(512MB) Raspberry Pi eats SD cards!


15 posts
by scruss » Thu Oct 25, 2012 3:01 am
I got a 512MB board, and put the 2012-09-18 Wheezy image on a known good 32 GB SD card. I copied the 512MB firmware (of 2012-10-21) into /boot, then started up the Raspberry Pi. After completing configuration, I rebooted and resized the root partition. Within a couple of minutes, the console was filling up with write error messages, usually (from memory) involving lots of mmc0 missing completion of DMA. Very soon, the system became unusable, and looking at the card on another machine showed that the file system was utterly hosed.

I tried the same with a different card - same results.

Tonight I tried again with a new 32GB card - but did not copy the new firmware across. It's been running fine for the last couple of hours as a 256 MB machine. I'm not seeing any of the same errors.

I was careful to copy the 512MB firmware files across, and am pretty sure they weren't corrupted. Anyone else had their 512MB board do this?
‘Remember the Golden Rule of Selling: “Do not resort to violence.”’ — McGlashan.
User avatar
Posts: 320
Joined: Sat Jun 09, 2012 12:25 pm
Location: Toronto, ON
by toxibunny » Thu Oct 25, 2012 3:12 am
Nope, but I have a camera that destroys any SD card you put into it...
note: I may or may not know what I'm talking about...
Posts: 1095
Joined: Thu Aug 18, 2011 9:21 pm
by Joe Schmoe » Thu Oct 25, 2012 3:16 am
toxibunny wrote:Nope, but I have a camera that destroys any SD card you put into it...


I have a dishwasher that does that, too.
Never answer the question you are asked. Rather, answer the question you wish you had been asked.

- Robert S. McNamara - quoted in "Fog of War" -
Posts: 2605
Joined: Sun Jan 15, 2012 1:11 pm
by daveca » Thu Oct 25, 2012 3:49 am
The main point, looking at contents on another machine and hosed, has no meaning. Whats hosed?

Realize there is a HUGE difference between using an old, abused card (or flash drive for that matter) with scratched up terminals and possibly internally damaged, as a store for an OPERATING SYSTEM.

The camera doesnt much care if it takes a couple extra mS to read a card through a flaky connector terminal, it can wreck an OS.

Ive seen it happen with Ubuntu on a flash drive. Worn USB connector = instant crash when the flash drive was wiggled slightly.
Posts: 15
Joined: Wed Oct 24, 2012 9:19 pm
by scruss » Fri Oct 26, 2012 12:07 am
Ah, as you were. Seems I downloaded a not quite complete firmware. Using rpi-update got me a working 512MB system.
‘Remember the Golden Rule of Selling: “Do not resort to violence.”’ — McGlashan.
User avatar
Posts: 320
Joined: Sat Jun 09, 2012 12:25 pm
Location: Toronto, ON
by scruss » Fri Oct 26, 2012 12:01 pm
Maybe I spoke too soon. This is what I got when trying to install some packages via apt-get:

Code: Select all
[36218.109865] mmc0: final write to SD card still running
[36228.126345] mmc0: Timeout waiting for hardware interrupt - cmd12.
[36228.127534] mmcblk0: error -110 sending stop command, original cmd response 0x900, card status 0x900
[36248.152121] mmc0: final write to SD card still running
[36258.163655] mmc0: Timeout waiting for hardware interrupt - cmd12.
[36258.164865] mmcblk0: error -110 sending stop command, original cmd response 0x900, card status 0x900
[36269.084446] mmc0: final write to SD card still running
[36279.101766] mmc0: Timeout waiting for hardware interrupt - cmd12.
[36279.102953] mmcblk0: error -110 sending stop command, original cmd response 0x900, card status 0x900
[36309.899006] mmc0: Timeout waiting for hardware interrupt - cmd25.
[36309.899047] mmc0: resetting ongoing cmd 25DMA before 4096/4096 [84]/[96] complete
[36309.902774] mmcblk0: error -110 transferring data, sector 1721928, nr 848, cmd response 0x900, card status 0xc00
[36309.902964] mmc0: DMA IRQ 6 ignored - results were reset
[36309.903200] end_request: I/O error, dev mmcblk0, sector 1722649
[36309.903227] end_request: I/O error, dev mmcblk0, sector 1722656
 ...
[36309.903462] end_request: I/O error, dev mmcblk0, sector 1722768
[36309.903823] Aborting journal on device mmcblk0p2-8.
[36310.460263] journal commit I/O error
[36310.653420] EXT4-fs error (device mmcblk0p2): ext4_journal_start_sb:327: Detected aborted journal
[36310.667118] EXT4-fs (mmcblk0p2): Remounting filesystem read-only


This is with a new Transcend Class 10 32GB card. It's no longer accepting commands:

Code: Select all
$ sync
-bash: /bin/sync: cannot execute binary file
$ sudo halt
/sbin/halt: 1: /sbin/halt: Syntax error: word unexpected (expecting ")")


These are exactly the same problems I had before.
‘Remember the Golden Rule of Selling: “Do not resort to violence.”’ — McGlashan.
User avatar
Posts: 320
Joined: Sat Jun 09, 2012 12:25 pm
Location: Toronto, ON
by LinuxDoku » Tue Oct 30, 2012 11:00 am
Same problem here, but I have the old 256MB model.

