This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: [PATCH/RFA] Ambiguity problem of REL objects on ELF sh*-*-*
- From: Alan Modra <amodra at bigpond dot net dot au>
- To: kaz Kojima <kkojima at rr dot iij4u dot or dot jp>
- Cc: binutils at sources dot redhat dot com
- Date: Fri, 1 Nov 2002 20:58:12 +1030
- Subject: Re: [PATCH/RFA] Ambiguity problem of REL objects on ELF sh*-*-*
- References: <200210280043.g9S0hZO17072@r-rr.iij4u.or.jp>
On Mon, Oct 28, 2002 at 09:20:25AM +0900, kaz Kojima wrote:
> The idea is to select a representative target from the ambiguity set
> if possible. This is already done when the ambiguity set includes
> the default target vector. I add a list of the "associated targets"
> to the bfd_target structure and if one of them for the default target
> is a member of the ambiguity set, then use it as a representative.
> It's not so good, but is simple and is better than nothing.
Hi Kaz,
Thanks for looking at this problem. I'd like to see a solution
that is a little more maintainable though. I suggest putting
the associated targets into config.bfd and modifying the existing
mechanism for extracting DEFAULT_VECTOR to also define
ASSOCIATED_VECS, which will then be stored in a static var
declared in target.c. That way we keep target selection in one
place. Hmm, perhaps you could even use targ_selvecs, choosing
the first of targ_selvecs that matches any of the set of
ambiguous vectors.
--
Alan Modra
IBM OzLabs - Linux Technology Centre