Re: "kernel too old" SOLVED? (was: Version Skew and chroot)

On Sun, Apr 20, 2003 at 09:51:50AM -0500, D. Joe Anderson wrote:
> On Sat, Apr 19, 2003 at 03:46:29PM -0700, Nick Moffitt wrote:
> > begin  Chris Hart  quotation:
> > > I've just realised there will definitely be two build environments
> > > for a while for the BBC. Is there hope of the second chroot tree
> > > becoming available? I'm beginning to like them!
> 
> > 	Now, there is a trick to getting your kernel version straight
> > that will interfere with your ability to build HEAD's glibc, but that
> > is outside the scope of a chroot.  Inkblot may be able to comment
> > better on that situation.
> 
> What he's had to so far on this is 
> 
> http://zork.net/pipermail/lnx-bbc-bugs/2003-April/001333.html
> 
> 
> Yesterday, having made the indicated change to the --enable-kernel
> argument in the devel/glibc/Makefile, I was able to get meta/lnx-bbc
> from HEAD to build to completion within a Debian testing chroot
> running under kernel 2.4.18 from Debian stable.
> 
> I'm downloading it now to burn and then test to see if/how well it
> runs.

It booted.  It ran hackedbox and some other things linked against
libstdc++ which had been the hallmark bugaboo of version skew in
March.  

I first went looking for non-X executables that linked against
libstdc++, because I first booted this from a machine that doesn't
work with the framebuffer necessary to run X Windows.  

What I found was python and python2.2, and they segfault.  I haven't
taken a good look at an strace of a segfaulting invocation of either
yet, but I'm wondering if this is yet another, more subtle, instance
of skew.

--Joe