[Bug fortran/60717] New: Wrong code with recursive procedure with unlimited polymorphic dummy argument

[Bug fortran/60717] New: Wrong code with recursive procedure with unlimited polymorphic dummy argument

[vladimir.fuka at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60717

            Bug ID: 60717
           Summary: Wrong code with recursive procedure with unlimited
                    polymorphic dummy argument
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: vladimir.fuka at gmail dot com

reported on comp.lang.fortran by Thomas Schnurrenberger

The output of the two parts of the following program should be the same, as
they are with some other compilers (tested with ifort 14). The output with 4.8
and 4.9 is:


 Output of show_real:
   0.   1.   2.   3.   4.   5.
   0.
   1.   2.   3.   4.   5.
   1.
   2.   3.   4.   5.
   2.
   3.   4.   5.
   3.
   4.   5.
   4.
   5.
   5.
 Output of show_generic:
   0.   1.   2.   3.   4.   5.
   0.
   1.   2.   3.   4.   5.
   2.
   2.   3.   4.   5.
   3.
   3.   4.   5.
   4.
   4.   5.
   5.
   5.
   0.









module m
  !
  implicit none
  !
contains
  !
  recursive subroutine show_real(a)
    real, intent(in) :: a(:)
    !
    if (size(a) > 0) then
      print '(10f5.0)', a
      print '(10f5.0)', a(1)
      call show_real(a(2:))
    end if
    return
  end subroutine show_real
  !
  recursive subroutine show_generic(a)
    class(*), intent(in) :: a(:)
    !
    if (size(a) > 0) then
      select type (a)
      type is (real)
        print '(10f5.0)', a
        print '(10f5.0)', a(1)
      end select
      call show_generic(a(2:))
    end if
    return
  end subroutine show_generic
  !
end module m

program test
  !
  use :: m
  !
  implicit none
  !
  real :: array(1:6) = (/ 0, 1, 2, 3, 4, 5 /)
  !
  print *, 'Output of show_real:'
  call show_real(array)
  print *, 'Output of show_generic:'
  call show_generic(array)
  !
end program test

[Bug fortran/60717] Wrong code with recursive procedure with unlimited polymorphic dummy argument

[pault at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60717

Paul Thomas <pault at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-03-31
                 CC|                            |pault at gcc dot gnu.org
           Assignee|unassigned at gcc dot gnu.org      |pault at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #1 from Paul Thomas <pault at gcc dot gnu.org> ---
Confirmed. You beat me to it by about 30 minutes, Vladimir!

The problem is the recursive call within 'show_generic':

            D.3571 = a->_data.dim[0].ubound;
            parm.2.dtype = 345;
            D.3575 = a->_data.dim[0].stride;
            parm.2.dim[0].lbound = 1;
            parm.2.dim[0].ubound = D.3571 + -1;
            parm.2.dim[0].stride = NON_LVALUE_EXPR <D.3575>;
            parm.2.data = a->_data.data + (sizetype) (a->_vptr->_size *
NON_LVALUE_EXPR <D.3575>);
            parm.2.offset = 0;
            class.3._data = VIEW_CONVERT_EXPR<struct array1_unknown>(parm.2);
            class.3._vptr = a->_vptr;
            show_generic (&class.3);

The offset is 0, whereas it should be -1.

The call to 'show_generic' from the main program is done correctly:

    parm.17.dtype = 281;
    parm.17.dim[0].lbound = 1;
    parm.17.dim[0].ubound = 6;
    parm.17.dim[0].stride = 1;
    parm.17.data = (void *) &array[0];
    parm.17.offset = -1;
    class.16._data = parm.17;
    show_generic (&class.16);

This tells us exactly where the problem is.

I am taking it.

Paul

[Bug fortran/60717] Wrong code with recursive procedure with unlimited polymorphic dummy argument

[pault at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60717

--- Comment #2 from Paul Thomas <pault at gcc dot gnu.org> ---
Created attachment 32509
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32509&action=edit
Patch for the PR

This patch bootstraps and regtests OK. I will tidy it up a bit tonight.

I also wish to check if it can be used to clean up:

trans-expr.c(gfc_trans_alloc_subarray_assign)
trans-expr.c(gfc_trans_pointer_assign)
trans-expr.c(fncall_realloc_result)
trans-array.c(trans_associate_var)

each of which contains calculation of the offset.

