Re: Making the same mistake as the broken C interface to fortran

[Chris M Moore <zmower@ntlworld.com>]

On 05/07/2019 18:44, Simon Wright wrote:
> Chris M Moore <zmower@ntlworld.com> writes:
> 
>> On 04/07/2019 09:38, Simon Wright wrote:
>>> procedure Call is
>>>      procedure Callee_C (C : Character)
>>>      with
>>>        Import,
>>>        Convention => Fortran,
>>>        External_Name => "callee_";
>>>      procedure Callee_F (C : Interfaces.Fortran.Fortran_Character)
>>>      with
>>>        Import,
>>>        Convention => Fortran,
>>>        External_Name => "callee_";
>>>      procedure Callee_S (S : String)
>>>      with
>>>        Import,
>>>        Convention => Fortran,
>>>        External_Name => "callee_";
>>> begin
>>>      Callee_C ('c');
>>
>> I've looked at the assembler for this.  Passing 'c' results in only a
>> byte pushed to the stack.  And then (aha!) the address on the stack is
>> placed in the rdi register.
> 
> Because all Fortran parameters are passed by reference.


I spoke too soon when I said

 > I'm sure GNAT does the right thing if you're using Fortran_Character.

If I change callee.f to

       subroutine callee (c)
       character (len=*), intent (in) :: c

       print *, 'parameter c is ', c

       end

then STORAGE_ERROR is the order of the day no matter the call used. 
Looking at the assembler, this is because GNAT does not pass the length 
of the string.

I compared it to fcall.f:

       program fcall
       call callee("OK")
       call callee("Oh noes")
       stop
       end

and this unsurprisingly does pass the lengths.

I've used the webform on the Community section of the GNAT website to 
provide feedback.  I've pointed out that the issue also affects single 
chacter parameters.

Chris

-- 
sig pending (since 1995)