From lukas.stampf at bat.at  Thu Jul  2 18:00:48 2020
From: lukas.stampf at bat.at (STAMPF Lukas)
Date: Thu, 2 Jul 2020 18:00:48 +0000
Subject: JDK-8229011: Hints
Message-ID: <VI1PR0802MB2446D9B6CFC1CD0AE2A18D41E56D0@VI1PR0802MB2446.eurprd08.prod.outlook.com>

Hello,

I hope I don't bother anyone with this.
Out of curiosity I  looked into this and it seems that the LambdaForms, MethodHandles, etc created by the DynamicLinker are nerver garbage collected. Before
https://bugs.openjdk.java.net/browse/JDK-8141538
this wasn't so bad, because there was always just one static DynamicLinker, therefore the amount of leaking objects was limited. With this change, there is one DynamicLinker per Context(so on per ScriptEngine) creating a memory leak.
This can be shown by adding
            if (linker == null)
            {
                linker = getLinker(scriptEngine);
            }
            setLinker(scriptEngine, linker);
to the loop in the testcript linked in Jira. Set/getLinker are simply using reflection to get/set the final field of jdk.nashorn.internal.runtime.Context dynamicLinker.


The question I can't answer yet is why the Objects created by the DynamicLinker are not garbage collected. JProfiler and Eclipse MAT seem to be unable to analyze these dynamic objects properly. Most of the objects are marked as unreachable but not collected.
I would be grateful for every input on why it could be this way or what I could try to  find out whats keeping these unreachable objects alive.

Kind regards,
Lukas

From sundararajan.athijegannathan at oracle.com  Wed Jul 15 15:58:14 2020
From: sundararajan.athijegannathan at oracle.com (sundararajan.athijegannathan at oracle.com)
Date: Wed, 15 Jul 2020 21:28:14 +0530
Subject: RFR 8246113: re-visit javax/script test that still requires
 jdk.scripting.nashorn module
Message-ID: <49174970-2ccd-af7a-8af4-16ef8343ce48@oracle.com>

Hi,

Please review - removing these tests.

Bug: https://bugs.openjdk.java.net/browse/JDK-8246113

Webrev: http://cr.openjdk.java.net/~sundar/8246113/webrev.00/

Thanks

-Sundar


From james.laskey at oracle.com  Wed Jul 15 16:04:19 2020
From: james.laskey at oracle.com (Jim Laskey)
Date: Wed, 15 Jul 2020 13:04:19 -0300
Subject: RFR 8246113: re-visit javax/script test that still requires
 jdk.scripting.nashorn module
In-Reply-To: <49174970-2ccd-af7a-8af4-16ef8343ce48@oracle.com>
References: <49174970-2ccd-af7a-8af4-16ef8343ce48@oracle.com>
Message-ID: <46FDFA20-8135-4B4C-A208-499B1A1097FA@oracle.com>

+1

> On Jul 15, 2020, at 12:58 PM, Sundararajan.athijegannathan at oracle.com wrote:
> 
> Hi,
> 
> Please review - removing these tests.
> 
> Bug: https://bugs.openjdk.java.net/browse/JDK-8246113
> 
> Webrev: http://cr.openjdk.java.net/~sundar/8246113/webrev.00/
> 
> Thanks
> 
> -Sundar
>