: https://issues.apache.org/jira/browse/POOL-279
>             Project: Commons Pool
>          Issue Type: Bug
>    Affects Versions: 2.2
>            Reporter: Jacopo Cappellato
>            Priority: Minor
>         Attachments: POOL-279-unit-test.patch, POOL-279.patch, POOL-279.patch, POOL-279.patch, POOL-279.patch
>
>
> Under unlucky thread concurrency the getIdleTimeMillis() method of DefaultPooledObject can return a negative value.
> I have attached a Junit test that fails most of the times and a simple fix, that doesn't use synchronization: with this fix the Junit test always succeed.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)