RFR: 8241881: ZGC: Add tests for JFR events

Per Liden per.liden at oracle.com
Wed Apr 1 05:38:50 UTC 2020


Hi Erik,

On 3/31/20 4:46 PM, Erik Gahlin wrote:
> Hi Per,
> 
> Would it be possible to verify the values in the event?
> 
> I don’t understand the semantics of the events, so it is hard for me to suggest what should actually be verified, but some sort of sanity check seems reasonable. For example, can capacityBefore be less than capacityAfter, or vice versa, in ZUncommit.


Thanks for reviewing. In general, checking these values is hard since 
they are not always predictable, especially in a jtreg context where 
additional threads are running. However, you do have a point about 
ZUncommit, where we at least could do some internal consistency checks 
of the values. I've also updated ZAllocationStall/ZPageAllocation events 
to check that the size is at least the min page size (2M).

Updated webrev: http://cr.openjdk.java.net/~pliden/8241881/webrev.1

cheers,
Per

> 
> Thanks
> Erik
> 
>> On 30 Mar 2020, at 22:38, Per Liden <per.liden at oracle.com> wrote:
>>
>> Add tests for the newly added ZGC-specific JFR events that we intend to make non-experimental as part of JEP 377. These events are:
>>
>> ZAllocationStall
>> ZPageAllocation
>> ZPageCacheFlush
>> ZRelocationSet
>> ZRelocationSetGroup
>> ZUncommit
>>
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8241881
>> Webrev: http://cr.openjdk.java.net/~pliden/8241881/webrev.0
>>
>> /Per
> 


More information about the hotspot-jfr-dev mailing list