Infelicity in OpenJDK

[Gary Benson]

Hi Ed,

Good catch, I'll get a fix committed on Monday or Tuesday when I'm
back in.

Cheers,
Gary

Edward Nevill wrote:
> There is a potential insecurity in native_entry() in cppInterpreter_zero.cpp
>  
> It does
>                 void **arguments;
>                 {
>                                 ...
>                                 void **dst = arguments;
>                                 ...
>  
>                                 void *mirror = NULL;
>                                 if (method->is_static()) {
>                                                 ...
>                                                 *(dst++) = &mirror;
> // ECN: Address of mirror stored in argument list here
>                                 }
>                                 ...
>  
>                 }
>  
> // ECN: Now follows more code which processes the argument list
> // however, 'mirror' is no longer in scope so the address is
> // potentially invalid.
>  
> What needs to happen is the definition of 'mirror' needs to be moved
> up to the same scope as 'arguments'.
>  
> Regards,
> Ed.

-- 
http://gbenson.net/