RFR (S) 8244326: Shenandoah: global statistics should not accept bogus samples

[Aleksey Shipilev]

On 5/5/20 11:40 AM, Roman Kennke wrote:
>>> Why not declare uninitialized a const field?
>>
>> Wanted to maintain the symmetry against ShenandoahWorkerData::uninitialized(), which is
>> WorkerDataArray<T>::uninitialized() in disguise:
>>
>> template <>
>> size_t WorkerDataArray<size_t>::uninitialized() {
>>   return (size_t)-1;
>> }
>>
>> template <>
>> double WorkerDataArray<double>::uninitialized() {
>>   return -1.0;
>> }
> 
> Hmm, ok then. The compiler will know what to do with it (I hope).

Yes. Even if compiler screws up, the flushes are done outside the pause ;)

-- 
Thanks,
-Aleksey