RFR(XXS) 8222188: Shenandoah: Adjust Shenandoah work gang types

Zhengyu Gu zgu at redhat.com
Tue Apr 9 14:59:37 UTC 2019



On 4/9/19 10:19 AM, Aleksey Shipilev wrote:
> On 4/9/19 4:13 PM, Zhengyu Gu wrote:
>> This is a cleanup in preparation for concurrent class unloading.
>>
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8222188
>> Webrev: http://cr.openjdk.java.net/~zgu/JDK-8222188/webrev.00/index.html
> 
> OK. Changing _safepoint_workers to be GC_Task_threads does not violate correctness?

Reverted this part.

Was following G1, but ZGC setup the same way as we do. I guess that 
shared code does not depends on this property.

> 
> Nits:
> 
>   *) Changeset synopsis is odd?
>     "1. cleanup for concurrent class unloading"

This just my local mq message.

> 
>   *) Whitespace missing after "*/" at L453 here:
> 
>   452   _workers = new ShenandoahWorkGang("Shenandoah GC Threads", _max_workers,
>   453                             /* are_GC_task_threads */true,
>   454                             /* are_ConcurrentGC_threads */ true);
> 
>   *) Better indenting maybe, like this?
> 
>     assert(!Thread::current()->is_Worker_thread() &&
>                (Thread::current()->is_VM_thread() ||
>                 Thread::current()->is_ConcurrentGC_thread())

Missed another occurrence in shenandoah/shenandoahTimingTracker.cpp

Updated: http://cr.openjdk.java.net/~zgu/JDK-8222188/webrev.01/


Thanks,

-Zhengyu



> 
> Thanks,
> -Aleksey
> 



More information about the hotspot-gc-dev mailing list