RFR: Handle single-threaded GC correctly

[Aleksey Shipilev]

On 09/27/2016 01:47 PM, Roman Kennke wrote:
> The problem that Aleksey reported comes from the fact that we
> initialize the ReferenceProcessor as single-threaded, and this
> repeatedly calls the same terminator. The fix makes ReferenceProcessor
> work multithreaded, even when only using a single thread. This may not
> be optimal, but it does the correct thing.
> 
> Ok to commit?
> http://cr.openjdk.java.net/~rkennke/1threadedgc/wevrev.00/

+1, do it.

This is the fourth time in last 24 hrs I trigger this assert :)

Thanks,
-Aleksey