RFR: Restore string table rehashing task in cleanup tasks

[Zhengyu Gu]

On 06/14/2018 08:39 AM, Aleksey Shipilev wrote:
> On 06/14/2018 02:00 PM, Zhengyu Gu wrote:
>> I wrongly removed string table rehashing task from safeppont cleanup phase, thought it should be
>> performed concurrently.
>>
>> I believe that StringInternCleanup test failure is due to upstream bug - I filed JDK-8205006
>> (https://bugs.openjdk.java.net/browse/JDK-8205006) and temporary disable the test till the bug is
>> fixed.
>>
>> Webrev: http://cr.openjdk.java.net/~zgu/shenandoah/restore_stringtable_rehashing/webrev.00/
> 
> I think ignoring the test masks the issue. Can we keep the test around, but disable rehashing with
> UseShenandoahGC? E.g.:
> 
>   // Workaround upstream bug in StringTable rehashing, see JDK-8205006.
>   if (StringTable::needs_rehashing() && !UseShenandoahGC) {
> 
This functional equivalents to what we have right now, another option is 
to defer this till JDK-8205006 fixed upstream?

-Zhengyu

> This way we have non-broken VM -- with performance hit, but still non-broken?
> 
> Thanks,
> -Aleksey
>