RFR: 8251451: Shenandoah: Remark ObjectSynchronizer roots with I-U

Zhengyu Gu zgu at redhat.com
Wed Aug 12 13:52:16 UTC 2020



On 8/12/20 9:45 AM, Roman Kennke wrote:
> 
>>
>> On 8/12/20 9:02 AM, Roman Kennke wrote:
>>> On Wed, 2020-08-12 at 08:47 -0400, Zhengyu Gu wrote:
>>>> Hmmm... seems that it duplicated thread remarking work also ...
>>>
>>> Why? Can you explain?
>>
>> http://hg.openjdk.java.net/jdk/jdk/file/e7109ed4bbb0/src/hotspot/share/gc/shenandoah/shenandoahConcurrentMark.cpp#l297
>>
>> or L#307 executed by all workers.
> 
> Ah yes. But:
> - It was pre-existing and has not been changed by this patch.
> - The actual thread-claiming is done in the closure.
Ah, right.

-Zhengyu



> 
> Should be good?
> 
> Roman
> 
> 
>> -Zhengyu
>>
>>
>>> Thanks,
>>> Roman
>>>
>>>
>>>> -Zhengyu
>>>>
>>>> On 8/12/20 6:42 AM, Aleksey Shipilev wrote:
>>>>> On 8/12/20 12:36 PM, Roman Kennke wrote:
>>>>>> http://cr.openjdk.java.net/~rkennke/JDK-8251451/webrev.01/
>>>>>
>>>>> Yes, that's better. The indenting is off, though: lots of space
>>>>> before "_claimed", and "public" is
>>>>> moved. Should be:
>>>>>
>>>>> private:
>>>>>      ShenandoahConcurrentMark* _cm;
>>>>>      TaskTerminator*           _terminator;
>>>>>      bool                      _dedup_string;
>>>>>      ShenandoahSharedFlag      _claimed_syncroots;
>>>>>
>>>>> public:
>>>>>      ShenandoahFinalMarkingTask(ShenandoahConcurrentMark* cm,
>>>>> TaskTerminator* terminator, bool
>>>>> dedup_string) :
>>>>>
>>>>> No need for another review.
>>>>>
> 



More information about the shenandoah-dev mailing list