Re: Patch: RFC: alignment fix in resolve.cc

[Andrew Haley <aph at redhat dot com>]

Tom Tromey writes:
 > Andrew> I think you've got it back to front.  All that the new code does is
 > Andrew> fix things so that CNI code is compliant with what gcj has always
 > Andrew> done.  clz->superclass->size() should already take this into account.
 > 
 > Thanks.
 > 
 > Andrew> The Right Thing to do is always to pad the size of a class so that its
 > Andrew> size is an integer multiple of its alignment.  As long as that's done
 > Andrew> everything will be fine, and the extra padding is already there.
 > 
 > Ok, I think we don't do this.  After we add the last field we don't do
 > any rounding of `instance_size', which we then use as the class's
 > size.
 > 
 > So it looks like either my patch or a different one (it would be
 > simpler to round before assigning to clz->size_in_bytes) is still
 > needed.  I'll see if I can come up with an example.

It's easiest to do this on a 64-bit host that requires Java longs to
be 8-aligned. 

Andrew.