http://bugzilla.novell.com/show_bug.cgi?id=557565
http://bugzilla.novell.com/show_bug.cgi?id=557565#c8
--- Comment #8 from Kay Sievers 2009-12-18 07:01:45 UTC ---
With race, you mean the direct repeated blkid calls without re-validation after
media change? They should not happen with mount(8).
blkid(8) and mount(8) use the same library/cache file. But mount(8) uses it
only as a hint and always forces a verify of the result of the libblkid call,
so there should be no race here, only in the "debugging tool" blkid(8), when
called in cache lookup mode, which is the default behavior.
Udev also always calls blkid(8) on every media change detection, but bypasses
any logic which uses any cache file or scanning of other devices, and will
therefore always cause a direct read of the device directly with the first
open() of the device node after a media change.
It's likely the probing for the filesystem type itself, which causes the noisy
seeks on the drive. We also try to find the name(label) of the inserted media
which involve seeks. I don't think there is an obvious fix for the media
scanning/seeking, which seems to make an unusual sound on this specific drive.
If it helps, sure, udev could be made to update the timestamp of the device
node on media change, but that would not change the issue mentioned here,
right?
--
Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.