RFR: Use PLAB for evacuations instead of TLAB

[Roman Kennke]

Am 09.05.2018 um 19:02 schrieb Roman Kennke:
> I propose to use PLAB for evacuations (write-barrier and GC workers). It
> has the following benefits:
> - It reverts/removes a *whole* lot of code to upstream state, thus
> significantly improving/reducing our upstream exposure
> - It is a much better fit for what we need (e.g. already has rollback
> functionality)
> - It avoids disturbing TLAB stats by GCLAB activity (which has much
> different characteristics)
> - I implemented it to not mix up mutator and collector GCLAB stats,
> which have very different characteristics (churn vs. very occasional allocs)
> 
> I hope I got the mechanics right. This might need some experimentation
> and tweaking afterwards, and would then be backportable.
> 
> http://cr.openjdk.java.net/~rkennke/plab-gclab/webrev.00/
> 
> Testing: hotspot_gc_shenandoah ok
> 

Need to move around some UseTLAB guards. The GCLABs are enabled
regardless of UseTLAB or not. Do we want to guard this with a separate flag?

Diff:
http://cr.openjdk.java.net/~rkennke/plab-gclab/webrev.01.diff/
Full:
http://cr.openjdk.java.net/~rkennke/plab-gclab/webrev.01/


Roman