Richard Earnshaw wrote:
> 
> > Richard Earnshaw wrote:
> > >
> > > > Richard Earnshaw wrote:
> > > > >
> > > > > > Richard Earnshaw wrote:
> > > > > > >
> > > > > > > Leaving asside the issue of the correctness of write_register_bytes (note
> > > > > > > to self, must finish of my register patches), I don't think this is
> > > > > > > correct -- in fact, I think it's also wrong for little-endian as well.
> > > > > > >
> > > > > > > What should happen is that the smaller-than-word value should be
> > > > > > > zero/sign-extended to 32 bits and then the whole thing stored in A1_REGNUM.
> > > > > >
> > > > > > Ah, thanks.  OK, how about this?
> > > > > >
> > > > > > 2002-11-06  Michael Snyder  <msnyder@redhat.com>
> > > > > >
> > > > > >       * arm-tdep.c (arm_store_return_value): Handle offset of
> > > > > >       small types on big-endian machines.
> > > > >
> > > > > And for little-endian?
> > > >
> > > > It already works for little-endian.  I've tested this with
> > > > arm-sim, arm-sim/-mbig-endian, and arm-sim/-mthumb.
> > >
> > > But it's not zero/sign extending properly for little-endian, so garbage is
> > > remaining in the top part of A1
> >
> > Ah; well, I didn't make it any worse!  ;-)
> > Can I leave that detail for someone else, and just
> > submit this minor improvement?
> 
> Given that to fix this for little-endian as well means that you just have
> to *remove* the endianness test from your patch, why is that so hard???!!!!

Oh, I see.  Yes, it's no longer necessary.  But if it's not so hard, 
and you already see what's necessary, why not just do it instead of
ragging on me to do it?   ;-)

This what you have in mind?  Please just take it from here, I really
didn't intend to spend this much time and energy on it.