This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: PATCH: PR ld/4409: --unresolved-symbols=ignore-all issues on ia64
On Thu, May 10, 2007 at 06:04:03AM -0700, H. J. Lu wrote:
> On Thu, May 10, 2007 at 02:15:39PM +0930, Alan Modra wrote:
> > On Wed, May 09, 2007 at 09:14:20PM -0700, H. J. Lu wrote:
> > > When an error from RELOC_FOR_GLOBAL_SYMBOL in executable is ignored
> > > on ia64, we shouldn't continue since unresolved symbol in executable
> > > is fatal on ia64. This patch fixes it.
> > >
> >
> > This would have been a lot less typing if you'd just changed
> > elfxx-ia64.c. The following does the same as your patch, I think.
> >
> > Index: bfd/elfxx-ia64.c
> > ===================================================================
> > RCS file: /cvs/src/src/bfd/elfxx-ia64.c,v
> > retrieving revision 1.198
> > diff -u -p -r1.198 elfxx-ia64.c
> > --- bfd/elfxx-ia64.c 26 Apr 2007 14:46:57 -0000 1.198
> > +++ bfd/elfxx-ia64.c 10 May 2007 04:39:41 -0000
> > @@ -4672,7 +4672,9 @@ elfNN_ia64_relocate_section (output_bfd,
> >
> > if (h->root.type == bfd_link_hash_undefweak)
> > undef_weak_ref = TRUE;
> > - else if (warned)
> > + else if (h->root.type != bfd_link_hash_defined
> > + && h->root.type != bfd_link_hash_defweak
> > + && (warned || info->executable))
> > continue;
> > }
> >
>
> They aren't the same for symbols with non-default visibility. We can
Yeah, I got it wrong.
> certainly copy the same logic to elfNN_ia64_relocate_section. But is
> this necessary given that I already did the typing :-(?
I'll defer that decision to Jim. I don't really know enough about
ia64 to OK the patch.
--
Alan Modra
IBM OzLabs - Linux Technology Centre