[patch, fortran] Fix PR 87689, wrong decls / ABI violation on POWER

[Janne Blomqvist]

On Mon, Feb 18, 2019 at 2:51 PM Richard Biener
<richard.guenther@gmail.com> wrote:
> The patch looks good to me.  I wonder how the frontend handles
> the 2nd call to doesntwork_p8 for
>
>       program main
>         implicit none
>         character :: c
>         character(len=20) :: res, doesntwork_p8
>         external doesntwork_p8
>         c = 'o'
>         res = doesntwork_p8(c,1,2,3,4,5,6)
>         res = doesntwork_p8(1,2)
>         if (res /= 'foo') stop 3
>       end program main
>
> at least I get no diagnostic and the patch suggests whatever call
> gets there first determines the backend-decl and its type?  At least
> when I omit res = from the second call I get
>
> t.f:4:47:
>
>     4 |         character(len=20) :: res, doesntwork_p8
>       |                                               1
> ......
>     8 |         call doesntwork_p8(1,2)
>       |
> Error: ‘doesntwork_p8’ at (1) has a type, which is not consistent with
> the CALL at (2)
>
> Does the Fortran standard say anything in how the above case should be handled?

Without actually checking, I'm pretty sure a procedure is either a
function (something that returns a value) or a subroutine (does NOT
return a value, like a void function in C), but not both (some of the
intrinsics in gfortran can be called both as a function or a
subroutine, but that's a gfortran-specific extension (inherited from
g77, AFAIK), not part of the language, and only for some intrinsics
that the compiler handles in a special way, not a generic user-defined
procedure). So the example you show above is invalid.


-- 
Janne Blomqvist