RFR: 8213890: Implementation of JEP 344: Abortable Mixed Collections for G1

Stefan Johansson stefan.johansson at oracle.com
Wed Nov 28 09:30:59 UTC 2018



On 2018-11-27 22:31, Thomas Schatzl wrote:
> Hi,
> 
> On Tue, 2018-11-27 at 15:41 -0500, Kim Barrett wrote:
>>> On Nov 27, 2018, at 3:32 PM, Thomas Schatzl <
>>> thomas.schatzl at oracle.com> wrote:
>>>
>>> Hi,
>>>
>>> On Tue, 2018-11-27 at 15:19 -0500, Kim Barrett wrote:
>>>>> On Nov 27, 2018, at 2:36 PM, Stefan Johansson <
>>>>> stefan.johansson at oracle.com> wrote:
>>>>>
>>>>> Thanks for the review Kim,
>>>>>
>>>>> Two new webrevs, I'll let you decide which way to go, I kind of
>>>>> prefer version b.
>>>>> Full a: http://cr.openjdk.java.net/~sjohanss/8213890/03a/
>>>>> Inc a:  http://cr.openjdk.java.net/~sjohanss/8213890/02-03a/
>>>>> Full b: http://cr.openjdk.java.net/~sjohanss/8213890/03b/
>>>>> Inc b:  http://cr.openjdk.java.net/~sjohanss/8213890/02-03b/
>>>>
>>>> Version b looks good to me.
>>>> (I skimmed version a, but I also preferred b.)
>>>>
>>>
>>>   - in G1OopStarChunkedList::push, the increase of _used_memory can
>>> be moved out of the if- and else-blocks. :)
>>
>> I don’t think so.  _used_memory shouldn’t be increased for the
>> (implicit) case of having an existing list that isn’t full.
> 
> Okay, my fault. Looks good as is.
> 
> Thanks,
>    Thomas
> 

Thanks for the review guys, let's go with version b then.

I'll continue running perf, functional and stress testing on this until 
the JEP gets targeted.

Thanks,
Stefan



More information about the hotspot-gc-dev mailing list