RFR (S): 8004172: Update jstat counter names to reflect metaspace changes

Erik Helin erik.helin at oracle.com
Thu Feb 7 12:36:06 UTC 2013

Hi Mandy,

thanks for the review! See comments inline.

On 02/05/2013 09:14 PM, Mandy Chung wrote:
> Hi Erik,
> On 2/4/13 8:01 AM, Erik Helin wrote:
>> There have been some updates to the hotspot code based on feedback from
>> the hotspot-gc-dev mailing list. The only change to the JDK code is that
>> the namespace is back to "sun.gc" :)
>> New webrevs:
>> - jdk: http://cr.openjdk.java.net/~ehelin/8004172/jdk/webrev.03/
>> - hotspot: http://cr.openjdk.java.net/~ehelin/8004172/hotspot/webrev.03/
> Looks good.  Minor nits:
> gcCapacityOutput1.awk - should the last column in L7 be removed?

I don't think so, jmap -gccause outputs the GC cause in the last column, 
which sometimes can be "No GC".

Why do you think it should be removed?

On 02/05/2013 09:14 PM, Mandy Chung wrote:
> jstatGcMetaCapacityOutput1.sh - it's a new test and the copyright year
> should be 2013.


See new webrev for jdk at:

>> Note that the fixed tests will continue to fail until the HotSpot
>> part of the change finds it way into the tl forest.

On 02/05/2013 09:14 PM, Mandy Chung wrote:
> How are you going to integrate the hotspot and jdk fixes in a
> synchronized fashion?  We shall make sure the TL sun/tools/jstat
> tests pass at all times.

Thanks for catching this. the tests in tl do passes (I was wrong about 
this). The change will have to be pushed in several steps:
1. Add new metaspace counters (and keep the old ones). This will be
    pushed to hotspot-gc.
2. Wait for the change to trickle down to tl.
3. Update the tests and jstat to use the new metaspace counters.
4. Wait for this change to trickle down to hotspot-gc.
5. Remove the old metaspace counters and push this to hotspot-gc.

Do you want me to send out an additional webrev for what the hotspot 
code will look like with both the old and new counters?


> Mandy

More information about the hotspot-gc-dev mailing list