Bug 8645 - (int-155405) External vfat mmc corrupted, file tables damaged? all files missing
(int-155405)
: External vfat mmc corrupted, file tables damaged? all files missing
Status: UNCONFIRMED
Product: Core
general
: 5.0/(2.2009.51-1)
: N900 Maemo
: Low critical (vote)
: ---
Assigned To: unassigned
: core-general-bugs
:
:
:
:
  Show dependency tree
 
Reported: 2010-01-29 10:35 UTC by Spencer Goh
Modified: 2014-03-08 10:43 UTC (History)
5 users (show)

See Also:


Attachments


Note

You need to log in before you can comment on or make changes to this bug.


Description Spencer Goh (reporter) 2010-01-29 10:35:02 UTC
SOFTWARE VERSION:
2.2009.51-1.203.2

EXACT STEPS LEADING TO PROBLEM: 
1. Device left charging overnight via mains charger
2. (preset clock) Alarm woke me up with beeping noise but not the specified
ringtone.
3. Device was very hot and non-responsive, disconnected charger
4. LED indicator was still pulsing orange. No response to lock-slider,
touchscreen, power button, long press power button to shutdown
5. Had to forcefully remove battery while still alive
6. Waited for device to cool & rebooted normally
7. Some widgets had been removed from desktop (I manually re-added -
"TouchSearch" "OMWeather" "Facebook"
8. Browsed /media/mmc1  and found all my files had vanished leaving only a info
directory there, which I had never noticed before (containing .md5sum and .list
files)
9. Shutdown & Removed MMC card, placed back into device, & rebooted
10. MMC still lots all data
11. umounted MMC and ran fsck -vvv /dev/mmcblk1p1 and found no problems
12. Ran fsdisk -l

Disk /dev/mmcblk1: 243200 cylinders, 4 heads, 16 sectors/track
Warning: The partition table looks like it was made for C/H/S=*/221/20 (instead
of 243200/4/16).
For this listing I'll assume that geometry

/dev/mmcblk1p1 1+ 3521- 3520- 7778304 b W95 FAT32
start: (c,h,s) expected (1,188,13) found (0,130,3)
end: (c,h,s) expected (1023,220,20) found (968,220,20)


http://talk.maemo.org/showthread.php?p=500358

EXPECTED OUTCOME:
1. All my saved data is on the /media/mmc1

ACTUAL OUTCOME:
2. All my files & directories are missing from /media/mmc1

REPRODUCIBILITY:
0/10

EXTRA SOFTWARE INSTALLED:

OTHER COMMENTS:
Have had occasional problems recently with the phone crashing and needed forced
shutdown by holding onto powerbutton) - possibly unrelated, not sure how this
was triggered. Am happy to do more investigation, but unsure what info you
need.
I've taken an image of the uSSD 16Gb card and hope to try recover all my data

dd if=/dev/mmcblk1 ds=512 | bzip2 > /home/user/MyDocs/mmcBackup.bz2 

User-Agent:       Mozilla/5.0 (X11; U; Linux x86_64; en-GB; rv:1.9.1.6)
Gecko/20100107 Fedora/3.5.6-1.fc12 Firefox/3.5.6
Comment 1 Andre Klapper maemo.org 2010-02-02 19:56:53 UTC
Thanks for reporting this.
Comment 2 Eero Tamminen nokia 2010-02-04 17:00:34 UTC
> Have had occasional problems recently with the phone crashing

Crashing how?  (Do you have another bug about them?)
Comment 3 Spencer Goh (reporter) 2010-02-05 03:47:33 UTC
Re: Crashing.

I didn't open a bug report about this because I hadn't fully isolated the cause
of it.

I was running WiFi connection all day long.
The Wifi was an ad-hoc network served from my PC,  which as I understand it
Ad-Hoc doesn't support powersave (hence I run out of battery very quickly!)
As a result I simultaneously connected to USB power (plugged into Linux PC) all
day.