The file IO errors occours suddenly, when I work on the sd card (like deleting a directory) or running apt-get update.

16GB Hama Class 10 SDHC
The errors occour intermediately when I overclock the pi or when not overclocking after running some undefined time (minutes, hours, days,...).

2GB SanDisk MicroSD Class 4
The errors occour on the first boot, it loads the kernel, some modules and then the errors appear and the system freezes.

Does anyone has a hint? I can not buy every week new sd cards, to get my pi workin :-/
Posts: 1
Joined: Tue Oct 30, 2012 10:54 am
by apoth » Tue Nov 06, 2012 10:21 pm
I've had this working on a 256MB system for ages. 512MB for a week and it's chewed itself up as per the OP.
Posts: 10
Joined: Mon Jul 30, 2012 12:37 pm
by jamesh » Tue Nov 06, 2012 10:28 pm
Are you overclocking? We've had some reports of SD card problems when overclocking. It's being looked in to.
Soon to be unemployed software engineer currently specialising in camera drivers and frameworks, but can put mind to most embedded tasks. Got a job in N.Cambridge or surroundings? I'm interested!
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 11657
Joined: Sat Jul 30, 2011 7:41 pm
by jesusdf » Tue Nov 06, 2012 11:54 pm
I think I have read somewhere in the wiki that class 8 and 10 cards used a lower voltage and are not recommended for the pi. The best approach would be a class 4 or 6 sd card.
User avatar
Posts: 6
Joined: Thu Nov 01, 2012 8:54 pm
by scruss » Wed Nov 07, 2012 3:55 am
jamesh wrote:Are you overclocking?


No. Adding
Code: Select all
over_voltage=2
seems to have helped, though.
‘Remember the Golden Rule of Selling: “Do not resort to violence.”’ — McGlashan.
User avatar
Posts: 320
Joined: Sat Jun 09, 2012 12:25 pm
Location: Toronto, ON
by Wanderlei » Wed Nov 07, 2012 5:19 am
I had the same issue with any sort of overclock on 512MB board that it would cause corruption. (Might happen without overclock too given enough time, I havent tested it for long periods) The gpu overclock had the biggest affect, even a overclock by 1 would suffer corruption on first reboot consistently.

Same SD cards, same power supplies, same operating systems and decent overclock on 256MB board and no corruption at all.

I think it is some batches of boards particular (512MB) are not the bestest.
Posts: 79
Joined: Tue Aug 21, 2012 2:01 pm
by jamesh » Wed Nov 07, 2012 7:57 am
Odd, the only real difference is the RAM chip, which I wouldn't expect to cause that sort of issue as of itself, so perhaps in something else as well.
Soon to be unemployed software engineer currently specialising in camera drivers and frameworks, but can put mind to most embedded tasks. Got a job in N.Cambridge or surroundings? I'm interested!
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 11657
Joined: Sat Jul 30, 2011 7:41 pm
by milhouse » Wed Nov 07, 2012 8:49 am
Wanderlei wrote:Same SD cards, same power supplies, same operating systems and decent overclock on 256MB board and no corruption at all.


Whereas I had* the exact opposite situation - 256MB core_freq couldn't be overclocked at all without causing SD corruption, yet a 512MB core_freq overclocks like a champ, and all with the same SD cards, same power supply, same images etc.

Wanderlei wrote:I think it is some batches of boards particular (512MB) are not the bestest.


I don't think much has changed, other than the memory increase - it's still pot luck whether you get a good overclocker, or not. Some Pis, for whatever reason, are particularly sensitive to core_freq overclocks and I don't think this has changed (improved) with the introduction of the 512MB Pi.

*I gave the 256MB board to a friend
Posts: 555
Joined: Mon Jan 16, 2012 12:59 pm
by marionee » Sat Apr 06, 2013 8:27 am
Hi, in order to prevent sd damage you have to configure a readonly Fs and mount some folders as tmpfs.
Take a look of bemWee image http://goo.gl/hW0bC
If you want to start from scratch look nere http://wiki.debian.org/ReadonlyRoot
Posts: 1
Joined: Sat Apr 06, 2013 8:20 am