RFR [11], 2018-07-16, Bulk backports to sh/jdk11

[Roman Kennke]

Am 16.07.2018 um 23:39 schrieb Aleksey Shipilev:
> http://cr.openjdk.java.net/~shade/shenandoah/backports/jdk11-20180716/webrev.02/
> 
> This backports latest work to sh/jdk11:
> 
>  [backport] Several C1 cleanups
>  [backport] Exponential backoff with pacing
>  [backport] More detailed pacing histogram
>  [backport] Proper units for allocation failure messages
>  [backport] Mark-compact adjust pointers can be specialized with UseShenandoahMatrix
>  [backport] Add task termination and enhanced task queue state tracking + weakrefs
>  [backport] Report actual free size in non-verbose FreeSet status
>  [backport] Pacer for evacuation should print "Avail" to capture discounting
>  [backport] Refactor allocation path to accept ShenandoahAllocRequest tuple
>  [backport] Elastic TLABs support for Shenandoah
>  [backport] Pacer should account actual size for elastic TLABs
>  [backport] Heap region count selection should only consider max heap size
>  [backport] Fix CAS-obj predicates and add expected-null-versions for cmpxchg-narrow-oop
>  [backport] Micro-optimize traversal GC alloc check
>  [backport] Hook up GCLABs to Elastic LAB support
>  [backport] Allocation tracker should really report bytes
>  [backport] Complete liveness for recently allocated regions outside the allocation path
>  [backport] GCLAB slowpath allocations should fit the object into GCLAB
>  [backport] fix aarch64 membar elision with shenandoah
>  [backport] Only include regular and humongous regions in traversal set
>  [backport] Print task queue statistics at the end of GC cycle
>  [backport] Fix TestSelectiveBarrierFlags
>  [backport] More efficient next bitmap resetting for traversal
> 
> Testing: tier3_gc_shenandoah {fastdebug|release}
> 
> Thanks,
> -Aleksey
> 

Looks good. Thanks!
Roman