[Bug fortran/107266] Reject kind=4 characters for BIND(C) – it invalid and generates wrong code

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107266

--- Comment #3 from kargl at gcc dot gnu.org ---
(In reply to kargl from comment #2)
> (In reply to Tobias Burnus from comment #1)
> > Better example:
> > 
> > character(kind=4) function f(x) bind(C)
> >   character(kind=4), value :: x
> > end
> > 
> > 
> > Has the dump:
> > 
> > __attribute__((fn spec (". w ")))
> > character(kind=1) f (character(kind=1) x)
> > {
> >   (void) 0;
> > }
> > 
> > 
> > Note the change to kind=1
> 
> So, you want gfortran to reject the function.
> 
> It gets real ugly with
> 
> function f(x) bind(C) result(a)
>    character(kind=4) a
>    character(kind=4), value :: x
>    a = x
> end
> 
> leading to
> 
> __attribute__((fn spec (". w ")))
> character(kind=1) f (character(kind=1) x)
> {
>   character(kind=1) a;
> 
>   *(character(kind=4) *) &a = *(character(kind=4) *) &x;
>   return a;
> }

Looks like resolve.cc:resolve_symbol() might be able to dela with this.
For my function with the result variable one has

(gdb) p *sym
$4 = {name = 0x2041ec100 "f", module = 0x0, declared_at = {
    nextc = 0x203a38160, lb = 0x203a38140}, ts = {type = BT_UNKNOWN, kind = 0, 
...
    in_equivalence = 0, function = 1, subroutine = 0, procedure = 0, 
    generic = 0, generic_copy = 0, implicit_type = 0, untyped = 0, 
    is_bind_c = 1, extension = 0, is_class = 0, class_ok = 0, vtab = 0, 
...
   value = 0x0, as = 0x0, result = 0x203a57600, components = 0x0, 

(gdb) p *sym->result
$5 = {name = 0x2041ec110 "a", module = 0x0, declared_at = {
    nextc = 0x203a381dc, lb = 0x203a38140}, ts = {type = BT_CHARACTER, 
    kind = 4, u = {derived = 0x203a245a0, cl = 0x203a245a0, pad = 60966304}, 
...
    dummy = 0, result = 1, assign = 0, threadprivate = 0, 

so gfortran can check for kind=4 versus kind=1 (aka C_CHAR).