RFR [10]: Revert G1 changes

Aleksey Shipilev shade at redhat.com
Tue Jul 11 09:35:24 UTC 2017


On 07/11/2017 11:33 AM, Roman Kennke wrote:
> Am 11.07.2017 um 11:30 schrieb Aleksey Shipilev:
>> Hi,
>>
>> As discussed last week, we want to minimize our exposure in G1-specific code.
>> This reverts G1 to mainline version in shenandoah-jdk10:
>>    http://cr.openjdk.java.net/~shade/shenandoah/revert-g1-jdk10/webrev.01/
>>
>> Testing: hotspot_gc_shenandoah
>>
>> Thanks,
>> -Aleksey
>>
>>
> The change is good.
> 
> I am almost surprised that you don't get linker conflicts... different
> definitions of CMBitMap, SATBMarkQueue, etc in two .cpp files?

CMBitMap vs. *G1*CMBitMap yields no conflicts.

SATBMarkQueue is not split, we still maintain requires_marking and
filter<HeapType> there, which seems innocuous during merges.

-Aleksey






More information about the shenandoah-dev mailing list