Red Hat Bugzilla – Bug 37072
Very bad: (SCSI) Magneto Optical disks can be ejected when mounted!!
Last modified: 2005-10-31 17:00:50 EST
From Bugzilla Helper:
User-Agent: Mozilla/4.77 [en] (X11; U; Linux 2.2.16-3smp i686)
With RH7.1 SeaWolf I can eject a MagnetoOptical disk (the drive is a
Fujitsu DynaMO640) even though it is mounted and i.e. I cd to it.
If I press the eject button the disk comes out with all possible
consequences (I trashed already two disks).
It did not occur on all previous versions of RH neither in
Wolverine as far as I know (I installed it only partially). I do not know
whether the problem is in mount or somewhere else, however it seems that
mounting does no longer tell the hardware to inhibit the eject. I met the
problem on a Latitude C800 with a APA1480A
SCSI card and the disk formatted as superfloppy (all drive
formatted as one single file system, "mke2fs /dev/sdb")
The same hardware combination (and the same disk) worked nicely with RH7.0,
a similar machine (Latitude Cpx500) with the same SCSI card and MO drive
worked nicely with RH6.1,6.2 and 7.0
Steps to Reproduce:
1.format the MO disk ("mke2fs -b 2048/dev/sdb")
2.mount it ("mount -t ext2 /dev/sdb /diskmo") write some data, cd to it
3.press the eject button
Actual Results: the disk is ejected (and the system sees it still mounted,
of course after a while the system heavily complains)
Expected Results: the disk should not be ejected
Hi, I am surprised to see no reaction to such a serious bug (trashing disks is
not the best advertising for Linux). I can confirm that the bug is there with
both the new and the old pcmcia drivers. If it is a bug of the SCSI driver
(aic7xxx) it will affect a large amount of cards due to the diffusion of Adaptec
hardware, if it is a bug of mount it should affect all cards...
I had another bug for mount (less serious but still disturbing) which was
submitted february 26th (29542, it affects both RH7.0 and RH7.1). No reaction on
that side too, it is still NEW not even ASSIGNed, is the maintainer of mount in
indefinite vacation? When I posted the bug to the RedHat lists RH guys answered
in the usual way "put it into bugzilla" however they did not add it was a way to
forget it forever...
Reported against the wrong component, the kernel handles device locking.
Reassigning. This may well be a bug in your device's firmware, though.
Is magicdev running ?
Hi, sorry for bothering you again.... after the APA1480 stuff. I'll tell you in
half a hour (I have to go back home to check) if magicdev is running or not. I
would exclude the device firmware since it is the same exact device which
worked nicely (=> blocking the ejection) under RH6.0,6.1,6.2 and 7.0 (the last
till one week ago until I upgraded to 7.1).
Two other identical devices (same model same firmware revision) are regularly
blocking the ejection on two 6.2 systems as well.
No magicdev is not running. Let me know whether you nedd further infos.
*** This bug has been marked as a duplicate of 38871 ***