RFR: Concurrent uncommit should be recorded as GC event + perf

[Zhengyu Gu]

Good!

-Zhengyu

On 07/06/2018 01:59 PM, Aleksey Shipilev wrote:
> On 07/06/2018 07:20 PM, Aleksey Shipilev wrote:
>> On 07/06/2018 07:17 PM, Zhengyu Gu wrote:
>>> Is this right?
>>>
>>>       ShenandoahHeapRegion* r = get_region(i);
>>>       if (r->is_empty_committed() && (r->empty_time() < shrink_before)) {
>>> +      ShenandoahHeapLocker locker(lock());
>>> +      if (r->is_empty_uncommitted()) {
>>>         r->make_uncommitted();
>>>         count++;
>>>
>>> should it be?
>>>
>>> +      if (r->is_empty_committed()) {
>>>         r->make_uncommitted();
>>>         count++;
>>>
>>
>> Damn. Good spot! I tested the previous incarnation of the patch, which does what you suggested.
>> Re-testing again...
> 
> Retested. tier3_gc_shenandoah is fine, adhoc tests are fine.
> 
> -Aleksey
> 
>