[OPIE 0001813]: mount applet not consistent with real state

The following issue has been set as RELATED TO issue 0001810.
======================================================================
http://opie-bugs.oszine.de/view.php?id=1813
======================================================================
Reported By: Laibsch
Assigned To: BlueLightning
======================================================================
Project: OPIE
Issue ID: 1813
Category: Taskbar/Launcher/Plugins
Reproducibility: always
Severity: major
Priority: normal
Status: feedback
======================================================================
Date Submitted: 08-20-2006 15:49 CEST
Last Modified: 12-22-2007 12:50 CET
======================================================================
Summary: mount applet not consistent with real state
Description:
Hi,

it happens that the mount applet claims to have unmounted the CF card when
in fact the CF card is still mounted (as verified by the mount command from
the console). This can lead to MAJOR data loss (it did for me, see my rant
at http://bugs.openembedded.org/show_bug.cgi?id=944)

Regards

Rolf
======================================================================
Relationships ID Summary
----------------------------------------------------------------------
related to 0001810 no applet to unmount SD card
child of 0001873 Opie 1.2.4 Release
======================================================================

----------------------------------------------------------------------
 BlueLightning - 12-14-07 07:13
----------------------------------------------------------------------
Actually, it seems that (with the Familiar images I have tested with at
least) neither the old cardmon applet nor the pcmcia applet make any
attempt to unmount cards before doing a software eject, and the eject will
happily go through leaving the card mounted.

Now, I have no idea if any of this is system related, and I have yet to
even determine a way of programmatically finding the link between a PC card
slot and a mounted block device, but this clearly needs to be fixed.

----------------------------------------------------------------------
 Laibsch - 12-14-07 14:52
----------------------------------------------------------------------
Thanks Paul, for taking this up.

----------------------------------------------------------------------
 BlueLightning - 12-15-07 07:27
----------------------------------------------------------------------
I have done some more investigation into this. I found a bug in the PCMCIA
scripts being used in Familiar 0.8.4 which resulted in unmount errors not
being sent back to cardmgr. However, when I fixed that bug it still allowed
the card to be removed with "cardctl eject". Perhaps that is expected
behaviour, I don't know.

I'd love to be able to test this on a device with PCMCIA slots using
up-to-date software, but unfortunately none of my iPAQs that have PCMCIA
slots can run 2.6 at the moment. Perhaps I will try it on my laptop.

----------------------------------------------------------------------
 BlueLightning - 12-19-07 12:57
----------------------------------------------------------------------
It seems pccardctl (as used in 2.6.13+) is no better here - it will also
pull the rug out from under a device without properly unmounting it first.

I think what is really needed here is a new card applet that can handle
both PCMCIA with proper unmounting for storage cards, and any other mounted
removable file systems (eg. SD cards, USB sticks). I will look into writing
this.

----------------------------------------------------------------------
 BlueLightning - 12-22-07 12:43
----------------------------------------------------------------------
OK, a new applet called "cardapplet" is now in CVS under
noncore/applets/cardapplet. When ejecting it first attempts to unmount any
associated block devices, and if that fails it shows an error and cancels
the eject. It should also show and allow safely unmounting SD cards and
other mounted storage devices, provided that the kernel has marked them as
removable (currently I notice the kernel does not do this for the SD slot
on my h2200, I'll have to follow that up).

There may be bugs to fix, so some testing would be appreciated :)

Issue History
Date Modified Username Field Change
======================================================================
08-20-06 15:49 Laibsch New Issue
12-14-07 07:13 BlueLightning Note Added: 0004757
12-14-07 07:13 BlueLightning Status new => confirmed
12-14-07 07:13 BlueLightning Relationship added child of 0001873
12-14-07 14:52 Laibsch Note Added: 0004759
12-15-07 07:27 BlueLightning Note Added: 0004760
12-15-07 08:11 BlueLightning Status confirmed => assigned
12-15-07 08:11 BlueLightning Assigned To => BlueLightning
12-19-07 12:57 BlueLightning Note Added: 0004774
12-22-07 12:43 BlueLightning Note Added: 0004777
12-22-07 12:44 BlueLightning Status assigned => feedback
12-22-07 12:50 BlueLightning Relationship added related to 0001810
======================================================================
Received on Sat Dec 22 2007 - 06:50:34 EST