The root cause for SEGV in evaluating fortran function call, any solution or suggestion?

[Jim Blandy]

Daniel Jacobowitz <drow@false.org> writes:
> On Wed, Nov 09, 2005 at 04:55:27PM -0800, Jim Blandy wrote:
>> 
>> Daniel Jacobowitz <drow@false.org> writes:
>> > Maybe if you pass the sp value by reference to value_arg_coerce and
>> > adjust it there...
>> 
>> Or you could pass a regcache.
>
> We're not trying to adjust the saved sp in the regcache, but the
> interim SP being used in setting up the function - it's not in the
> regcache until much later (after push_dummy_call).

My thought was that a language-specific value_arg_coerce might need to
do all manner of strange things, perhaps affecting other registers.
It's really all of infcall.c that is C-specific.  But I may be getting
ahead of myself.