[RFA:] ar test failure fix

[Hans-Peter Nilsson]

> Date: Wed, 8 Jun 2005 12:45:43 +0930
> From: Alan Modra <amodra@bigpond.net.au>

> On Wed, Jun 08, 2005 at 05:01:57AM +0200, Hans-Peter Nilsson wrote:
> > My patch is for bfd_fopen which only did that on error, not bfd_fdopenr.
> > 
> > So there *is* a latent bug.
> 
> Well, there was.  I'm applying this fix instead of yours.
> 
> 	* opncls.c (bfd_fopen): Don't set bfd_error unconditionally.

But not the same in bfd_fdopenr?

It it was me, I'd save the previous bfd_error at function start,
and restore before successful exit.

brgds, H-P