RFR: 8057037: Verification in ClassLoaderData::is_alive is too slow

Stefan Karlsson stefan.karlsson at oracle.com
Wed Apr 8 12:56:51 UTC 2015


Hi Volker,

On 2015-04-08 14:45, Volker Simonis wrote:
> Hi Stefan,
>
> have you considered downporting this to 8u-dev?

This should have been backported earlier, sorry.

> I would do it myself, but unfortunately it is in shared code so I
> would need a sponsor anyway.
>
> This hurts us badly especially on our slow porting platforms where we
> get GC-times of several minutes now.

The patch applies cleanly to:
http://hg.openjdk.java.net/jdk8u/hs-dev/hotspot/

so I'm going to push the fix to that repository.

Thanks,
StefanK
>
> Thank you and best regards,
> Volker
>
> On Tue, Sep 2, 2014 at 3:03 PM, Stefan Karlsson
> <stefan.karlsson at oracle.com> wrote:
>> Hi all,
>>
>> Please review this patch to remove some verification code that was added
>> during the G1 Class Unloading project. The code was added to make sure we
>> catch any class unloading bugs as early as possible. This verification code
>> is run too often and is causing large GC times in our fastdebug test runs.
>> In one of the tests I investigated, the Full GCs take 8 seconds with this
>> verification and 0.2 seconds without it. Most of the verified pointers are
>> already verified before and after GCs in ClassLoaderDataGraph::verify().
>>
>> http://cr.openjdk.java.net/~stefank/8057037/webrev.00/
>>
>> thanks,
>> StefanK



More information about the hotspot-gc-dev mailing list