RFR: JDK-8215196: [Graal] vmTestbase/nsk/jvmti/PopFrame/popframe003/TestDescription.java fails with "changes for the arguments of the popped frame's method, did not remain current argument values"

Alex Menkov alexey.menkov at oracle.com
Sat Nov 9 01:57:36 UTC 2019



On 11/08/2019 16:55, Chris Plummer wrote:
> Hi Alex,
> 
> Comments below:
> 
> On 11/8/19 4:39 PM, Alex Menkov wrote:
>>
>>
>> On 11/08/2019 15:22, Alex Menkov wrote:
>>> Hi all,
>>>
>>> Please review the fix for
>>> https://bugs.openjdk.java.net/browse/JDK-8215196
>>> webrev:
>>> http://cr.openjdk.java.net/~amenkov/jdk14/popframe_args/webrev/
> I don't really see a resolution in the JDK-8215196 comments as to what 
> is actually broken. Are we sure we want to fix this in the test, and not 
> require different behavior by the compiler (and also clarify the spec)?

In the test activeMeth method changes its arguments values and then 
don't use them later.
I think dropping useless code is good compiler optimization and I'd 
prefer to not restrict to do the optimization.

>>>
>>> Currently PopFrame is disabled with JVMCI by [1], so for testing I 
>>> reverted [1] changes.
>>
>> Just to be clear - I temporary reverted [1] for test runs.
>>
> The description for JDK-8218025 says that the intention is to only 
> disable these capabilities for JDK12. Is there a CR to re-enabled them?

https://bugs.openjdk.java.net/browse/JDK-8218885
Unfortunately the problem why the capabilities were disabled are still 
unresolved and looks like won't be resolved in 14, so for now it's 
targeted to tbd.

--alex

> 
> thanks,
> 
> Chris
>> --alex
>>
>>>
>>> [1] https://bugs.openjdk.java.net/browse/JDK-8218025
>>>
>>> --alex
> 


More information about the serviceability-dev mailing list