Re: [PATCH] calls: Introduce cxx17_empty_base_field_p [PR94383]

[Jakub Jelinek via Gcc-patches]

On Wed, Apr 22, 2020 at 12:17:02PM +0100, Richard Sandiford wrote:
> But my point was that, if the DECL_NAME does actually act to exclude

I'm fine with dropping DECL_NAME test there, on the other side would like to
add
  && TYPE_SIZE (TREE_TYPE (field))
  && !integer_zerop (TYPE_SIZE (TREE_TYPE (field)))
in there because that is what all these empty bases should satisfy too.

If the predicate says that it is the C++17 empty base field, then it better
should be a narrow check.
Now, in the backend, one has the -Wpsabi diagnostics that also talks about
C++17, so I think it is better to use that predicate in there.
But, what one could do is verify that all other
if (DECL_SIZE (field) && integer_zerop (DECL_SIZE (field)))
fields don't matter for the ABI decisions.
So
            if (TREE_CODE (field) != FIELD_DECL)
              continue;

            if (cxx17_empty_base_field_p (field))
              {
                /* hint -Wpsabi warning here somehow.  */
                continue;
              }
   
            sub_count = aapcs_vfp_sub_candidate (TREE_TYPE (field), modep);

            /* Verify that other zero sized fields don't affect the
               ABI decisions.  */
            if (DECL_SIZE (field) && integer_zerop (DECL_SIZE (field)))
              gcc_assert (sub_count == 0);

            if (sub_count < 0)
              return -1;
            count += sub_count;
?

        Jakub