https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80605
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jason at gcc dot gnu.org
--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
check_bases does:
/* ...either has no non-static data members in the most-derived
class and at most one base class with non-static data
members, or has no base classes with non-static data
members */
for (basefield = TYPE_FIELDS (basetype); basefield;
basefield = DECL_CHAIN (basefield))
if (TREE_CODE (basefield) == FIELD_DECL)
{
if (field)
CLASSTYPE_NON_STD_LAYOUT (t) = 1;
else
field = basefield;
break;
}
and we hit this with field being x and basefield being the artificial
FIELD_DECL created from:
4615 /* We used to not create a FIELD_DECL for empty base classes
because of
4616 back end issues with overlapping FIELD_DECLs, but that doesn't
seem to
4617 be a problem anymore. We need them to handle initialization
of C++17
4618 aggregate bases. */
4619 if (cxx_dialect >= cxx1z && !BINFO_VIRTUAL_P (binfo))
4620 {
4621 tree decl = build_base_field_1 (t, basetype, next_field);
4622 DECL_FIELD_OFFSET (decl) = BINFO_OFFSET (binfo);
4623 DECL_FIELD_BIT_OFFSET (decl) = bitsize_zero_node;
4624 SET_DECL_OFFSET_ALIGN (decl, BITS_PER_UNIT);
4625 }
So, do we need to ignore those artificial FIELD_DECLs in the
CLASSTYPE_NON_STD_LAYOUT computation? Do we have some way how to identify the
FIELD_DECLs created in this case?
