RFR(S): 8013591 compiler/ciReplay/TestSA.sh fails in nightly

[Staffan Larsen]

On 8 maj 2013, at 13:50, Coleen Phillimore <coleen.phillimore at oracle.com> wrote:

> 
> This looks fine.   What tests did you run to find this bug?   I think Jiangli ran the sajdi and probably tmtools tests and the previous changes looked good to me.   They passed nightly which also should have exercised this code.   Can we add a test somewhere since the SA insists on bit-rotting with everything we touch?

The test name is in the subject. It is part of the hotspot jtreg suite, but added relatively recently: 2013-04-25. Jiangli's changes were made on 2013-04-10. It's possible that this is the first sighting of running with these two changes together.

/Staffan

> 
> thanks,
> Coleen
> 
> On 5/8/2013 4:17 AM, Staffan Larsen wrote:
>> When JDK-8010862 (Method counter fields used for profiling can be allocated lazily) was fixed, SA was not updated to reflect this change.
>> 
>> The problem is with Method.getInvocationCounter() and Method.getBackedgeCounter() which do not handle the case where MethodCounters are not allocated.
>> 
>> The fix is to check for null and return 0 as the counter value.
>> 
>> webrev: http://cr.openjdk.java.net/~sla/8013591/webrev.00/
>> bug: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8013591
>> 
>> Thanks,
>> /Staffan
>