Arch Linux

Okay, thank you. I will report this upstream.

Emil: coreutils test suite applies root:object_r:tmp_t:s0 to a directory and in your case this fails with "chcon: invalid context: root:object_r:tmp_t:s0: Invalid argument" (cf. http://git.savannah.gnu.org/cgit/coreutils.git/tree/tests/mkdir/restorecon.sh?id=e0afeb0099c899c2caa7453072e2d223d8f0ceb9#n27 ).
As far as I understand, this is an upstream limitation, which expects to be able to change any context in the test suite even though this may not be allowed if there is a policy loaded in enforced mode. Moreover it only supports MLS policy (because of the :s0 part of the context), which makes the test suite also fails hard on system with a non-MLS policy. Could you please report this upstream? (for example at http://debbugs.gnu.org/cgi/pkgreport.cgi?pkg=coreutils). A nice solution might be to skip the test if "chcon 'root:object_r:tmp_t:s0' subdir" fails in tests/mkdir/restorecon.sh.
Anyway building with makepkg --nocheck should succeed.

I am getting this error when I try to build this. http://pastebin.com/B7qiHvb3

Thanks for the info. I'll update this package as soon as possible.

You will want to include this patch until the next release is made:
http://lists.gnu.org/archive/html/coreutils/2014-01/msg00012.html

Renamed to coreutils-selinux

su is part of util-linux as of some recent version, so you should have current selinux-util-linux installed.

I think it broke su command, su, it says command not found

Ok, I deleted the line to make the backup warning go away. I'll install the systemd package that you made, I hope there is no fatal outcome :P

On the testing machine I still have sysvinit. Since it's being dropped from Arch, I will try to find out, how to do it with systemd. As for the warning, this file was moved to util-linux package, so there is not backup here.

By the way, just a warning was raised, the file etc/pam.d/su doesn't exist from the backup input

Thanks Nicky726, I'll install one by one and if I encounter any issues I'll tell you about it, I'll try to install every selinux package and set it to permissive to test. By the way, are you still using sysvinit or switched to systemd?

All packages from selinux-system-utilities group + the kernel should now be up-to-date with [core]. I gave them some testing in my virtualized machine, but I'm currently stuck with loading SELinux policy, so I tested them in non-SELinux system. I may give it one more try today, or I'll look into it next weekend.

I'll update the selinux-system-utilities packages.

Should I just change the version to 8.20? or just change the PKGBUILD file for the kernel so it doesn't need this package but the default coreutils?

If I change the version number only it tryes to patch with the old patches, are those related to selinux or I should only delete them?

With version 8.20 it builds and installs fine. It also seems to work on on non-SELinux testmachine.

Hi, I've come this far, now selinux-coreutils fails to install but not to build. It complains about conflicting files /bin/su, /etc/pam.d/su and /usr/share/man/man1/su.1.gz

I have tested the install of selinux-pam with siosm's and it compiles and installs without problems but when I come to selinux-coreutils it fails to install, I will see if I can update the pkgbuild file to see if it's a version problem.

You're right. The problem isn't with the selinux support. Even the core package from ABS fails the test. Sorry for the trouble.

Hi, I just build if without troubles. Does the test also fail if you build coreutils from abs?

One of the tests also fails for me: stty-pairs. The log for the test is full of lines like
stty: standard input: unable to perform all requested operations
and I always seem to get that error after stty commands including "parodd", however not all commands with "parodd" result in an error.

Any idea why this happens? Or can I just safely ignore it?

Thanks,
gsx1022

I wouldn't think it's lack of RAM, server has 4GB RAM and 2GB swap, 1.5GB is used by services.
The names of the test are 'cp-a-selinux' and 'many-dir-entries-vs-OOM'
I don't know what many-dir-entries checks, but I think cp-a-selinux mounts a loop and tries to copy from it or whatnot?

Was it killed due to lack of RAM, or is it just the name of the test?

Not sure why but I always get problems when compiling this. Today with 8.16-2, 2 tests failed;
FAIL: cp/cp-a-selinux (exit: 99)
FAIL: rm/many-dir-entries-vs-OOM (exit: 1)
And here's a bit of the text with the second failed test;
+ du -sh d
du: fts_read failed: d: Cannot allocate memory
+ fail=1
+ chmod -R 700 d
./rm/many-dir-entries-vs-OOM: line 35: 3092 Killed chmod -R 7$
+ fail=1
+ rm -rf d
./rm/many-dir-entries-vs-OOM: line 36: 3095 Killed rm -rf d
+ fail=1

Not really sure why either are failing if I'm honest.

Thanks for the quick update :)

Added, thank you.

This patch is required for coreutils 8.14 to compile: http://pastie.org/3003839

I've modified the patch so that it applies correctly to coreutils 8.14. There original patch is here: http://thread.gmane.org/gmane.comp.lib.gnulib.bugs/28825

The 'test-parse-datetime.c' test apparently fails because of daylight time savings in certain areas of the world.

Thanx, fixed.

Small mistake here: conflicts=("${_org *i* name}" 'mktemp')
It should be: conflicts=("${_or *i* gname}" 'mktemp')

Updated to 8.12-2 which brings the hostname program as the coreutils in [core].