Hi Michael,

you wrote in <9503160530.AA24693@gemini.oscs.montana.edu>:
>   I don't see anything wrong with that code, which would rule out a
> compiler bug.  I don't see anything that should cause any problems on
> the 68040, unless it might be some kind of timing problem with the two
> successive byte stores [a2@(13) and a2@(14)].

What about inserting a "NOP" between the two instruction?
(Does GCC support this?) That should fix the timing problem.

-- 
Matthias Scheler
tron@lyssa.owl.de