Red Hat Bugzilla – Bug 826519
cups 1.5.3-2.fc17 interferes with udisksctl unmount
Last modified: 2012-09-19 04:52:53 EDT
Description of problem:
After installing the recent cups testing update, "udisksctl unmount" fails to remove the directory created for a mounted USB disk.
Version-Release number of selected component (if applicable):
Reproducible - after some further testing, I think cups prevents udisksd cleaning up the directory for USB disks that were mounted when cups was started.
Steps to Reproduce:
1. systemctl stop cups.service
2. Mount a USB disk with "udisksctl mount"
3. open a print dialog to auto start the cups service
4. Unmount the USB disk with "udisksctl unmount"
The disk is unmounted but the directory created remains.
May 30 12:36:23 obsidian udisksd: Cleaning up mount point /run/media/ejs/UDISK 2.0 (device 8:17 is not mounted)
May 30 12:36:23 obsidian udisksd: Error cleaning up mount point /run/media/ejs/UDISK 2.0: Error removing directory: Device or resource busy
The directory gets removed.
If you mount the USB disk again, it gets mounted in a new directory with a 1 appended to the name. And then udisks2 gets confused about what to unmount:
$ ls -l /run/media/ejs/
drwx------ 2 root root 40 May 30 12:35 UDISK 2.0
drwx------ 8 ejs ejs 16384 Jan 1 1970 UDISK 2.01
$ udisksctl unmount -b /dev/sdb1
Error unmounting /dev/sdb1: GDBus.Error:org.freedesktop.UDisks2.Error.Failed: Error unmounting /dev/sdb1: Command-line `umount "/run/media/ejs/UDISK 2.0"' exited with non-zero exit status 32: umount: /run/media/ejs/UDISK 2.0: not mounted
If I do "systemctl stop cups.service", udisksd does clean up the old directory (eventually, I mounted and unmounted another disk to trigger the cleanup).
Possibly related to bug #747114 or bug #808109 but this started with cups-1.5.3-2.fc17.
This could indeed be related to bug #747114, bug #808109.
Also the 1.5.3 provides an improved USB backend based on libusb 1.0,
which could possibly somehow interfere with this.
Do you have any printer configured on your system ?
I have a USB Deskjet-5900-series (hplip) configured but it's not usually plugged in nor switched on.
I'm able to reproduce it even with cups-1.5.2-12.fc17.x86_64 so I'd say it's a duplicate of the above mentioned bugs. Let's wait for Tim (he was investigating them more deeply) to decide.
Marking as duplicate.
*** This bug has been marked as a duplicate of bug 747114 ***