[Bug fortran/52351] New: Wrong bounds when passing an array section to an intent-in pointer dummy

[Bug fortran/52351] New: Wrong bounds when passing an array section to an intent-in pointer dummy

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52351

             Bug #: 52351
           Summary: Wrong bounds when passing an array section to an
                    intent-in pointer dummy
    Classification: Unclassified
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: burnus@gcc.gnu.org


The following program should print:
           3           6
           1           4
but it wrongly prints
           3           6
           3           6


The program uses the Fortran 2008 feature that one can pass a nonpointer to an
intent(in) pointer argument.


>From the standard:

"12.5.2.7 Pointer dummy variables"
"If the dummy argument does not have the INTENT (IN), the actual argument shall
be a pointer. Otherwise, the actual argument shall be a pointer or a valid
target for the dummy pointer in a pointer assignment statement. If the actual
argument is not a pointer, the dummy pointer becomes pointer associated with
the actual argument."

"16.5.2 Pointer association"
"If the pointer has deferred type parameters or shape, their values are assumed
from the target."


As lbound(a(:)) == 1 (always) and lbound(a) == 3 (in this example), the
respective result should be obtained for lbound(x).



integer, allocatable, target :: A(:)
allocate(A(3:6))

call sub (a)
call sub (a(:))

contains

  subroutine sub(x)
    integer, pointer, intent(in) :: x(:)
    print *, lbound(x), ubound(x)
  end subroutine sub
end

[Bug fortran/52351] Wrong bounds when passing an array section to an intent-in pointer dummy

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52351

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

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

--- Comment #1 from Tobias Burnus <burnus at gcc dot gnu.org> 2012-02-23 10:49:22 UTC ---
It works if one any other array section than "(:)" such as:
  call sub (a(4:))  ! gives:  1  3
or even
  call sub (a(3:))  ! gives:  1  4

[Bug fortran/52351] Wrong bounds when passing an array section to an intent-in pointer dummy

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52351

--- Comment #2 from Tobias Burnus <burnus at gcc dot gnu.org> 2012-02-23 22:34:31 UTC ---
The issue seems to be in gfc_conv_array_parameter - or rather in the called
gfc_conv_expr_descriptor. There, one has:

        full = gfc_full_array_ref_p (info->ref, NULL);
      if (full && !transposed_dims (ss))
          else if (se->want_pointer)
              se->expr = gfc_build_addr_expr (NULL_TREE, desc);
          return;

However, that's wrong if one needs to modify the bounds. Thus, in this special
case (fsym == pointer/class_pointer && fsym->attr.intent == INTENT_IN plus last
ref != AR_FULL), one has to create a new descriptor.

[Bug fortran/52351] Wrong bounds when passing an array section to an intent-in pointer dummy

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52351

Dominique d'Humieres <dominiq at lps dot ens.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2013-06-20
     Ever confirmed|0                           |1

--- Comment #3 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
Still present at revision 200247.

[Bug fortran/52351] Wrong bounds when passing an array section to an intent-in pointer dummy

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=52351

--- Comment #5 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Thomas Kथà¤nig <tkoenig@gcc.gnu.org>:

https://gcc.gnu.org/g:2ff0f48819c8a7ed5d7c03e2bfc02e5907e2ff1a

commit r11-1230-g2ff0f48819c8a7ed5d7c03e2bfc02e5907e2ff1a
Author: José Rui Faustino de Sousa <jrfsousa@gmail.com>
Date:   Thu Jun 11 14:14:30 2020 +0200

    Wrong array section bounds when passing to an intent-in pointer dummy.

    Add code to allow for the creation a new descriptor for array
    sections with the correct one based indexing.

    Rework the generated descriptors indexing (hopefully) fixing the
    wrong offsets generated.

    gcc/fortran/ChangeLog:

    2020-06-11  José Rui Faustino de Sousa  <jrfsousa@gmail.com>

            PR fortran/52351
            PR fortran/85868
            * trans-array.c (gfc_conv_expr_descriptor): Enable the
            creation of a new descriptor with the correct one based
            indexing for array sections.  Rework array descriptor
            indexing offset calculation.

    gcc/testsuite/ChangeLog:

    2020-06-11  José Rui Faustino de Sousa  <jrfsousa@gmail.com>

            PR fortran/52351
            PR fortran/85868
            * gfortran.dg/coarray_lib_comm_1.f90: Adjust match test for
            the newly generated descriptor.
            * gfortran.dg/PR85868A.f90: New test.
            * gfortran.dg/PR85868B.f90: New test.

[Bug fortran/52351] Wrong bounds when passing an array section to an intent-in pointer dummy

[tkoenig at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=52351

Thomas Koenig <tkoenig at gcc dot gnu.org> changed:

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

--- Comment #6 from Thomas Koenig <tkoenig at gcc dot gnu.org> ---
Fixed on master.