RFR: 8148745: [testbug] Test gc/g1/plab/TestPLABPromotion.java fails in nightly

Dmitry Fazunenko dmitry.fazunenko at oracle.com
Mon Feb 8 13:35:10 UTC 2016


Hi Misha,

new version looks good.

Thanks,
Dima

On 08.02.2016 14:58, Michail Chernov wrote:
> Hi Dima,
>
> Thanks for you comment. Added comments to checkRatio and 
> checkDifferenceRatio methods in TestPLABPromotion.java.
>
> Webrev:
> http://cr.openjdk.java.net/~mchernov/8148745/webrev.01/
>
> Heap sizes are limited by OldSize (see PLABUtils.java) and NewSize and 
> MaxNewSize (see TestPLABPromotion.java).
>
> Thanks
> Michail
>
> On 05.02.2016 19:02, Dmitry Fazunenko wrote:
>> Hi Misha,
>>
>> In general the fix looks good to me.
>> A few comments:
>>
>> Would add short comment to the new introduced methods, pls
>>
>> 195 private static boolean checkRatio(long checkedValue, long 
>> controlValue) {
>> 199 private static boolean checkDifferenceRatio(long checkedValue, 
>> long controlValue) {
>>
>> > Checking of size of promoted objects was changed because difference 
>> of real allocated objects and
>> > promoted can exceed existing limit with big heap size.
>>
>> Would you consider to limit the heap size by -Xmx?
>>
>> Thanks
>> Dima
>>
>>
>> On 05.02.2016 18:09, Michail Chernov wrote:
>>> Hi all,
>>>
>>> Could I have a couple of review for this change:
>>>
>>> Webrev: http://cr.openjdk.java.net/~mchernov/8148745/webrev.00/
>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8148745
>>>
>>> Added -XX:MaxTenuringThreshold=1 to achieve objects promotion at 
>>> every GC.
>>> Added timeout for TestPLABPromotion.java because it can work 
>>> sufficient time with -Xcomp.
>>> Added @requires to exclude test execution with Flight Recorder, 
>>> because test cannot handle case when some additional objects are 
>>> created before or during test running.
>>> Checking of size of promoted objects was changed because difference 
>>> of real allocated objects and promoted can exceed existing limit 
>>> with big heap size.
>>>
>>> Thanks
>>> Michail
>>
>>
>




More information about the hotspot-gc-dev mailing list