Always enforce ACC_SUPER

Florian Weimer fweimer at redhat.com
Fri May 18 06:32:03 PDT 2012


On 05/18/2012 02:44 PM, Keith McGuigan wrote:
> I haven't investigated (maybe you have), but does the JVMS allow this
> sort of behavior?

I'm not sure.  It probably doesn't.  This part of the JVMS is for 
compatibility with pre-1.0.2 implementations, and 1.0.2 was the first 
publicly released version.  So I hope that this deviation is acceptable.

> If it does (and if it doesn't, then that's the place
> to start), might it make sense to limit setting this flag implicitly to
> newer versions of classfiles (say >=51), or only to classes which have
> an ACC_SUPER set in their set of non-Object superclasses, or something
> like that? Seems like we should have something in place to mitigate any
> backward-compatibility concerns in order to go forward with this.

Unfortunately, we have to make this change in a way such that classes 
can't opt out of it.  Otherwise, it will not have any affect.

-- 
Florian Weimer / Red Hat Product Security Team


More information about the hotspot-dev mailing list