[Bug fortran/86100] Spurious error with -fcheck=bounds and allocatable class(*) array components

["anlauf at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86100

anlauf at gcc dot gnu.org changed:

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

--- Comment #2 from anlauf at gcc dot gnu.org ---
The problem seems to generally occur for class components of derived types.

program p
  implicit none
  type t
     integer :: i = 0
  end type t
  type any_matrix
     class(t), allocatable :: m(:,:)
  end type any_matrix
  type(any_matrix) :: a, b
  allocate(a%m(3,4))
  call foo ()
contains
  subroutine foo ()
    b = a      ! Array bound mismatch for dimension 1 of array '<<unknown>>'
(12/3)
    !b%m = a%m ! no runtime error with -fcheck=bounds
  end subroutine foo
end

This fails as indicated.

Note that 12=3*4 is the total size of the array, and 3 is the size of the
first dimension, which - along with the '<<unknown>>' - points to
gfc_copy_class_to_class:

          from_len = gfc_conv_descriptor_size (from_data, 1);
          from_len = fold_convert (TREE_TYPE (orig_nelems), from_len);
          tmp = fold_build2_loc (input_location, NE_EXPR,
                                  logical_type_node, from_len, orig_nelems);
          msg = xasprintf ("Array bound mismatch for dimension %d "
                           "of array '%s' (%%ld/%%ld)",
                           1, name);

          gfc_trans_runtime_check (true, false, tmp, &body,
                                   &gfc_current_locus, msg,
                             fold_convert (long_integer_type_node,
orig_nelems),
                               fold_convert (long_integer_type_node,
from_len));

So we compare dimension 1 and the full size of the rhs?
Shouldn't we compare lhs and rhs shapes?