Patch: Remove JvAllocObject and add fields to Class for interpreter/JIT data

[Bryce McKinlay]

Tom Tromey wrote:

>Bryce>  extern inline jobject
>Bryce>  JvAllocObject (jclass cls, jsize sz)
>Bryce>  {
>Bryce> -  return _Jv_AllocObject (cls, sz);
>Bryce> +  return _Jv_AllocObject (cls);
>Bryce>  }
>
>Does this function even make sense any more?
>I think we should remove it and update gcj.texi.
>  
>

My naive attempt to get rid of these turned into a major patch when I 
noticed that defineClass.cc actually does call JvAllocObject with a size 
argument that is not equal to the size_in_bytes of the class it is 
creating. The problem is that the interpreter actually overrides 
java.lang.Class to store additional, interpreter specific fields. Since 
the goal is pluggable JITs, I decided it would make more sense to add 
fields to java.lang.Class where JIT-specific data can be stored.

Please comment on the following patch, particularly the names and nature 
of the new fields in Class.h

Thanks

Bryce.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: interp-alloc-2.patch
Type: text/x-patch
Size: 20661 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/java-patches/attachments/20040417/b1c71368/attachment.bin>