[PATCH/RFA] Update m68k function return value handling

[Michael Snyder]

Andreas Schwab wrote:
> Michael Snyder <msnyder@redhat.com> writes:
> 
> 
>>FYI, there's 80 or more  fails for an m68k-elf cross target.
>>I'm not sure if it's better or worse with your patch.
>>Andreas, it seems like some of your recent changes (last six
>>months or so) have broken the cross target.
> 
> 
> The last time I changed something for m68k was on 2003-09-25, which is
> already more than 6 months ago.  The biggest changes of mine happend about
> one year ago, with another batch about 8 months ago, where I converted the
> m68k backend to move away from the deprecated interfaces.

Yeah, it seems likely that no one has built an m68k cross toolchain
in the intervening time.  ;-(


> Would it be possible for you to find out which changes are bad?  I have
> them all tested natively on m68k-linux as much as possible.

I'm investigating.  The first problem seems to be the choice of
pcc_struct_return as the default.  GCC doesn't use that, whether
for native or cross.  After that, I'm suspecting that we still
need stack_align.  One changelog entry says that you replaced
stack_align with gdbarch_parm_boundary -- the trouble is that
they aren't the same thing.