RFR: Fix TestUnifiedLoggingSwitchStress test failure

[Zhengyu Gu]

On 06/01/2018 12:44 PM, Aleksey Shipilev wrote:
> On 06/01/2018 06:42 PM, Zhengyu Gu wrote:
>>>> okay: http://cr.openjdk.java.net/~zgu/shenandoah/degen_tq_stats/webrev.01/
>>>
>>> Looks good.
>>>
>>> But I still wonder why can't we simplify the whole thing? Remove the TQ stats block from
>>> SCM::mark_from_roots() completely, move reset past the print in SCM::finish_mark_from_roots()? Let
>>> the queues finish, and print+reset.
>>>
>>> For example:
>>>
>>> http://cr.openjdk.java.net/~shade/shenandoah/tq-why/webrev.01/src/hotspot/share/gc/shenandoah/shenandoahConcurrentMark.cpp.sdiff.html
>>>
>>
>> I can try this, but I suspect it only works with all success cycles.
> 
> Please do. I think Degenerated should finish up all queue work in the same
> SCM::finish_mark_from_roots, so by the time we are printing, everything should be in order. It makes
> little sense trying to print queue stats while there is still work in them.

Make sense. Although, there is not an issue to print stats at the end of 
conc-mark, just not to reset queue stats, in case there is degen cycle
followed.

Webrev: http://cr.openjdk.java.net/~zgu/shenandoah/degen_tq_stats/webrev.02/

Test:

Reran hotspot_gc_shenandoah.

Thanks,

-Zhengyu

> 
> -Aleksey
>