On 30/09/2011 00:44, Bruce Dubbs wrote: > Matthew Burgess wrote: >> Hi, >> >> For a while now, I've had 2 test failures in coreutils, but have just >> ignored them. With the latest version I saw a 3rd failure, which made >> me relook at Coreutils' test suite. The failures I see are: >> >> misc/help-version >> misc/invalid-opt >> rm/many-dir-entries-vs-OOM >> >> The 1st 2 tests pass if I don't apply the i18n patch. The 3rd one, from >> a brief look at the coreutils mailing list, looks like it may be an >> issue with the test. >> >> If someone could confirm that they see similar failures, I'll add a '-k' >> switch to coreutils 'make check' invocation with a note as to which >> tests are expected to fail. > > On a 64-bit system I got: > > make NON_ROOT_USERNAME=nobody check-root > > All 14 tests passed > (9 tests were not run) > Not run due to selinux, xattr, libcap, python components missing. > > 2 of 401 tests failed > (57 tests were not run) > > I did get the help-version and invalid-opt errors, but the > many-dir-entries-vs-OOM test passed for me. You may want to check > tests/rm/many-dir-entries-vs-OOM.log > > The help-version test is doing: > > for i in $built_programs; do > v=$(env $i --version | sed -n '1s/.* //p;q') > break > done > > And failing with 'env: cd: No such file or directory' > > $built_programs is returning: > > cd ..&& /bin/sh ./config.status src/Makefile depfiles config.status: > creating src/Makefile config.status: executing depfiles commands [ > base64 basename cat chcon chgrp chmod chown chroot cksum comm cp csplit > cut date dd df dir dircolors dirname du echo env expand expr factor > false fmt fold ginstall groups head hostid id join link ln logname ls > md5sum mkdir mkfifo mknod mktemp mv nice nl nohup nproc od paste pathchk > pinky pr printenv printf ptx pwd readlink rm rmdir runcon seq sha1sum > sha224sum sha256sum sha384sum sha512sum shred shuf sleep sort split stat > stdbuf stty sum sync tac tail tee test timeout touch tr true truncate > tsort tty uname unexpand uniq unlink users vdir wc who whoami yes > > It should not have anything before the '[' command. > > The invalid-opt test is erroring out because of the same incorrect > $built_programs variable. > > The $built_programs variable is set in tests/Makefile. If I remove the > section modifying tests/Makefile.am, I get no errors and there is a 3rd > set of tests that are not run otherwise:
Ah yes, that all rings a bell now. It's triggered by the call to 'make NON_ROOT_USERNAME=nobody check-root'. The root cause appears to be the fact that the coreutils build machinery expects automake-1.11a (where is this I wonder, it's not on gnu.org!), and won't regenerate the tests/Makefile from the patched tests/Makefile.am without it. That leads to the automake related rules appearing in the .built_programs file. I've just tested a coreutils-8.13-i18n-2.patch that patches tests/Makefile.in as well as tests/Makefile.am, therefore avoiding the automatic automake calls, resulting in a correctly generated .built-programs file. It's late now, so I'll update the book tomorrow. Given that you can't reproduce the 3rd test failure, it looks again to be caused by my woefully memory constrained VM environment. That bug may just be fixed by Mrs Claus in a few months though! Thanks, Matt. -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
