Re: [RFC] Providing init_fini_syms earlier?

[Alan Modra <amodra at bigpond dot net dot au>]

On Wed, Jul 13, 2005 at 11:02:11PM +0930, Alan Modra wrote:
> On Tue, Jul 12, 2005 at 05:47:12PM -0700, H. J. Lu wrote:
> > The final values have to be absulte.
> 
> Why?
> 
> > How does it support relaxation
> > when it will be called more then once?
> 
> If they are section relative, then there is no problem with relaxation
> unless the init or fini sections change size, which I think is
> unlikely to happen.  There is one disadvantage of section relative syms;
> They cause their output section to be kept.  Hmm, I suppose we could
> change that.  We might even be able to do without bfd_mark_used_section
> entirely now that vma is set correctly for stripped output sections.

Oh, except you would need to convert the syms to absolute if their
output section was stripped.  Was that why you were saying they needed
to be absolute?

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre