RFR [11], 2018-07-30, Bulk backports to sh/jdk11

[Roman Kennke]

Am 30.07.2018 um 19:34 schrieb Aleksey Shipilev:
> http://cr.openjdk.java.net/~shade/shenandoah/backports/jdk11-20180730/webrev.01/
> 
> This backports latest work to sh/jdk11:
> 
>  [backport] Skipping null strings when popping dedup string candidates
>  [backport] Trace and report total allocation latency and sizes
>  [backport] -XX:-UseTLAB should disable GCLABs too
>  [backport] Exponential thread-local GCLAB sizing
>  [backport] Elastic TLABs may avoid peeking into the free set racily
>  [backport] Refactoring ShenandoahStrDedupStress test to reduce test time
>  [backport] Optimize traversal mark: constify, avoid polling traversal set and is_marked_next
>  [backport] Traversal should resize TLABs
>  [backport] Refactor to group marking bitmap and TAMS structure in one class ShenandoahMarkingContext
>  [backport] Refactor alive-closures to deal better with new marking contexts
>  [backport] Avoid indirection to next-mark-context
>  [backport] TLAB sizing policy should converge faster with Shenandoah
>  [backport] Enable worker scaling for ref-proc
>  [backport] Move periodic GC decision making to GC heuristics base class
>  [backport] Fix single-threaded reference processing
>  [backport] Improve C2 barrier filtering: heap-stable check should test for HAS_FORWARDED only
>  [backport] Fix task queue assertion with traversal GC
>  [backport] Reset TAMS for uncommitted empty regions to avoid leftover old values
>  [backport] Handle missing ShenandoahWriteBarrierRB case
>  [backport] Move Shenandoah stress tests to tier3
> 
> Testing: tier3_gc_shenandoah {fastdebug|release}
> 
> Thanks,
> -Aleksey
> 

Good! Thanks!
Roman