mips64 n32 and n64 support in dl-machine.h

[Alexandre Oliva]

On Mar 18, 2003, Andreas Schwab <schwab@suse.de> wrote:

> Have you tried my suggestion yet?  I still think that would be the
> cleaner approach.

Your suggestion might work today, but I've already explained why a
smarter compiler might be able to defeat it, if not now, at some point
in which unrelated `clean-ups' change void* to the actual types in
callers.  I don't see what's not clean about this approach, anyway.
It adds no instructions whatsoever, and it makes sure no assumptions
can possibly be made by the compiler about the alignment of the
relocation value, which is exactly what we want.  I.e., it's more
robust than any change elsewhere.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                 aoliva@{redhat.com, gcc.gnu.org}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist                Professional serial bug killer