Re: glibc broken

[Roland McGrath]

> On Tue, May 22, 2001 at 11:04:54PM -0400, Roland McGrath wrote:
> > It is important to be clear about what works and what doesn't:
> > 
> > * statically linked user programs
> > * statically linked user programs that use cthreads
> 
> I don't see how a statically linked program can be affected by a glibc
> upgrade.  Note that I didn't recompile the Hurd with the new glibc.

Ah!  I presumed you meant you'd recompiled everything, so I was thinking
about the statically-linked boot fs.

Be aware that when you use glibc versions in between releases, you might
need to compile sometimes.  The binary compatibility occasionally gets
broken by some test version of glibc and then fixed in a way that is not
compatible with binaries compiled against that broken glibc.

> > Also test whether running a program via ld.so works,
> > i.e. /lib/ld.so.1 /bin/foo arg1 arg2
> > 
> > This is how exec gets run.  Try /lib/ld.so.1 /hurd/exec --help.
> 
> I believe I did this and it worked, but I will try again.

If this works, then I really start to wonder.  If you can run all those old
hurd binaries (e.g. just testing with --help) using the new libc, then the
problem must be quite obscure.

> > As to debugging a hurd boot problem, is there a reason you haven't done
> > this under a sub-hurd and given more detailed information?
> 
> Only lack of time.  That's the next step.