RFR: Upstream merge to jdk-12+19

Aleksey Shipilev shade at redhat.com
Thu Nov 8 14:56:35 UTC 2018


On 11/08/2018 03:44 PM, Roman Kennke wrote:
> Lots of good stuff from upstream:
> 
> - 8212610: Fix handling of memory in PhaseIdealLoop::clone_loop_predicates()
> - 8212243: More gc interface tweaks for arraycopy
> - 8213384: Move G1/C2 barrier verification into G1BarrierSetC2
> - 8213381: Hook to allow GC to inject Node::Ideal() calls
> - 8213371: GC/C2 abstraction and cleanup to handle custom offset for GC
> memory accesses
> - 8213469: Remove/fix leftovers from JDK-8213384: Move G1/C2 barrier
> verification into G1BarrierSetC2

Good! OK.

> I needed to add this to shenandoahCodeRoots.cpp:
> diff --git a/src/hotspot/share/gc/shenandoah/shenandoahCodeRoots.cpp
> b/src/hotspot/share/gc/shenandoah/shenandoahCodeRoots.cpp
> --- a/src/hotspot/share/gc/shenandoah/shenandoahCodeRoots.cpp
> +++ b/src/hotspot/share/gc/shenandoah/shenandoahCodeRoots.cpp
> @@ -205,6 +205,7 @@
>        break;
>      }
>      case 2: {
> +      ResourceMark rm;
>        ShenandoahNMethodOopDetector detector;
>        nm->oops_do(&detector, /* allow_zombie = */ true);
> 
> 
> 
> Because of ShenandoahNMethodOopDetector allocating a GrowableArray as
> resource. Not sure why this started failing now, and not earlier and if
> needs backporting.

It probably needs backporting. Just RFR it separately for sh/jdk11, and we shall pick it up from there.

-Aleksey



More information about the shenandoah-dev mailing list