RFR (S) 8031045: Access checks should precede additional per-instruction checks

[Lois Foltan]

Hi Harold,

I think this looks good.  One minor nit is that the problem report 
justifies the change by citing section 5.4.3.3 Method Resolution of the 
JVM spec, after looking at JSR 335, I actually think this falls under 
section 5.4.3.4 Interface Method Resolution.  The invokeinterface 
bytecode is generated via visitMethodInsn where the last parameter is 
"true" indicating an IMR.

Lois

On 1/16/2014 1:23 PM, harold seigel wrote:
> Hi,
>
> Please review this small fix for bug 8031045.  The fix changes the 
> implementation so that it does an access check before doing the 
> non-static method ICCE error check.  This ensures that if the method 
> is both not accessible and not static, that an IllegalAccessError 
> exception will get throw.
>
> The fix was tested with the test in the bug report, the JCK lang and 
> vm tests, the defmeth tests, the ute mlvm tests, and the hotspot, 
> lambda, java util, and java lang jtreg tests.
>
> Open webrev at: http://cr.openjdk.java.net/~hseigel/bug_8031045/
> <http://cr.openjdk.java.net/%7Ehseigel/bug_8031045/>
> Bug: https://bugs.openjdk.java.net/browse/JDK-8031045
>
> Thanks! Harold