v4.1.x: Fix inconsistent configure-time check

[skosukhin]

This fixes the inconsistency between what is tested at the configure time and what is actually used in the source code. Note that the issue was solved in the main with #9445. As far as I understand, there are no plans to backport that PR to 4.1.x because it potentially changes the ABI compatibility. Therefore, this PR fixes the problem the other way around.

Starting version 23.3, NVHPC compiler supports type(*) and the configure script sets ompi_cv_fortran_ignore_tkr_data to 1:type(*), dimension(*):!GCC$ ATTRIBUTES NO_ARG_CHECK ::, which makes the mpi Fortran module unusable in some cases. For example, we cannot compile the following:

program main
  use mpi
  implicit none

  real(kind=selected_real_kind(12, 307)):: buf
  integer :: count, datatype, dest, tag, comm, ierror

  call mpi_send(buf, count, datatype, dest, tag, comm, ierror)
end program main

The compiler reports:

NVFORTRAN-S-0155-Could not resolve generic procedure mpi_send (./test2.f90: 8)
  0 inform,   0 warnings,   1 severes, 0 fatal for main

As was mentioned in #9445, the unnamed interfaces are more forgiving when it comes to TKR mismatches. Indeed, the code snippet run by the configure script uses the unnamed interface and is too weak for the following use case. This PR makes the configure-time check stricter by switching to the named interface, which is what is actually used in the code. With this change, the configure script sets ompi_cv_fortran_ignore_tkr_data to 1:real, dimension(*):!DIR$ IGNORE_TKR when using NVHPC 23.3+ (the same value as for the older versions of the compiler) and the code snipped above compiles successfully.

bot:notacherrypick

[github-actions]

Hello! The Git Commit Checker CI bot found a few problems with this PR:

c3fbe69: Fix inconsistent configure-time check

• check_signed_off: does not contain a valid Signed-off-by line
• check_cherry_pick: does not include a cherry pick message (did you need to bot:notacherrypick?)

Please fix these problems and, if necessary, force-push new commits back up to the PR branch. Thanks!

[github-actions]

Hello! The Git Commit Checker CI bot found a few problems with this PR:

044adbe: Fix inconsistent configure-time check

• check_cherry_pick: does not include a cherry pick message (did you need to bot:notacherrypick?)

Please fix these problems and, if necessary, force-push new commits back up to the PR branch. Thanks!

[github-actions]

Hello! The Git Commit Checker CI bot found a few problems with this PR:

0f8e73b: Fix inconsistent configure-time check

• check_cherry_pick: does not include a cherry pick message (did you need to bot:notacherrypick?)

Please fix these problems and, if necessary, force-push new commits back up to the PR branch. Thanks!

[github-actions]

Hello! The Git Commit Checker CI bot found a few problems with this PR:

0f8e73b: Fix inconsistent configure-time check

• check_cherry_pick: does not include a cherry pick message (did you need to bot:notacherrypick?)

Please fix these problems and, if necessary, force-push new commits back up to the PR branch. Thanks!

[jsquyres]

@skosukhin Can you add more information to the commit message about the "what" and "why" this commit is necessary? You put a lot of information in the PR body, which is great -- at least some of that should probably also be in the git commit message.

I understand your rationale for why this is not a cherry pick; you need to put the notacherrypick bot token in the PR description, not the git commit message. Then the CI bot will pass.

[github-actions]

Hello! The Git Commit Checker CI bot found a few problems with this PR:

0f8e73b: Fix inconsistent configure-time check

• check_cherry_pick: does not include a cherry pick message (did you need to bot:notacherrypick?)

Please fix these problems and, if necessary, force-push new commits back up to the PR branch. Thanks!

[jsquyres]

Thank you!