This is the mail archive of the
libc-alpha@sources.redhat.com
mailing list for the glibc project.
Re: Problematic linking between glibc and shared libgcc
- To: Franz Sirl <Franz dot Sirl-kernel at lauterbach dot com>
- Subject: Re: Problematic linking between glibc and shared libgcc
- From: "H . J . Lu" <hjl at valinux dot com>
- Date: Thu, 15 Feb 2001 15:45:29 -0800
- Cc: gcc at gcc dot gnu dot org, libc-alpha at sources dot redhat dot com
- References: <01021523111501.01075@enzo.bigblue.local>
On Thu, Feb 15, 2001 at 11:11:15PM +0100, Franz Sirl wrote:
> Hi,
>
> yesterday I built a glibc-2.2.2 RPM with gcc-3_0-branch, tested and installed
> it, all was fine. Then I rebooted and my system didn't come up again,
> complaining about a missing libgcc_s.so.0.
>
> The problem is that / and /usr are on different partitions (yes, I know this
> is a silly setup with todays HD sizes, but a lot of systems out there are
> configured this way) and libgcc_s.so.0 is not available directly after boot.
> So I guess we have to install the shared libgcc into /lib on native glibc
> systems when the given prefix is /usr?
>
Is that really a surprise? I have been telling everyone about the
issuses with the shared libgcc under Linux. The current scheme simply
won't work well for Linux. IMHO, if we want to make the shared libgcc,
we have to
1. Make a ABI and stick to it.
2. Move it out of gcc and treat it like libstdc++.
3. Install it undr /lib for Linux.
4. Since we have a published ABI, we can have one in glibc and gcc
can use it instead of its own under Linux.
The key issue here is a well-defined, published and stable ABI for
libgcc.
--
H.J. Lu (hjl@valinux.com)