Re: [patch] Honour SHF_LINK_ORDER

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

On Thu, Jul 22, 2004 at 01:07:54PM +0100, Nick Clifton wrote:
> Hi H. J.
> 
> >>I suspect this means elf_fixup_link_order returned FALSE. Looking at it 
> >>again, this should probably issue an error/warning when this happens.
> 
> >Here is a patch.
> 
> Approved - please apply.
> 
> One question though: Is the IA64 toolchain broken because it is 
> producing both ordered and unordered sections, or is it a feature of the 
> toolchain and the problem is just that the BFD library does not support 
> this feature (yet) ?

The test is invalid. The linker script:

NOCROSSREFS ( .text .data )
SECTIONS
{
  .dynsym : { *(.dynsym) }
  .dynstr : { *(.dynstr) }
  .hash : { *(.hash) }
  .toc  : { *(.toc) }
  .opd  : { *(.opd) }
  .text : { tmpdir/cross1.o }
  .data : { tmpdir/cross2.o }
}

will fail on any targets which use SHF_LINK_ORDER since

  .text : { tmpdir/cross1.o }
  .data : { tmpdir/cross2.o }

may put both ordered and unordered sections into .text or .data.

H.J.