RFR: 8079275: Remove CollectedHeap::use_parallel_gc_threads

Stefan Karlsson stefan.karlsson at oracle.com
Mon May 4 18:33:56 UTC 2015


On 2015-05-04 19:34, Kim Barrett wrote:
> On May 4, 2015, at 11:52 AM, Stefan Karlsson <stefan.karlsson at oracle.com> wrote:
>> Hi,
>>
>> On 2015-05-04 17:28, Stefan Johansson wrote:
>>> Hi Stefan,
>>>
>>> On 2015-05-04 15:34, Stefan Karlsson wrote:
>>>> Hi,
>>>>
>>>> Please review this patch to remove CollectedHeap::use_parallel_gc_threads.
>>>>
>>>> http://cr.openjdk.java.net/~stefank/8079275/webrev.01/
>>> Change looks good, nice cleanup.
>>>
>>> If I'm not missing anything you could remove the typedef in concurrentMarkSweepGeneration.cpp:
>>>      typedef CMSParGCThreadState* CMSParGCThreadStatePtr;
>>> -    _par_gc_thread_states =
>>> -      NEW_C_HEAP_ARRAY(CMSParGCThreadStatePtr, ParallelGCThreads, mtGC);
>>> +  _par_gc_thread_states = NEW_C_HEAP_ARRAY(CMSParGCThreadStatePtr, ParallelGCThreads, mtGC);
>>>
>>> The typedef is only used for in the NEW_C_HEAP_ARRAY-call and I can't see any good reason for keeping it.
>> OK. I removed it. I also removed the unused _modUnionClosure.
>>
>> http://cr.openjdk.java.net/~stefank/8079275/webrev.02.delta/
>> http://cr.openjdk.java.net/~stefank/8079275/webrev.02/
> Looks good.

Thanks.

>
> However, I think the class ModUnionClosure is no longer used directly, only as a base class for ModUnionClosurePar and as a type in signatures that always get passed the latter.  If that’s correct, then some class hierarchy cleanup can be done, perhaps with some naming cleanup too.  I’ll keep looking at new versions of this if you want to continue the cleanup here, or you can defer such cleanup to a new CR.  Or you can tell me I missed something.

It's used in CMSParRemarkTask::do_dirty_card_rescan_tasks and 
CMSCollector::do_remark_non_parallel.

StefanK

>




More information about the hotspot-gc-dev mailing list