Review request (L) 8003419: NPG: Clean up metadata created during class loading if failure

Coleen Phillimore coleen.phillimore at oracle.com
Tue Mar 5 11:30:46 PST 2013


Resend to hotspot-dev.   I don't think anyone reads hotspot-runtime-dev 
anymore.
Please review!!
Thanks,
Coleen

On 03/04/2013 03:13 PM, Coleen Phillimore wrote:
> Summary: Store metadata on ClassFileParser instance to be cleaned up 
> by destructor.  This enabled some refactoring of the enormous 
> parseClassFile function.
>
> Note: I moved out code that lays out the java object into it's own 
> function, but the diffs made it look like I made more than minor 
> changes.  The changes I made was changing class_name to _class_name 
> and cp to _cp, and not more than that.   To see these changes, you'll 
> have to look at before/after code and not the diffs.  I also tried 
> moving this layout_fields() function around to it wouldn't false match 
> but I wanted to keep it next to parseClassFile() and not move it too 
> far away.
>
> Tested with runThese jck, jck8 lang and vm, NSK vm.quick.testlist, 
> java/lang/invoke jtreg tests and java/lang/instrument jtreg tests.
>
> open webrev at http://cr.openjdk.java.net/~coleenp/8003419/
> bug link at http://bugs.sun.com/view_bug.do?bug_id=8003419
>
> Thanks,
> Coleen
>



More information about the hotspot-dev mailing list