[lfs-dev] glibc-2.21 failed tests and minor fix

Ken Moffat zarniwhoop at ntlworld.com
Sun Mar 1 17:48:31 PST 2015

On Sun, Mar 01, 2015 at 10:16:44PM -0300, Fernando de Oliveira wrote:
> 5.7. Glibc-2.21:
> s/Fix is regression/Fix regression/
> 6.9. Glibc-2.21
> Seven failures in the tests. Following three not in the book.
> FAIL: grp/testgrp
> FAIL: posix/globtest
> FAIL: pwd/tst-getpw
> Problem seems to be the necessity of /etc/passwd and /etc/group with
> user:group 0:0 all defined. After copied the ones from the host (Mint),
> those tests passed:
 Nice analysis.  I did not see these failures because I always create
/etc/passwd and /etc/group on my way into chroot.  My notes suggest
that is because I got pissed off with a lack of name resolution,
which might have been 'I have no name', or might have been issues
with trying to chown files and logs to a non-root user, or even
just issues with conditional tests when I had to re-enter chroot
after leaving it.  Apart from jhalfs, there are many "approximately
correct" ways of scripting as there are builders.

 While we are talking about failures, I continue to see around 17 or
18 gcc failures on my x86_64 gcc builds, mostly in the torture
tests, and 3 in perl.  But hey, they're tests, nobody should care
;-)  And I have not build i686 on real hardware so far.

Nanny Ogg usually went to bed early. After all, she was an old lady.
Sometimes she went to bed as early as 6 a.m.

More information about the lfs-dev mailing list