Le Friday 14 September 2007 10:37:25 Gallien Matthieu, vous avez écrit :
> Le Thursday 13 September 2007 18:18:54 Aurelien Jarno, vous avez écrit :
> > Gallien Matthieu a écrit :
> > > Setting up libc6 (2.6.1-3) ...
> > > + set -e
> > > + export LC_ALL=C
> > > + LC_ALL=C
> > > + type=configure
> > > + preversion=2.6.1-1+b1
> > > + '[' configure = configure ']'
> > > + '[' -f /usr/share/debconf/confmodule ']'
> > > + . /usr/share/debconf/confmodule
> > > ++ '[' '!' '' ']'
> > > ++ PERL_DL_NONLAZY=1
> > > ++ export PERL_DL_NONLAZY
> > > ++ '[' '' ']'
> > > ++ exec /usr/share/debconf/frontend /var/lib/dpkg/info/libc6.postinst
> > > configure 2.6.1-1+b1
> > > dpkg: error processing libc6 (--configure):
> > >  subprocess post-installation script killed by signal (Segmentation
> > > fault)
> >
> > Wow it's very strange that /usr/share/debconf/frontend segfault in such
> > conditions. It is a perl program.
> >
> > While a version gap between libc6 and libc6-i686 could really cause such
> > causes of problem, the difference is very small in your case, and I
> > don't really see how it could cause such a problem.
> >
> > Does a /etc/ld.so.nohwcap file exists when the versions of libc6 and
> > libc6-i686 mismatch (after a failure in the postinst script)?
>
> Yes there is an empty one
>

> > - Are you sure there is a correlation between the segfault and the
> > presence of libc6-i686?
>
> No it seems that there is other path to reproduce the error.
> At least :
>
> if libc6 and libc6-i686 are correctly installed from unstable.
> Then I remove libc6-i686. I downgrade to libc6/testing.

There is no /etc/ld.so.nohwcap and:

wahiba:~# ldd /usr/bin/perl
        linux-gate.so.1 =>  (0xffffe000)
        libdl.so.2 => /lib/libdl.so.2 (0xb7edb000)
        libm.so.6 => /lib/libm.so.6 (0xb7eb5000)
        libpthread.so.0 => /lib/libpthread.so.0 (0xb7e9e000)
        libc.so.6 => /lib/libc.so.6 (0xb7d56000)
        libcrypt.so.1 => /lib/libcrypt.so.1 (0xb7d28000)
        /lib/ld-linux.so.2 (0xb7efc000)

> And I try again to upgrade. There is now a faillure.

There is a /etc/ld.so.nohwcap and:

wahiba:~# ldd /usr/bin/perl
        linux-gate.so.1 =>  (0xffffe000)
        libdl.so.2 => /lib/libdl.so.2 (0xb7f39000)
        libm.so.6 => /lib/libm.so.6 (0xb7f13000)
        libpthread.so.0 => /lib/libpthread.so.0 (0xb7efb000)
        libc.so.6 => /lib/libc.so.6 (0xb7db3000)
        libcrypt.so.1 => /lib/libcrypt.so.1 (0xb7d85000)
        /lib/ld-linux.so.2 (0xb7f59000)

and if I manualy rm /etc/ld.so.nohwcap.

the configuration of libc6 works.

So the problem of segfault seems related to the presence 
of /etc/ld.so.nohwcap.

> If I do:
> dpkg --ignore-depends=libc6 -i
> /var/cache/apt/archives/libc6-i686_2.6.1-3_i386.deb
>
> after the manual forced installation of libc6-i686:
>
> wahiba:~# ldd /usr/bin/perl
>         linux-gate.so.1 =>  (0xffffe000)
>         libdl.so.2 => /lib/i686/cmov/libdl.so.2 (0xb7f5c000)
>         libm.so.6 => /lib/i686/cmov/libm.so.6 (0xb7f37000)
>         libpthread.so.0 => /lib/i686/cmov/libpthread.so.0 (0xb7f20000)
>         libc.so.6 => /lib/i686/cmov/libc.so.6 (0xb7dd8000)
>         libcrypt.so.1 => /lib/i686/cmov/libcrypt.so.1 (0xb7daa000)
>         /lib/ld-linux.so.2 (0xb7f7d000)

and there is no /etc/ld.so.nohwcap

and after the dpkg --configure -a

the result is still the same

>
> Hope this help



-- 
Gallien Matthieu


Reply via email to