glib2 problems on amd64

Bernd Knorr bernd at knorze.de
Fri Nov 18 03:10:11 PST 2005


OK, I could more or less solve the problem. Everything works well if the 
32-bit-version is not present at compile time. Maybe it has something to do 
with arch-dependent headers. I'm too lazy at the moment to go further into 
this problem.

Am Freitag, 11. November 2005 21:45 schrieb Ken Moffat:
> On Fri, 11 Nov 2005, Bernd Knorr wrote:
> > I tried precompiled glib and different versions but nothing helped.
> > I had the same problem with a CLFS build about two weeks ago.
> > On my old CLFS-system, which I built using ryan's scripts the problem
> > didn't occur.
> > So I think it might have something to do with the different glibc/gcc
> > versions on my old system (glibc-2.3.5 and gcc-3.4.3).
> >
> > I'd like to know if anybody has the same problem.
> > Could it have sth to do with the boot or chroot question (I chrooted) ?
>
>   My guess is glibc, mainly because it throws spanners into the works
> when you least expect it :)  Did you run make check on glibc when you
> built it?  I really don't see how booting or chrooting will cause you to
> create different code in the final system, because in either case you
> use the same tools.  Different kernel versions can sometimes upset the
> testsuites, particularly for glibc, but that isn't directly attributable
> to 'boot or chroot'.
>
>   My own multilib is now officially *old* (started 20051017 with a newer
> glibc than was then in the book) and I think we've upgraded glibc since
> then. I'm also probably using older package versions (gnome-2.10 era).
> I'm intermittently struggling through blfs, but with the aim of building
> a 32-bit firefox.  I've now got past glib2 : had to specify --build and
> --host for -m32, but the 64-bit version gave me no problems at all.
> Before that, I used the 20050930 book to build x86_64-64 and had no
> problems with glib2 or firefox.
>
> Ken



More information about the cross-lfs mailing list