I found that during the day the N900 would completely grind to a halt.
Touch screen non-responsive... task switch would take seconds to start
responding... and the transition animations would not show... just jumped from
one screen to next... general **extreme** lag.. 

I could barely run anything, couldn't even launch terminal to run top or
shutdown gracefully.   Perhaps some event feeback loop, because as soon as I
started interacting with the device it got even slower, grinding to a halt.

Once I managed to disconnect the WiFi using the N900 applet, and then it went
back to normal.  However on subsequent crashes I couldn't repeat this - was
forced to powerdown the device due to non-responsiveness  (Hold powerbutton)

I have hard-powered it down a few times... but only once or twice had to pull
out the battery (as I did the other day) due to this massive overheating.
Comment 4 Eero Tamminen nokia 2010-02-09 14:13:46 UTC
> 8. Browsed /media/mmc1  and found all my files had vanished leaving only
> a info directory there, which I had never noticed before (containing
> .md5sum and .list files)

Maybe a stupid question, but did/does "mount" report your card to be mounted?
(see bug 8235)


(In reply to comment #3)

Thanks.  The issues you describe might be related to stuff in bug 5712.


> Touch screen non-responsive... task switch would take seconds to start
> responding... and the transition animations would not show... just jumped
> from one screen to next... general **extreme** lag.. 

If this happens again, please check does "dmesg | grep SGX" dmesg show
anything?


> due to this massive overheating.

And if you're still able to start XTerm or ssh into the device, please check
with "top" whether something is using CPU.
Comment 5 Spencer Goh (reporter) 2010-02-09 14:25:07 UTC
>> Maybe a stupid question, but did/does "mount" report your card to be mounted?

The card was mounted no problem, except looked like someone had rm -fr'd all my
data!  I unmounted then ran fsck at which point it told me of those
inconsistencies I reported.  But did not state that the card was corrupted.  I
decided not to write anything in case I corrupted the FS, and then dumped the
mmc with dd.

>> Re: https://bugs.maemo.org/show_bug.cgi?id=5712

I think i've definitely experienced this...  swap hell would explain the whole
thing grinding a halt, and seemed very consistent with the ad-hoc wifi.

I'll try your other suggestions should it happen again, but for now it looks
liked I've lost all my data permanently  :-((

Thanks for looking at this.
Comment 6 Uwe Kaminski 2010-03-16 23:02:44 UTC
Thanks for reporting this.
This particular issue has already been reported in our bug tracking system, but
please feel free to report any further bugs you find. Also feel encouraged to
vote for the original bug report if interested.

Please feel free to search for existing reports first to avoid filing
duplicates.


*** This bug has been marked as a duplicate of bug 5712 ***
Comment 7 Uwe Kaminski 2010-03-16 23:21:25 UTC
The reason for this bug seems to be a duplicate but not the consequence - hence
reopening.
Comment 8 Andre Klapper maemo.org 2010-06-30 22:48:20 UTC
*** Bug 9657 has been marked as a duplicate of this bug. ***
Comment 9 Wille 2010-07-06 09:41:46 UTC
I'm pretty sure I have also experienced this but I didn’t have any chance to
look in to it when I was away. The phone was burning hot and when I finally
managed to check top it was a daemon checking the MMC card that used all of my
cpu. I don’t remember the name of it now and I forgot my card at my friends
place. I will get it back next week and then I plug it in to a computer and
check if I got the same problems.. I really hope I haven’t lost all my files…

//Marcus
Comment 10 Andre Klapper maemo.org 2010-08-18 16:52:07 UTC
"This is tricky: it is darn difficult to deal with case of external card,
granted list of reasons.
- some card manufacturers deserve to be fried for good-enough-to-sell quality
(quality barely at level NOT being returned to shop next day);
- I've seen cases when *normal* cards being swapped between Windows and Linux
machine acting nuts in one of them;
- asking for particular card in each case is the only correct (technically)
option but we have problems with both logistics (complexity and costs) and
privacy;
- factors above together with low reproduction rate make it next to impossible
to find CAUSE."