Glibc test failures

Clint Eaker ceaker at mindspring.com
Tue Sep 25 21:24:15 PDT 2007


> 
> On 9/25/07, Clint Eaker <ceaker at mindspring.com> wrote:
> > On Tue, 2007-09-25 at 19:20 -0700, Dan Nicholson wrote:
> > >
> > > Could you show the error from one of them? Can't say I'll be much 
> > > help, but it would be interesting to see them. FWIW, these errors 
> > > have
> >
> > > happened to people in the glibc-2.5.x line, but I can't 
> recall the 
> > > source of the problem.
> >
> > Since I'm not sure which one would be most helpful and they aren't 
> > especially large, I've attached all six.  I'm still a few versions 
> > back from glibc-2.5.x (see below).
> 
> Thanks for those. I was actually referring to building 
> glibc-2.5.x. Hopefully, your host's glibc shouldn't be 
> affecting things by that point.
> 
> > Linux version 2.6.18-5-686 (Debian 2.6.18.dfsg.1-13etch2)
> > (dannf at debian.org) (gcc version 4.1.2 20061115 (prerelease) (Debian
> > 4.1.1-21)) #1 SMP Thu Aug 30 02:19:07 UTC 2007
> 
> A little Google on "glibc cleanup handler not called for 
> 'read'" got me what I wanted.
> 
> http://sourceware.org/ml/libc-alpha/2007-01/msg00004.html
> 
> Apparently, enabling CONFIG_COMPAT_VDSO on linux-2.6.18 causes a lot
> of failures for glibc's testsuite. Can you check if your kernel has
> that set? If so, then we know the problem. Then, you can either assume
> that the failures are just due to the environment and continue (safe,
> IMO). Or, you can build a new kernel for Etch with CONFIG_COMPAT_VDSO
> off and try running the glibc tests again.
> 
> Of course, if the issue isn't CONFIG_COMPAT_VDSO, then there's some
> more digging.

That looks like the problem...

lfs at eaker-lfs1:~$ grep CONFIG_COMPAT_VDSO /boot/config-2.6.18-5-686
CONFIG_COMPAT_VDSO=y

Thanks for the help.  (I did *try* to google for something that looked
related...honest.  ;-)

Given the length of time it would take to rebuild the kernel on a PIII
machine, I think I'll make the assumption that the failures are due to
the environment.

Clint





More information about the lfs-support mailing list