RFR (S) 8243503: InstanceKlass::_array_name is not needed and leaks

coleen.phillimore at oracle.com coleen.phillimore at oracle.com
Fri Apr 24 19:36:32 UTC 2020


Thanks, Lois!
Coleen

On 4/24/20 3:21 PM, Lois Foltan wrote:
> On 4/24/2020 2:22 PM, coleen.phillimore at oracle.com wrote:
>>
>>
>> On 4/24/20 1:32 PM, Lois Foltan wrote:
>>> On 4/23/2020 7:07 PM, coleen.phillimore at oracle.com wrote:
>>>> Summary: decrement the refcount for all Klass in class unloading.
>>>>
>>>> See bug for more details.  Tested with tier1-6.
>>>>
>>>> open webrev at 
>>>> http://cr.openjdk.java.net/~coleenp/2020/8243503.01/webrev
>>>> bug link https://bugs.openjdk.java.net/browse/JDK-8243503
>>>>
>>>> Thanks,
>>>> Coleen
>>>
>>> Hi Coleen,
>>>
>>> Looks good.  Minor nit, can you keep part of the comment in 
>>> instanceKlass.cpp line #590, "which includes reference counting 
>>> symbol names"?  I find that helpful.
>>
>> Hi Lois,
>>
>> I restored the comment.  It made me notice that I had a bug. The 
>> Klass::release_C_heap_structures should be called by 
>> InstanceKlass::release_C_heap_structures_internal() so that the _name 
>> refcount is decremented for redefinition too.  I retested with the 
>> redefinition tests (which exercise this code) and runThese which does 
>> a lot of class unloading and will sanity check this with tier1.
>>
>> open webrev at 
>> http://cr.openjdk.java.net/~coleenp/2020/8243503.02/webrev
>> bug link https://bugs.openjdk.java.net/browse/JDK-8243503
>
> Looks good!
> Lois
>
>>
>> thanks,
>> Coleen
>>>
>>> Thanks,
>>> Lois
>>
>



More information about the hotspot-runtime-dev mailing list