This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: QNX binutils targets
- From: <gp at qnx dot com>
- To: "Segher Boessenkool" <segher at koffie dot nl>, "Alan Modra" <amodra at bigpond dot net dot au>
- Cc: "Graeme Peterson" <gp at qnx dot com>, <binutils at sources dot redhat dot com>
- Date: Sat, 26 Oct 2002 17:48:59 -0000
- Subject: Re: QNX binutils targets
Hi all.
I think I will be putting together a patch that removes the QNX bfd and QNX
specific code that is causing the grief. The problem our code is meant to
solve is something we want to have supported, but it is not core to our
products yet, and I don't think I want to break backward compatibility with
our existing binaries and tools (both GNU and QNX in-house).
I will then look into other more friendly ways to achieve what the current
elf32-qnx.[ch] does, and re-submit the support.
This will then change all of the QNX support to use the base vector's that I
had originally extended, and hopefully I will be able to come up with an
alternate solution for the other stuff, which may boil down to implementing
H.J's ELFOSABI_QNX solution.
But in the meantime, QNX host/target support will not break other
hosts/targets when configured enabling all targets.
Comments?
Thanks.
GP
Segher Boessenkool <segher@koffie.nl> said:
> Alan Modra wrote:
> >
> > On Thu, Oct 24, 2002 at 11:32:00PM +0200, Segher Boessenkool wrote:
> > [zapping targets from _bfd_target_vector]
> > > which makes things work as I like. It seems this is done for some other
> > > ambiguously detected formats as well.
> >
> > Yes, that works. The trouble is that your selection of targets to
> > "make things work as I like" may not suit everybody.
>
> Sure, but this broke my cross to i686-unknown-linux-gnu -- it might
> not suit everybody, but it'll probably suit more people. Or maybe
> not many people are cross compiling *to* x86-linux?
>
> Nice to see this is being solved so rapidly, btw :)
>
>
> Segher
>
--