Hi,
On Fri, May 10, 2013 at 07:24:06PM +0200, Jan Hubicka wrote:
> > 2013-05-10 Martin Jambor <mjam...@suse.cz>
> >
> > * tree.c (get_binfo_at_offset): Use types_compatible_p to compare
> > types.
> >
> > Index: src/gcc/tree.c
> > ===================================================================
> > --- src.orig/gcc/tree.c
> > +++ src/gcc/tree.c
> > @@ -11483,7 +11483,7 @@ get_binfo_at_offset (tree binfo, HOST_WI
> > tree fld;
> > int i;
> >
> > - if (TYPE_MAIN_VARIANT (type) == TYPE_MAIN_VARIANT (expected_type))
> > + if (types_compatible_p (type, expected_type))
> As discussed on lunch, it seems fine to me, but I am not an expert ;) I
> wonder about
> the following test few lines bellow:
> /* Offset 0 indicates the primary base, whose vtable contents are
> represented in the binfo for the derived class. */
> else if (offset != 0)
> {
> tree base_binfo, found_binfo = NULL_TREE;
> for (i = 0; BINFO_BASE_ITERATE (binfo, i, base_binfo); i++)
> if (TREE_TYPE (base_binfo) == TREE_TYPE (fld))
>
> don't we want also more relaxed testing?
we most probably do, good catch. I'll update the patch, re-test and
commit it early next week unless someone objects.
>
> "grep "TYPE.* ==.*TYPE " *.c" seems to show some extra dubious cases i.e.
> in fold-const, tree-ssa-coaelesce and other places where we most probably care
> only about semantical equivalence of the type...
>
Yeah, examinig those is also probably a good idea.
Thanks,
Martin
