[Bug fortran/53818] New: -fno-automatic cannot have initializer

[Bug fortran/53818] New: -fno-automatic cannot have initializer

[j.moyard at voila dot fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53818

             Bug #: 53818
           Summary: -fno-automatic cannot have initializer
    Classification: Unclassified
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: j.moyard@voila.fr


I try to compile this fortran code on Red Hat 5.7 linux x86_64 platform :
logical function testing(date1, date2) result(test)
integer  date1, date2
test = ( (date1 < date2) .or. ( date1==date2 ))
end function testing

But when I integrate '-fno-automatic' option, I obtain this error message :
function testing(date1, date2) result(test)
                                          1
Error: Function result 'test' at (1) cannot have an initializer

Is-it a normal issue?

[Bug fortran/53818] -fno-automatic cannot have initializer

[steven at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53818

Steven Bosscher <steven at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |steven at gcc dot gnu.org

--- Comment #1 from Steven Bosscher <steven at gcc dot gnu.org> 2012-06-30 13:38:10 UTC ---
What version of gfortran are you using? (gfortran --version)

If you're using the standard RH5.7 gfortran, it's probably version 4.1.x, which
is from the days that the environment was still clean and dinosaurs still
roamed the earth. Well, almost anyway. Try upgrading.

[Bug fortran/53818] -fno-automatic cannot have initializer

[j.moyard at voila dot fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53818

--- Comment #2 from johnjohn <j.moyard at voila dot fr> 2012-06-30 13:51:13 UTC ---
(In reply to comment #1)
> What version of gfortran are you using? (gfortran --version)
> 
> If you're using the standard RH5.7 gfortran, it's probably version 4.1.x, which
> is from the days that the environment was still clean and dinosaurs still
> roamed the earth. Well, almost anyway. Try upgrading.

No no, the problem does not appear with the gfortran 4.1.x packaged with Red
Hat, The problem appears with another gfortran version 4.6.0.

[Bug fortran/53818] -finit-local-zero -fno-automatic: Function result cannot have initializer

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53818

Tobias Burnus <burnus at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |rejects-valid
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-06-30
                 CC|                            |burnus at gcc dot gnu.org
            Summary|-fno-automatic cannot have  |-finit-local-zero
                   |initializer                 |-fno-automatic: Function
                   |                            |result cannot have
                   |                            |initializer
     Ever Confirmed|0                           |1

--- Comment #3 from Tobias Burnus <burnus at gcc dot gnu.org> 2012-06-30 14:31:29 UTC ---
(In reply to comment #0)
> But when I integrate '-fno-automatic' option, I obtain this error message :
> function testing(date1, date2) result(test)
>                                           1
> Error: Function result 'test' at (1) cannot have an initializer

I helps if you tell up all options. The issue seemingly only occurs with
-fno-automatic *and* -finit-local-zero (or -finit-logical=true|false).


Two things surprise me:

a) Why are there two initializer, when looking at the dump
(-fdump-tree-original). (Requires that one does not have -fno-automatic) - the
should be only one, shouldn't it?


b) Why does this only trigger with -fno-automatic? The code currently has the
following. At a glance, it should always trigger and the
apply_default_init_local should be after the checks (together with the
simplify):

resolve_fl_variable (gfc_symbol *sym, int mp_flag)
...
  if (sym->value == NULL && sym->attr.referenced)
    apply_default_init_local (sym); /* Try to apply a default initialization. 
*/
...
  /* Reject illegal initializers.  */
  if (!sym->mark && sym->value)
...
      else if (sym->attr.result)
        gfc_error ("Function result '%s' at %L cannot have an initializer",
                   sym->name, &sym->declared_at);

[Bug fortran/53818] -finit-local-zero -fno-automatic: Function result cannot have initializer

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53818

--- Comment #4 from Tobias Burnus <burnus at gcc dot gnu.org> 2012-06-30 14:41:08 UTC ---
Draft patch - completely untested.

--- a/gcc/fortran/resolve.c
+++ b/gcc/fortran/resolve.c
@@ -10271,3 +10330,3 @@ apply_default_init_local (gfc_symbol *sym)
   if (sym->attr.save || sym->ns->save_all 
-      || (gfc_option.flag_max_stack_var_size == 0
+      || (gfc_option.flag_max_stack_var_size == 0 && !sym->attr.result
          && (!sym->attr.dimension || !is_non_constant_shape_array (sym))))

[Bug fortran/53818] -finit-local-zero -fno-automatic: Function result cannot have initializer

[j.moyard at voila dot fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53818

--- Comment #5 from johnjohn <j.moyard at voila dot fr> 2012-07-05 16:32:44 UTC ---
(In reply to comment #3)
> (In reply to comment #0)
> > But when I integrate '-fno-automatic' option, I obtain this error message :
> > function testing(date1, date2) result(test)
> >                                           1
> > Error: Function result 'test' at (1) cannot have an initializer
> 
> I helps if you tell up all options. The issue seemingly only occurs with
> -fno-automatic *and* -finit-local-zero (or -finit-logical=true|false).
> 
> 
> Two things surprise me:
> 
> a) Why are there two initializer, when looking at the dump
> (-fdump-tree-original). (Requires that one does not have -fno-automatic) - the
> should be only one, shouldn't it?
> 
> 
> b) Why does this only trigger with -fno-automatic? The code currently has the
> following. At a glance, it should always trigger and the
> apply_default_init_local should be after the checks (together with the
> simplify):
> 
> resolve_fl_variable (gfc_symbol *sym, int mp_flag)
> ...
>   if (sym->value == NULL && sym->attr.referenced)
>     apply_default_init_local (sym); /* Try to apply a default initialization. 
> */
> ...
>   /* Reject illegal initializers.  */
>   if (!sym->mark && sym->value)
> ...
>       else if (sym->attr.result)
>         gfc_error ("Function result '%s' at %L cannot have an initializer",
>                    sym->name, &sym->declared_at);

Yes in fact, I use finit-local with fno-automatic!

[Bug fortran/53818] -finit-local-zero -fno-automatic: Function result cannot have initializer

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53818

--- Comment #6 from Tobias Burnus <burnus at gcc dot gnu.org> 2013-02-15 08:50:43 UTC ---
Author: burnus
Date: Fri Feb 15 08:50:37 2013
New Revision: 196073

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=196073
Log:
2013-02-15  Tobias Burnus  <burnus@net-b.de>

        PR fortran/53818
        * resolve.c (apply_default_init_local): Don't create an
        initializer for a result variable.

2013-02-15  Tobias Burnus  <burnus@net-b.de>

        PR fortran/53818
        * gfortran.dg/init_flag_11.f90: New.


Added:
    trunk/gcc/testsuite/gfortran.dg/init_flag_11.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/resolve.c
    trunk/gcc/testsuite/ChangeLog

[Bug fortran/53818] -finit-local-zero -fno-automatic: Function result cannot have initializer

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53818

Tobias Burnus <burnus at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |burnus at gcc dot gnu.org
         Resolution|                            |FIXED

--- Comment #7 from Tobias Burnus <burnus at gcc dot gnu.org> 2013-02-15 08:52:32 UTC ---
Finally FIXED – the approval was quite a while ago
http://gcc.gnu.org/ml/fortran/2012-07/msg00003.html

Thanks for the bugreport and sorry for the delay.