[PATCH] ld: Ignore the new weak definition if needed

[H.J. Lu]

On Mon, Jul 20, 2020 at 4:17 PM Alan Modra <amodra@gmail.com> wrote:
>
> On Mon, Jul 20, 2020 at 06:34:29AM -0700, H.J. Lu via Binutils wrote:
> > --- a/bfd/elflink.c
> > +++ b/bfd/elflink.c
> > @@ -1269,6 +1269,18 @@ _bfd_elf_merge_symbol (bfd *abfd,
> >           && h->root.type != bfd_link_hash_undefweak
> >           && h->root.type != bfd_link_hash_common);
> >
> > +  /* NB: Ignore the new weak definition if the old definition comes
> > +     from the LTO IR object since plugin_notice will turn it into
> > +     undefweak.  */
> > +  if (olddef
> > +      && oldbfd
> > +      && (oldbfd->flags & BFD_PLUGIN) != 0
> > +      && newweak)
> > +    {
> > +      *skip = TRUE;
> > +      return TRUE;
> > +    }
> > +
> >    /* NEWFUNC and OLDFUNC indicate whether the new or old symbol,
> >       respectively, appear to be a function.  */
> >
>
> How do you square this with later code in _bfd_elf_merge_symbol?
> See PR12696.
>
>   /* Skip weak definitions of symbols that are already defined.  */
>   if (newdef && olddef && newweak)
>     {
>       /* Don't skip new non-IR weak syms.  */
>       if (!(oldbfd != NULL
>             && (oldbfd->flags & BFD_PLUGIN) != 0
>             && (abfd->flags & BFD_PLUGIN) == 0))
>         {
>           newdef = FALSE;
>           *skip = TRUE;
>         }
>

My patch is wrong.


-- 
H.J.