[11u] RFR 8251487: Shenandoah: missing detail timing tracking for final mark cleaning phase

[Zhengyu Gu]

On 8/13/20 5:24 PM, Roman Kennke wrote:
> Hi Zhengyu,
> 
> - the order of placement of _phase and its initializer don't match.
> Some compilers are not going to like this.
> - the indentation of _phase wrt the other fields doesn't match
> - Alignment don't match here:
> 
> +  ParallelCleaningTask(ShenandoahPhaseTimings::Phase phase,
> BoolObjectClosure* is_alive, bool process_strings,
> +    bool process_symbols, uint num_workers, bool unloading_occurred);

Updated: http://cr.openjdk.java.net/~zgu/JDK-8251487/webrev.01/index.html

Okay now?

Thanks,

-Zhengyu

> 
> 
> The rest looks good to me.
> 
> Thank you!
> Roman
> 
> On Wed, 2020-08-12 at 14:10 -0400, Zhengyu Gu wrote:
>> Please review this small enhancement that adds detail timing
>> tracking
>> for final mark cleaning phase. The timing information is very useful
>> to
>> diagnose latency issues.
>>
>> The patch is 11u specific, since 11u code structure is quite
>> different
>> from jdk/jdk, mainly due to concurrent class unloading in jdk/jdk.
>> The change is completed isolated in Shenandoah.
>>
>>
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8251487
>> Webrev: http://cr.openjdk.java.net/~zgu/JDK-8251487/webrev.00/
>>
>> Test:
>>     hotspot_gc_shenandoah
>>
>> Thanks,
>>
>> -Zhengyu
>>
>