Heads-up: sh/jdk CTW failure with Traversal

[Roman Kennke]

> Testing looks good. Patch also looks good. Push it! :-)

I meant to say, file bug and RFR it properly, then push to jdk/jdk ;-)

Roman


> Roman
> 
> 
>> Fix is below. If one enqueue barrier feeds into another one, we only
>> need to keep one.
>>
>> Roland.
>>
>> diff -r 6277fcfd1269 
>> src/hotspot/share/gc/shenandoah/c2/shenandoahSupport.cpp
>> --- a/src/hotspot/share/gc/shenandoah/c2/shenandoahSupport.cpp    Mon 
>> Mar 25 19:57:08 2019 +0100
>> +++ b/src/hotspot/share/gc/shenandoah/c2/shenandoahSupport.cpp    Wed 
>> Mar 27 15:25:24 2019 +0100
>> @@ -2101,6 +2101,7 @@
>>   int ShenandoahEnqueueBarrierNode::needed(Node* n) {
>>     if (n == NULL ||
>>         n->is_Allocate() ||
>> +      n->Opcode() == Op_ShenandoahEnqueueBarrier ||
>>         n->bottom_type() == TypePtr::NULL_PTR ||
>>         (n->bottom_type()->make_oopptr() != NULL && 
>> n->bottom_type()->make_oopptr()->const_oop() != NULL)) {
>>       return NotNeeded;
>>