Re: [buildd] gcc buildd failures
On Wed, 20 May 2009, Stephen R Marenka wrote:
> On Tue, Apr 21, 2009 at 09:22:19PM -0500, Stephen R Marenka wrote:
> > gcc-4.1 [0] and gcc-4.2 [1] failed with the following. wtf?
> >
> > | echo @set srcdir /build/buildd/gcc-4.2-4.2.4/build/gcc/../../src/gcc >> gcc-vers.texiT
> > | mv -f gcc-vers.texiT gcc-vers.texi
> > | mv: /build/buildd/gcc-4.2-4.2.4/build/./gcc/libgcc_s.so.2: version `GCC_4.3.0' not found (required by /lib/libselinux.so.1)
The command in the build log just prior to this creates a libgcc_s.so.2 in
the current working directory (it uses "mv" to do that).
The build failure occurs the very next time mv is called... It looks to me
like the dynamic linker links "mv" with the new libgcc_s that was just
created in the current working directory? (seems this new library wants to
link with libselinux.)
Could be a problem with glibc-2.5-11 dynalic linker? Or just a screwed up
library search path?
Finn
> >
> > [0] http://buildd.debian-ports.org/fetch.php?&pkg=gcc-4.1&ver=4.1.2-25&arch=m68k&stamp=1232439281&file=log&as=raw
> > [1] http://buildd.debian-ports.org/fetch.php?&pkg=gcc-4.2&ver=4.2.4-6&arch=m68k&stamp=1233229744&file=log&as=raw
>
> Stumbling around blindly in the dark I find that:
>
> m68k: libselinux1_1.32-3
> stephen@theia:~$ ldd /lib/libselinux.so.1
> libdl.so.2 => /lib/libdl.so.2 (0xc0018000)
> libsepol.so.1 => /lib/libsepol.so.1 (0xc001c000)
> libc.so.6 => /lib/libc.so.6 (0xc005d000)
> /lib/ld.so.1 (0xd0000000)
>
> m68k: libselinux1_2.0.71-1
> stephen@theia:~$ ldd chroot/unstable/lib/libselinux.so.1
> libdl.so.2 => /lib/libdl.so.2 (0xc001e000)
> libgcc_s.so.2 => /lib/libgcc_s.so.2 (0xc0022000)
> libc.so.6 => /lib/libc.so.6 (0xc0030000)
> /lib/ld.so.1 (0xd0000000)
>
> i386: libselinux1_2.0.71-1
> $ ldd /lib/libselinux.so.1
> linux-gate.so.1 => (0xb7f9d000)
> libdl.so.2 => /lib/libdl.so.2 (0xb7f78000)
> libc.so.6 => /lib/libc.so.6 (0xb7e36000)
> /lib/ld-linux.so.2 (0xb7f9e000)
>
> So why do we have a weird libgcc link? I rebuilt libselinux1, but
> it remains.
>
> What's the next step to chasing this down?
>
> Thanks,
>
> Stephen
>
>
Reply to: