RFR (S): 8020697: jniCheck.cpp:check_is_obj_array asserts on TypeArrayKlass::cast(aOop->klass())

David Simms david.simms at oracle.com
Thu Jul 18 00:24:37 PDT 2013


On 18/07/2013 5:03 a.m., David Holmes wrote:
> David,
>
> On 17/07/2013 9:10 PM, David Simms wrote:
>> G'day,
>>
>> Please review a small fix:
>>
>> http://cr.openjdk.java.net/~dsimms/8020697/
>> <http://cr.openjdk.java.net/%7Edsimms/8020697/>
>>
>> Bug: https://jbs.oracle.com/bugs/browse/JDK-8020697#comment-13359065
>>
>> Summary of fix:  JDK-6671508 (-Xcheck:jni) fix should have used original
>> "is_objArray()" rather than messing around with "element_type()"
>> (unnecessary and bad typeArray cast asserts)
>
> Sorry that this escaped notice. Looking at the original code again it 
> seems the T_OBJECT==elementType was more addressing an incorrect call 
> to the checkArray function than inspecting the array itself.
>
This was my bad :-)
> Your fix looks okay.
>
> Have you identified testing that would have caught this prior to pushing?
>
Yeah, I rerun the tests that failed nightly. There is a QA engineer 
looking into how we can run tests with JNI code...the existing shell 
scripts leave something to be desired.
>> Needs a sponsor.
>
> I can do that for you.
Please, and thank you sir.
>
> Thanks,
> David H.
>
>
>> Cheers
>> /David Simms



More information about the hotspot-runtime-dev mailing list