[jira] [Commented] (OPENEJB-2070) Potential deadlock in URLClassLoaderFirst

["Romain Manni-Bucau (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/OPENEJB-2070?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13893481#comment-13893481 ] 

Romain Manni-Bucau commented on OPENEJB-2070:
---------------------------------------------

we can also synchronized it, avoid a variable and doesn't change that much things regarding perf  (we didnt design yet our classloaders to be multithreaded)

> Potential deadlock in URLClassLoaderFirst
> -----------------------------------------
>
>                 Key: OPENEJB-2070
>                 URL: https://issues.apache.org/jira/browse/OPENEJB-2070
>             Project: OpenEJB
>          Issue Type: Bug
>          Components: container system
>    Affects Versions: 4.6.0
>         Environment: NA
>            Reporter: Andy Gumbrecht
>             Fix For: 4.6.1
>
>
> Calling org.apache.openejb.util.classloader.URLClassLoaderFirst#loadClass from one thread may cause a deadlock if Class.forName is called from another thread, as this internally calls ClassLoader.getCallerClassLoader() - Which may be the same URLClassLoaderFirst instance.
> Adding an ReentrantLock should be a viable solution.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)