RFR (S) JDK-8028741: Interface Method Resolution should skip static and non-public methods in j.l.Object

[Lois Foltan]

On 12/10/2013 2:25 PM, Leonid Mesnik wrote:
> Lois
>
> It is a bad practice to call System.exit(1) if test fails. It cause 
> issue in samevm mode.  It is better to throw exception.
> See more information about failures here:
> http://openjdk.java.net/jtreg/faq.html#question7.1

Hi Leonid,
Thank you for the code review!  I will remove the System.exit(1) call 
prior to committing.
Lois

>
> Leonid
> On 12/10/2013 04:40 AM, Lois Foltan wrote:
>> Please review the following fix:
>>
>> Webrev:
>>     http://cr.openjdk.java.net/~lfoltan/bug_jdk8028741/
>>
>> Fixes both: JDK-8028741: Interface Method Resolution should skip 
>> static and non-public methods in j.l.Object
>>                     JDK-8028626: invokeinterface selects the static 
>> Object.registerNatives method
>>
>> https://bugs.openjdk.java.net/browse/JDK-8028741
>> https://bugs.openjdk.java.net/browse/JDK-8028626
>>
>> Summary of fix:
>>     Implementation for the JDK 8 JVMS 5.4.3.4 specification change 
>> requested in JDK-8027581, to
>>     skip static and non-public methods of java.lang.Object, such as 
>> clone, finalize and registerNatives,
>>     for interface method resolution.  This fix also removes the 
>> temporary workaround that was put in
>>     place for JDK-8026394.  The workaround special cased checking for 
>> java.lang.Object methods
>>     of name "clone" and "finalize" during interface method 
>> resolution's interface method accessability
>>     checking.  Since non-public java.lang.Object methods such as 
>> clone and finalize, are now being ignored
>>     during method resolution, this temporary work around is no longer 
>> necessary.
>>
>> Tests:
>>     JCK lang & vm - passed
>>     JDK java/lang & java/util - passed
>>     vm.quick.testlist - passed
>>     Hotspot test/* - in progress
>>     default method tests - passed
>>
>> Thank you,
>> Lois
>>
>
>