make -k check during check of Glibc-2.5.1

Ken Moffat ken at linuxfromscratch.org
Thu Jul 31 14:33:04 PDT 2008


On Thu, Jul 31, 2008 at 09:30:55PM +0200, Wolfgang Messingschlager wrote:
> Hi all,
> 
> 
> 
> my host Linux system (AMD Athlon 2000 Hz) is a debian etch system with
> 
> - gcc 4.1.1-15
> - kernel 2.6.18-5-686
> - libc6  2.3.6.ds1-13etch5
> 
> I tried to build a LFS system from the LFS 6.3 book. However during page 
> 6.9 (Installation of Glibc) the test suite for Glibc is considered 
> critical. Unfortunaely I got the following error after
>     make -k check 2>&1 | tee glibc-check-log
> 
> ====================================================================
> package glibc:/usr/src/glibc/glibc-build> grep Error glibc-check-log
> make[2]: [/usr/src/glibc/glibc-build/posix/annexc.out] Error 1 (ignored)
> make[2]: *** [/usr/src/glibc/glibc-build/nptl/tst-mutex8.out] Error 1
> make[2]: *** [/usr/src/glibc/glibc-build/nptl/tst-mutexpi8.out] Error 1
> make[2]: *** [/usr/src/glibc/glibc-build/nptl/tst-cond8.out] Error 1
> make[2]: *** [/usr/src/glibc/glibc-build/nptl/tst-cancel24.out] Error 1
> make[2]: *** [/usr/src/glibc/glibc-build/nptl/tst-cancelx4.out] Error 1
> make[2]: *** [/usr/src/glibc/glibc-build/nptl/tst-cancelx5.out] Error 1
> make[2]: *** [/usr/src/glibc/glibc-build/nptl/tst-cancelx16.out] Error 1
> make[2]: *** [/usr/src/glibc/glibc-build/nptl/tst-cancelx20.out] Error 1
> make[2]: *** [/usr/src/glibc/glibc-build/nptl/tst-cancelx21.out] Error 1
> make[2]: *** [/usr/src/glibc/glibc-build/nptl/tst-cleanupx4.out] Error 1
> make[1]: *** [nptl/tests] Error 2
> make: *** [check] Error 2
> ====================================================================
> 
> However in ../glibc-2.5.1/nptl/tst-mutex8.c is mentioned
> 
> /* This test checks behavior not required by POSIX.  */
> 
> So I am not sure if I did something wrong or if I can continue
> 
> 
> Thanks in advance
> Messingschlager Wolfgang
 I don't think I've ever seen quite that many failures in an x86
build (in particular, mutexpi8 and cleanupx4 don't ring any bells -
the cancel variants are "not uncommon" in the list archives).  But,
the failures are only Error{1,2} ( if Error 127 or higher numbers
ever show up, something is missing or broken ).

 My personal experience with glibc (particularly on other
architectures), and with test suites in general, leads me to suggest
you shouldn't worry too much about these failures.  In particular,
your kernel appears to be quite a bit older than your LFS headers (I
say 'appears' because I've no idea what patches and backports it
contains).  I'd hazard a guess that either the running kernel or the
particular processor architecture you are running on is causing at
least some of these tests to fail.

 If you build frequently and the test results suddenly diverge, that
points a finger at something in the change (new kernel, different
book instructions, version upgrade somewhere).  In other cases, it
can be hard to interpret if the differences matter.

 Certainly, there have been various reports of nptl test failures in
the past.  I don't think anybody ever noted a problem in the
resulting systems that could be associated with this.

 But, as someone building from debian, please make sure that your
'awk' is gawk, ISTR the glibc install will fail if you are using
e.g. mawk.  A symlink to gawk should be fine.  So, as long as 'awk
--version' reports gawk >= 3.0 you are probably ok.  Perhaps I'm
unnecessarily doubting the book's 'host requirements' (which says to
run 'gawk --version').

 Actually, a quick google for mutexpi8 found a report from someone
building HLFS in i686 from debian etch.  Maybe something in debian
causes these tests to fail, but I doubt it will be a problem.

ĸen
-- 
das eine Mal als Tragödie, das andere Mal als Farce



More information about the lfs-support mailing list