[Bug fortran/98573] New: Dynamic type lost on assignment

[Bug fortran/98573] New: Dynamic type lost on assignment

[davidhneill at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98573

            Bug ID: 98573
           Summary: Dynamic type lost on assignment
           Product: gcc
           Version: 9.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: davidhneill at gmail dot com
  Target Milestone: ---

Created attachment 49904
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49904&action=edit
Minimal reproducer

In the following example, the dynamic type of an unlimited polymorphic array is
lost on an assignment but conserved (as expected) on a sourced allocation.

This happens in all versions I've tried: 7.5.0, 8.4.0, 9.3.0, 10.2.0

My system is Ubuntu 20.04 on Intel hardware with all gfortran versions
installed directly from the Ubuntu repositories.

$ cat type_lost.f90

module foo
  type, public:: box
    class(*), allocatable :: val(:)
  end type
contains
  subroutine store1(this, val)
    class(box), intent(out) :: this
    class(*), intent(in) :: val(:)
    this%val = val
  end subroutine store1
  subroutine store2(this, val)
    class(box), intent(out) :: this
    class(*), intent(in) :: val(:)
    allocate(this%val, source=val)
  end subroutine store2
  subroutine vector_type(val)
    class(*), intent(in) :: val(:)
    select type (val)
    type is (integer)
      print '("INTEGER")'
    class default
      print '("OTHER")'
    end select
  end subroutine vector_type
end module foo

program prog
  use foo
  type(box) :: b
  call store1(b, [1, 2, 3])
  call vector_type(b%val)  ! OTHER
  call store2(b, [1, 2, 3])
  call vector_type(b%val)  ! INTEGER

end program

$ gfortran -g -Wall -Wextra minimal.f90
$ ./a.out
OTHER
INTEGER

[Bug fortran/98573] Dynamic type lost on assignment

[davidhneill at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98573

--- Comment #1 from David Neill Asanza <davidhneill at gmail dot com> ---
Removing the derived type container triggers a segfault.

$ cat segfault.f90

module foo
contains
  subroutine store1(arr, val)
    class(*), allocatable, intent(out) :: arr(:)
    class(*), intent(in) :: val(:)
    arr = val
  end subroutine store1
  subroutine store2(arr, val)
    class(*), allocatable, intent(out) :: arr(:)
    class(*), intent(in) :: val(:)
    allocate(arr, source=val)
  end subroutine store2
end module foo

program prog
  use foo
  class(*), allocatable :: arr(:)
  call store1(arr, [1, 2, 3])  ! SEGFAULT
  call store2(arr, [1, 2, 3])  ! NO PROBLEM
end program


$ gfortran -g -Wall -Wextra minimal.f90
$ ./a.out

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

[Bug fortran/98573] Dynamic type lost on assignment

[davidhneill at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98573

--- Comment #2 from David Neill Asanza <davidhneill at gmail dot com> ---
Created attachment 49905
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49905&action=edit
Minimal reproducer for segmentation fault.

[Bug fortran/98573] Dynamic type lost on assignment

[davidhneill at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98573

--- Comment #3 from David Neill Asanza <davidhneill at gmail dot com> ---
A related case when the array is 0-length. In this case, the dynamic type is
lost in both assignment and sourced allocation.

$ cat type_lost_0_length.f90

module foo
  type, public:: box
    class(*), allocatable :: val(:)
  end type
contains
  subroutine store1(this, val)
    class(box), intent(out) :: this
    class(*), intent(in) :: val(:)
    this%val = val
  end subroutine store1
  subroutine store2(this, val)
    class(box), intent(out) :: this
    class(*), intent(in) :: val(:)
    allocate(this%val, source=val)
  end subroutine store2
  subroutine vector_type(val)
    class(*), intent(in) :: val(:)
    select type (val)
    type is (integer)
      print '("INTEGER")'
    class default
      print '("OTHER")'
    end select
  end subroutine vector_type
end module foo

program prog
  use foo
  type(box) :: b
  integer, allocatable :: arr1(:)
  integer, dimension(0) :: arr2

  allocate(arr1(0))
  call store1(b, arr1)
  call vector_type(b%val)  ! OTHER
  call store2(b, arr1)
  call vector_type(b%val)  ! OTHER

  call store1(b, arr2)
  call vector_type(b%val)  ! OTHER
  call store2(b, arr2)
  call vector_type(b%val)  ! OTHER
end program

$ gfortran -g -Wall -Wextra type_lost_0_length.f90
$ ./a.out
OTHER
OTHER
OTHER
OTHER

[Bug fortran/98573] Dynamic type lost on assignment

[davidhneill at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98573

--- Comment #4 from David Neill Asanza <davidhneill at gmail dot com> ---
Created attachment 49906
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49906&action=edit
Reproducer for 0-length case.

[Bug fortran/98573] Dynamic type lost on assignment

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98573

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pault at gcc dot gnu.org
           Assignee|unassigned at gcc dot gnu.org      |pault at gcc dot gnu.org

--- Comment #5 from Paul Thomas <pault at gcc dot gnu.org> ---
Created attachment 50023
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=50023&action=edit
Fix for all three tests

I believe that attached does the job although I don't quite see as many failing
cases as you.

It regtests too :-)

I will investigate a bit further before I submit.

Thanks for the report.

Paul

[Bug fortran/98573] Dynamic type lost on assignment

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98573

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2021-01-21
     Ever confirmed|0                           |1

[Bug fortran/98573] Dynamic type lost on assignment

[davidhneill at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98573

--- Comment #6 from David Neill Asanza <davidhneill at gmail dot com> ---
Thanks for looking into this Paul. 

I'm looking forward to having this fixed. :)

[Bug fortran/98573] Dynamic type lost on assignment

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98573

--- Comment #7 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
A patch has been submitted at

https://gcc.gnu.org/pipermail/fortran/2021-January/055624.html

and approved at

https://gcc.gnu.org/pipermail/fortran/2021-January/055644.html

More comments at

https://gcc.gnu.org/pipermail/fortran/2021-January/055651.html