Paul

[Bug fortran/60717] Wrong code with recursive procedure with unlimited polymorphic dummy argument

[vladimir.fuka at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60717

--- Comment #3 from Vladimir Fuka <vladimir.fuka at gmail dot com> ---
Could there be a connection http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58085 ?

[Bug fortran/60717] Wrong code with recursive procedure with unlimited polymorphic dummy argument

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60717

--- Comment #4 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> Could there be a connection http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58085 ?

May be, but pr58085 is not fixed by the patch in comment 2 (submitted at
http://gcc.gnu.org/ml/fortran/2014-04/msg00005.html).

[Bug fortran/60717] Wrong code with recursive procedure with unlimited polymorphic dummy argument

[pault at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60717

--- Comment #6 from Paul Thomas <pault at gcc dot gnu.org> ---
Author: pault
Date: Sun Apr 13 11:55:49 2014
New Revision: 209346

URL: http://gcc.gnu.org/viewcvs?rev=209346&root=gcc&view=rev
Log:
2014-04-13  Paul Thomas  <pault@gcc.gnu.org>

    PR fortran/58085
    PR fortran/60717
    * trans.h: Add 'use_offset' bitfield to gfc_se.
    * trans-array.c (gfc_conv_expr_descriptor): Use 'use_offset'
    as a trigger to unconditionally recalculate the offset for
    array slices and constant arrays.
    trans-expr.c (gfc_conv_intrinsic_to_class): Use it.
    trans-stmt.c (trans_associate_var): Ditto.
    (gfc_conv_procedure_call): Ditto.

2014-04-13  Paul Thomas  <pault@gcc.gnu.org>

    PR fortran/60717
    * gfortran.dg/unlimited_polymorphic_17.f90: New test.

    PR fortran/58085
    * gfortran.dg/associate_15.f90: New test.

Added:
    branches/gcc-4_9-branch/gcc/testsuite/gfortran.dg/associate_15.f90
   
branches/gcc-4_9-branch/gcc/testsuite/gfortran.dg/unlimited_polymorphic_17.f90
Modified:
    branches/gcc-4_9-branch/gcc/fortran/ChangeLog
    branches/gcc-4_9-branch/gcc/fortran/trans-array.c
    branches/gcc-4_9-branch/gcc/fortran/trans-expr.c
    branches/gcc-4_9-branch/gcc/fortran/trans-stmt.c
    branches/gcc-4_9-branch/gcc/fortran/trans.h
    branches/gcc-4_9-branch/gcc/testsuite/ChangeLog

[Bug fortran/60717] Wrong code with recursive procedure with unlimited polymorphic dummy argument

[pault at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60717

--- Comment #7 from Paul Thomas <pault at gcc dot gnu.org> ---
Author: pault
Date: Sun Apr 13 11:58:55 2014
New Revision: 209347

URL: http://gcc.gnu.org/viewcvs?rev=209347&root=gcc&view=rev
Log:
2014-04-13  Paul Thomas  <pault@gcc.gnu.org>

    PR fortran/58085
    PR fortran/60717
    * trans.h: Add 'use_offset' bitfield to gfc_se.
    * trans-array.c (gfc_conv_expr_descriptor): Use 'use_offset'
    as a trigger to unconditionally recalculate the offset for
    array slices and constant arrays.
    trans-expr.c (gfc_conv_intrinsic_to_class): Use it.
    trans-stmt.c (trans_associate_var): Ditto.
    (gfc_conv_procedure_call): Ditto.

2014-04-13  Paul Thomas  <pault@gcc.gnu.org>

    PR fortran/60717
    * gfortran.dg/unlimited_polymorphic_17.f90: New test.

    PR fortran/58085
    * gfortran.dg/associate_15.f90: New test.

Added:
    trunk/gcc/testsuite/gfortran.dg/associate_15.f90
    trunk/gcc/testsuite/gfortran.dg/unlimited_polymorphic_17.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/trans-array.c
    trunk/gcc/fortran/trans-expr.c
    trunk/gcc/fortran/trans-stmt.c
    trunk/gcc/fortran/trans.h
    trunk/gcc/testsuite/ChangeLog

[Bug fortran/60717] Wrong code with recursive procedure with unlimited polymorphic dummy argument

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60717

Paul Thomas <pault at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #8 from Paul Thomas <pault at gcc dot gnu.org> ---
It seems that I forgot to close this one.

Thanks for the report

Paul