Description of problem: The command "uname --kernel-version" doesn't display kernel version. Version-Release number of selected component (if applicable): coreutils-5.97-1.2 How reproducible: Always. Steps to Reproduce: 1. uname --kernel-version Actual results: #1 Sat Jul 1 13:03:45 EDT 2006 Expected results: 2.6.17 Additional info: The command "uname --kernel-release" works properly as it displays: "2.6.17-1.2145_FC5". I expect the "--kernel-version" flag to display "2.6.17" instead of a timestamp.
It's just reporting what the kernel tells it to from the uname() syscall. Reassigning.
Tim, I see no evidence that this ever did anything different. Here's a RHEL3 box running 2.4 kernel.. (20:16:16:davej@devserv:davej)$ uname --kernel-version #1 SMP Thu Dec 1 20:31:15 EST 2005 (20:16:16:davej@devserv:davej)$ uname -r 2.4.21-38.ELsmp uname -r works in FC5 too. Clearly uname can get the right string with uname -r, so this looks like a userspace bug to me ?
This is not util from the util-linux package: # rpm -qf /bin/uname coreutils-5.97-1.1 and ... I don't think there is any problem with uname: $ cat /proc/sys/kernel/version; uname --kernel-version #1 Tue Aug 8 15:30:55 EDT 2006 #1 Tue Aug 8 15:30:55 EDT 2006 $ cat /proc/sys/kernel/osrelease; uname --kernel-release 2.6.17-1.2174_FC5 2.6.17-1.2174_FC5 and ... "--kernel-version" != "-r", see man page. Reassigning to coreutils.
..back in a circle. No one can find any bug here. Closing.
What do you mean there's no bug? As far as I can tell, timestamps and kernel versions are 2 different things, right?
'#1' is the _version_ number, and you get the build timestamp as well -- you are thinking of the _release_ number. uname is just reporting what the kernel says. You ask for the version number, you get the version number. Ask for the release, you get the release instead. There is no bug here.