You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@commons.apache.org by "Phil Steitz (JIRA)" <ji...@apache.org> on 2014/12/31 17:28:14 UTC

[jira] [Closed] (POOL-202) GenericKeyedObjectPool.close() can interfere with evict(); eviction[Key]Iterator is not consistently synched.

     [ https://issues.apache.org/jira/browse/POOL-202?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Phil Steitz closed POOL-202.
----------------------------

> GenericKeyedObjectPool.close() can interfere with evict(); eviction[Key]Iterator is not consistently synched.
> -------------------------------------------------------------------------------------------------------------
>
>                 Key: POOL-202
>                 URL: https://issues.apache.org/jira/browse/POOL-202
>             Project: Commons Pool
>          Issue Type: Bug
>    Affects Versions: 2.0
>            Reporter: Sebb
>             Fix For: 2.0
>
>
> As far as I can tell, calling close during an evict() run can cause problems, because close() sets the eviction[Key]Iterator fields to null.
> I suspect close() and evict() should be mutually exclusive.
> Maybe close does not need to nullify the eviction[Key]Iterator fields, but there's probably other mutually incompatible behaviour.
> Or maybe the fields should be nulled in startEvictor.
> Also, the fields eviction[Key]Iterator aren't volatile and aren't always accessed under the same lock so might not be safely published.
> That would be solved by moving the nulling to startEvictor.



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