[Bug fortran/94660] New: Wrong subroutine called at runtime than the one called in the source code with deferred subroutines

[Bug fortran/94660] New: Wrong subroutine called at runtime than the one called in the source code with deferred subroutines

[innspg at yahoo dot fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94660

            Bug ID: 94660
           Summary: Wrong subroutine called at runtime than the one called
                    in the source code with deferred subroutines
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: innspg at yahoo dot fr
  Target Milestone: ---

[Bug fortran/94660] Wrong subroutine called at runtime than the one called in the source code with deferred subroutines

[innspg at yahoo dot fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94660

innspg at yahoo dot fr changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |innspg at yahoo dot fr

--- Comment #1 from innspg at yahoo dot fr ---
Created attachment 48306
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=48306&action=edit
tar files of fortran Source files that reproduce the bug

The attachment is a tar file containing 4 fortran sources codes to reproduce
the bug.

gfortran 4.8.5 produces the expected result
All other versions that I tried produced the wrong results; those versions are:
6.3.0, 7.3.0, 7.5.0, 8.2.0, 8.4.0 and 9.2.0.


The expected output when the program is ran should be:

Test: calling C%init
  B::init, calling setup
    C::setup
  B::init done ...............
Test: calling C%finalize
  B::finalize, calling free_resources
    C::free_resources
  B::finalize, calling cleanup
  B::cleanup
  B::finalize done ...................
Test:done.......................

I am getting this instead:

Test: calling C%init
  B::init, calling setup
  B::cleanup
  B::init done ...............
Test: calling C%finalize
  B::finalize, calling free_resources
    C::setup
  B::finalize, calling cleanup
  B::cleanup
  B::finalize done ...................
Test:done.......................

[Bug fortran/94660] Wrong subroutine called at runtime than the one called in the source code with deferred subroutines

[innspg at yahoo dot fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94660

--- Comment #2 from innspg at yahoo dot fr ---
The bug was reproduced with the following version of gfortran: 6.3.0, 7.3.0,
7.5.0, 8.2.0, 8.4.0 and 9.2.0.

gfortran 4.8.5 produces the expected result


The expected output when the program is ran should be:

Test: calling C%init
  B::init, calling setup
    C::setup
  B::init done ...............
Test: calling C%finalize
  B::finalize, calling free_resources
    C::free_resources
  B::finalize, calling cleanup
  B::cleanup
  B::finalize done ...................
Test:done.......................

I am getting this instead:

Test: calling C%init
  B::init, calling setup
  B::cleanup
  B::init done ...............
Test: calling C%finalize
  B::finalize, calling free_resources
    C::setup
  B::finalize, calling cleanup
  B::cleanup
  B::finalize done ...................
Test:done.......................

[Bug fortran/94660] Wrong subroutine called at runtime than the one called in the source code with deferred subroutines

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94660

kargl at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2020-04-19
                 CC|                            |kargl at gcc dot gnu.org
             Status|UNCONFIRMED                 |NEW
      Known to work|                            |4.8.5
     Ever confirmed|0                           |1
      Known to fail|                            |6.3.0, 7.3.0, 7.5.0, 8.2.0,
                   |                            |8.4.0, 9.2.0

[Bug fortran/94660] Wrong subroutine called at runtime than the one called in the source code with deferred subroutines

[tkoenig at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94660

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |WAITING
                 CC|                            |tkoenig at gcc dot gnu.org

--- Comment #3 from Thomas Koenig <tkoenig at gcc dot gnu.org> ---
I just ran this through nagfor, and it agrees with gfortran.
Nagfor is usually quite picky.

xlf, on the other hand, agrees with you, so it does not seem
to be straightforward, at least.

This will take some further analysis.

[Bug fortran/94660] Wrong subroutine called at runtime than the one called in the source code with deferred subroutines

[gnu.gcc.detection675 at passmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94660

Jeff Hole <gnu.gcc.detection675 at passmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |gnu.gcc.detection675@passma
                   |                            |il.com

--- Comment #4 from Jeff Hole <gnu.gcc.detection675 at passmail dot com> ---
Created attachment 56807
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=56807&action=edit
clue: combining modules into a single file gives expected results

Here's an interesting clue: the correct results occur when all the modules are
stored in the same source file (note, all modules are still separate, just in
the same file).

This is certainly not a solution, but it does provide a clue to the problem, as
well as and a way to get things working the way you want. Ultimately, it is
necessary that this works across separate module source files so that we
separate compilation steps and enable extensions of these types beyond the
single source file.

I compiled with gfortran:
    GNU Fortran (GCC) 10.3.1 20210422 (Red Hat 10.3.1-1)

I attached tarball "buginnspg-clue.tar".