This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Versioning problem with __register_frame
- From: "Ulrich Weigand" <Ulrich dot Weigand at de dot ibm dot com>
- To: Andreas Jaeger <aj at suse dot de>
- Cc: jakub at redhat dot com, schwab at suse dot de, "Martin Schwidefsky" <schwidefsky at de dot ibm dot com>, "Hartmut Penner" <HPENNER at de dot ibm dot com>, gcc-patches at gcc dot gnu dot org
- Date: Mon, 4 Feb 2002 22:04:57 +0100
- Subject: Re: Versioning problem with __register_frame
Andreas Jaeger wrote:
>The idea looks fine - but note that s390x might not be the only port
>to use it. x86-64 comes to mind - but I have to check whether it's
>really needed there. I would prefer to not have separate
>libgcc-glibc.ver for each architecture and instead only add GLIBC_2.2
>somewhere... Any ideas?
I think x86-64 is in yet a different situation than s390x, because
AFAIK there has never been any gcc 2.95.x compiler for x86-64.
Therefore you don't need to have the *register_frame* functions
in glibc at all, because you always have a shared libgcc.
However, on s390x, we are in the probably unique situation that we
have a gcc 2.95.x based compiler out there, but never had any other
glibc than version 2.2.x ...
In fact, maybe we should rather force every architecture that needs
it to have its own libgcc-glibc.ver, and have no default at all?
It looks like the default is wrong anyway for some arches, e.g. the
divdi3 etc. routines are bogus on 64-bit archs, as they were never
exported by any glibc. (On the other hand, with early compilers on
s390, we had __negdi2 accidentally exported by libstdc++ :-/)
Mit freundlichen Gruessen / Best Regards
Ulrich Weigand
--
Dr. Ulrich Weigand
Linux for S/390 Design & Development
IBM Deutschland Entwicklung GmbH, Schoenaicher Str. 220, 71032 Boeblingen
Phone: +49-7031/16-3727 --- Email: Ulrich.Weigand@de.ibm.com