I use KDE media:/ kio-slave but it doesn't work properly. Some devices like USB keys, CDROMSs appear all the time unmounted. This bug is similar to bug #119759 . The problem is not media kio-slave, it's HAL. With
I use KDE media:/ kio-slave but it doesn't work properly. Some devices like USB keys, CDROMSs appear all the time unmounted. This bug is similar to bug #119759 . The problem is not media kio-slave, it's HAL. With « lshal », I can see that my USB key is displayed as unmounted : udi = '/org/freedesktop/Hal/devices/volume_uuid_5DC9_57F0' volume.policy.desired_mount_point = 'USB DISK' (string) volume.policy.mount_filesystem = 'vfat' (string) volume.policy.should_mount = true (bool) [...] volume.is_mounted = false (bool) volume.mount_point = '' (string) With CDROMs : udi = '/org/freedesktop/Hal/devices/volume_label_xxxxx' info.udi = '/org/freedesktop/Hal/devices/volume_label_xxxxx' (string) info.product = 'xxxxx' (string) [...] volume.is_mounted = false (bool) volume.mount_point = '' (string) With / partition : udi = '/org/freedesktop/Hal/devices/volume_uuid_1581840b_54bb_4327_a965_fb52f29eeece' volume.policy.desired_mount_point = 'scsidisk' (string) volume.policy.mount_filesystem = 'reiserfs' (string) volume.policy.should_mount = true (bool) [...] volume.is_mounted = false (bool) volume.mount_point = '' (string) The unique « device » which works is my Windows partition : udi = '/org/freedesktop/Hal/devices/volume_uuid_DC3C36073C35DCE8' volume.policy.desired_mount_point = 'scsidisk' (string) volume.policy.mount_filesystem = 'ntfs' (string) volume.policy.should_mount = true (bool) [...] volume.is_mounted = true (bool) volume.mount_point = '/mnt/windows' (string) With « mount », I can see they are actually mounted : /dev/sda5 on / type reiserfs (rw,noatime,notail) /dev/hda on /media/hda type iso9660 (ro,noexec,nosuid,nodev,uid=1000,gid=100,iocharset=utf8) /dev/sdb1 on /media/sdb1 type vfat(rw,noexec,nosuid,nodev,quiet,shortname=winnt,uid=1000,gid=100,umask=077,iocharset=utf8) /dev/sda1 on /mnt/windows type ntfs (ro,umask=022) [...] Used versions : dbus-0.60-r4 hal-0.5.5.1-r3
I found this in HAL Changelog : Kernel 2.6.15 will have a poll()'able /proc/mounts file, which tells us about mount tree changes. Kernel 2.6.16 will no longer have any netlink event regarding block devices mount/claim. We watch only /proc/mounts from now, remove all netlink listening and depend on kernel version 2.6.15. With every mount event, we update all known devices now, to the current state found in /proc/mounts. The device name in /proc/mounts is ignored, but the dev_t of the underlying block device is looked up to find the hal device. That way, /dev/root and rootdev will also be recognized and the current mount state becomes visible. I have manually bumped HAL to version 0.5.7, and now this bug is fixed. So, you can add an ebuild for HAL 0.5.7 to close this bug and bug #119759 .
fixed in CVS
