Red Hat Bugzilla – Bug 196377
"Enabling local swap partitions:" always fails
Last modified: 2007-11-30 17:11:35 EST
Description of problem: "Enabling local swap partitions:" always fails with "swapon: /dev/cciss!c0d0p3: Device or resource busy" ("Fehlgeschlagen" means "failed") - sorry for the German screenshot. The problem is not SELinux related, I was just relabeling before, but this error is displayed during any startup. Swap seems to be available normal: $ cat /proc/swaps Filename Type Size Used Priority /dev/cciss/c0d0p3 partition 1048552 0 -1 $ $ free total used free shared buffers cached Mem: 1032956 926776 106180 0 191764 201876 -/+ buffers/cache: 533136 499820 Swap: 1048552 0 1048552 $ $ grep swap /etc/fstab /dev/cciss/c0d0p3 swap swap defaults 0 0 $ Version-Release number of selected component (if applicable): initscripts-8.35-1 Expected results: Would be very nice to see this fixed. Please reassign the bug to the quota package, if needed...
Created attachment 131383 [details] Screenshot of failure during booting time
Oh, blah. blkid returns the path with the '!' in it?
What's the output of 'blkid' on your system?
Oh, didn't get comment #2 per e-mail. Here's output of blkid (UUIDs have been trunicated) on my system: /dev/cciss/c0d0p1: LABEL="/boot" UUID="41e532d3cc77" SEC_TYPE="ext2" TYPE="ext3" /dev/cciss/c0d0p2: LABEL="/" UUID="b601fb50c59a" SEC_TYPE="ext2" TYPE="ext3" /dev/cciss/c0d0p3: TYPE="swap" /dev/cciss!c0d0p3: TYPE="swap" /dev/cciss!c0d0p2: LABEL="/" UUID="b601fb50c59a" SEC_TYPE="ext2" TYPE="ext3" /dev/cciss!c0d0p1: LABEL="/boot" UUID="41e532d3cc77" SEC_TYPE="ext2" TYPE="ext3"
OK, that needs fixed in libblkid; it should not be returning paths with the '!' in it...
No, it is no option that libblkid does not resolve partitions where the path contains special chars. This is no option at all. The question is: Where does these special entries come from? Assigning back to initscripts.
? These entries are returned by blkid... it's reading the '!' path from sysfs, and needs to be able to resolve it to the actual device.
The initial problem seems to be fixed for me at least in initscripts 8.46-1 and/or e2fsprogs 1.39-7.