IcedTea-web: Memory leak - java side

[Efrat Lefeber]

My problems are only with the java side. What I see are two major leaks:
1. PluginAppletViewer
2. ApplicationShutdownHook in JNLPClassLoader

Are there tests that run applets to the plugin in a loop on the same jvm? I get OutOfMemory exceptions very fast.

-----Original Message-----
From: Jiri Vanek [mailto:jvanek at redhat.com]
Sent: Thursday, January 16, 2014 3:32 PM
To: Efrat Lefeber; Deepak Bhole
Cc: distro-pkg-dev at openjdk.java.net; Andrew Azores; Adam Domurad
Subject: Re: IcedTea-web: Memory leak - java side

On 01/16/2014 10:16 AM, Efrat Lefeber wrote:
> Do you know why it was reverted? When will a new patch be released? I am working on a project and it is very urgent for me.
>
>
> -----Original Message-----
> From: Deepak Bhole [mailto:dbhole at redhat.com]
> Sent: Wednesday, January 15, 2014 11:07 PM
> To: Efrat Lefeber
> Cc: distro-pkg-dev at openjdk.java.net; jiri Vanek; Andrew Azores
> Subject: Re: IcedTea-web: Memory leak - java side
>
> * Efrat Lefeber <ELefeber at trustwave.com> [2014-01-15 10:27]:
>> I am using icedtea-6-plugin with firefox on linux debian (wheezy).
>>
>> I noticed that after opening applets in firefox tabs, and closing
>> them, some objects in java plugin are not released, which causes the
>> heap size of the jvm to keep increasing.
>>
>> For example, I see that the PluginAppletViewer object is not
>> released, also the applet object itself and each classloader of each applet.
>>
>> If I open and close a webpage containing a java applet, the jvm
>> eventually will get out of memory exception.
>>
>> Looking at the plugin code, I see that when closing the applet
>> window, the function PluginAppletViewer.destroyApplet is called, but
>> in this flow, not all objects are release.
>>
>>
>>
>> Is there a way to free the memory without taking down the jvm to
>> prevent the increase in memory?
>>
>
> I know some objects are kept around to make subsequent startups faster.
> But PluginAppletViewer does not sound like something that belongs on that list.
>
> Adam (Domurad) had proposed a patch to clean memory as much as possible but I think at one point it had to be reverted.

Did he?

He fixed many, really *many* leaks in C part. All are in, and none was ever reverted.
Here we are about java side. AFAIK the leaks here were never tried to be hunted.

Also freeing of PluginAppletViewer .... I doubt someone have ever tried...


I have cc'ed him but I do not wont to bother him during studies. So unless I really missed something...
>
> Jiri, Andrew, do you know the status of this?

J.



________________________________

This transmission may contain information that is privileged, confidential, and/or exempt from disclosure under applicable law. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution, or use of the information contained herein (including any reliance thereon) is strictly prohibited. If you received this transmission in error, please immediately contact the sender and destroy the material in its entirety, whether in electronic or hard copy format.