Description of problem: Rebuild of coreutils fails with "FAIL: chcon.log", because mcstrans(d) is not a must. So please correct the patch to work also without mcstrans(d) running. Version-Release number of selected component (if applicable): coreutils-6.10-11 How reproducible: Everytime, see above. Expected results: Working rebuild even without having mcstrans(d) running.
Created attachment 297381 [details] tests/misc/chcon.log
Thanks for report, will try to find suitable solution for unactive selinux translation... but I don't know about which patch you are talking about. Chcon implementation was in selinux patch in coreutils-6.9, coreutils-6.10 are using upstream implementation of chcon (which is very similar to implementation in coreutils-selinux.patch ).
Sorry, I didn't have an exact look to it, I thought that SELinux would come from that patch. The fix is IMHO just to allow both: With and without translation. But I don't know how the check is performed, maybe just an regexp?
In coreutils-6.9 all selinux things were from that selinux patch - coreutils-6.10 included selinux things by default - but unfortunately in some cases unlike with Red Hat patch. Therefore upstream solution is based on Red Hat patch, but sometimes not the same. Chcon check is following: perform some commands and compare results with expected results. So maybe the easiest way would be to skip the test when translation is not active - as is already done for the case of some only "non-root" tests executed by root.
Anyway - chcon is root only test, so it will be skipped when rebuild with non-root user. So that's an easy workaround for the issue. Same problem with not running mcstrans(d) was with cp-a-selinux test. While fixing that issue I found that the chcon test fails in selinux enforcing mode as well... Both issues fixed and built as coreutils-6.10-15.fc9, closing RAWHIDE.
Thanks for the report and investigation. I've fixed this upstream: http://thread.gmane.org/gmane.comp.gnu.coreutils.bugs/13098