RFR: Pinning humongous regions should be allowed

[Zhengyu Gu]

Good to me.

-Zhengyu

On 08/30/2017 08:15 AM, Aleksey Shipilev wrote:
> This is the regression from region FSM refactoring: we should allow pinning humongous regions. The
> good thing is that we do not have to mark humongous regions as "Pinned" or do critical_pins ops,
> because they are not moving. Humongous objects acquired via GetPrimitiveArrayCritical are referenced
> via JNI handles, so pinning info is also not needed to avoid their reclamation.
> 
> Fix:
>    http://cr.openjdk.java.net/~shade/shenandoah/region-humongous-pin/webrev.01/
> 
> Testing: hotspot_gc_shenandoah (also new test that used to fail)
> 
> Thanks,
> -Aleksey
>