Re: PATCH: PR 233: Check non-debug local references to discarded sections

["H. J. Lu" <hjl at lucon dot org>]

On Wed, Jun 30, 2004 at 01:25:07PM +0930, Alan Modra wrote:
> On Tue, Jun 29, 2004 at 09:13:16AM -0700, H. J. Lu wrote:
> > @@ -6277,6 +6277,10 @@ elf_section_ignore_discarded_relocs (ase
> >        && (*bed->elf_backend_ignore_discarded_relocs) (sec))
> >      return TRUE;
> >  
> > +  /* Catch all .eh_frame sections.  */
> > +  if (strcmp (sec->name, ".eh_frame") == 0)
> > +    return TRUE;
> > +
> >    return FALSE;
> >  }
> 
> Huh?  Why wasn't sec->sec_info_type == ELF_INFO_TYPE_EH_FRAME?

Because some ld tests have some bogus .eh_frame sections which won't
have it set.

> 
> > @@ -6593,9 +6596,11 @@ elf_link_input_bfd (struct elf_final_lin
> >  			  else
> >  			    finfo->info->callbacks->error_handler
> >  			      (LD_DEFINITION_IN_DISCARDED_SECTION,
> > -			       _("%T: discarded in section `%s' from %s\n"),
> > +			       _("`%T' referenced in section `%s' from %s: discarded in section `%s' from %s\n"),
> >  			       h->root.root.string,
> >  			       h->root.root.string,
> > +			       o->name,
> > +			       bfd_archive_filename (o->owner),
> >  			       h->root.u.def.section->name,
> >  			       bfd_archive_filename (h->root.u.def.section->owner));
> >  			}
> 
> You can't call bfd_archive_filename twice in the one error message,
> as it uses a static buffer.
> 

Here is the new one.


H.J.

Attachment: binutils-linkonce-local-2.patch
Description: Text document