FRF: Wiring GC events to JFR

[Aleksey Shipilev]

On 08/27/2018 05:04 PM, Zhengyu Gu wrote:
>>> http://cr.openjdk.java.net/~zgu/shenandoah/jfr-event-1/webrev.00/
>>
>> *) The introduction of GCTraceConcTime means it changes our logging, right? For example, it would
>> print "Concurrent marking" without "gc+start", right? This is inconvenient and makes the logs noisy.
>> Why do we need that change?
> 
> Well, GCTraceTime counts as pause time, that will distort the pause time displayed in JMC.

Okay, too bad. Let's bite the bullet and do GCTraceConcTime then.


>> *) The comment is oddly formatted, and it seems "JMC" -> "JFR" is in order? Suggestion:
>>
>>     // FIXME: It seems that JFR throws away level 0 events, which are the Shenandoah
>>     // pause events. Create this pseudo level 0 event to push real events to level 1
> 
> JFR actually sends level 0 events, but are dropped by JMC.

All right. But reformat the comment into two lines like above.

Thanks,
-Aleksey