SDHC Corruption since HF4

[mindlord]

I have a class 6 SDHC card that up until I installed Hotfix 4 worked just great. Never a problem. Way back, I installed Hotfix 4 beta 1, and noticed that high activity on the card would cause it to suddenly remount read-only. Even simply tasks like renaming a modest batch of files would cause io problems and force the card to remount read-only. fsck-ing the card would fix any errors, but this is a constant issue. The same card worked great up until HF4 and continues to work just fine on all my Ubuntu boxes.


So, did something change in HF4 that affects ext2 and ext3 that might cause io errors? Is anyone else having similar issues?

 

[EvilDragon]

Hm, no drivers or anything have been updated with HF4 B1... I'm not sure if it included a new kernel version, but I don't think so.


Have you tried saving your data, reformat the card and put the data back?


I know some guys are using ext2 / ext3 SD Cards, and they haven't reported any problems yet.

 

[Prometheus]

Actually, I had problems with ext2 cards under Hotfix 3 (final version), but they were different problems. The Pandora would hang at the blinking cursor and fail to actually power down every time I ran PNDs from my main card (which was ext2) which meant that every time I used my Pandora I had to force a reset once it hung during shut-down and then shut down once it rebooted. It ultimately corrupted the card, and I had to switch to using FAT32.


I never had the time to look into it more, unfortunately, and I didn't feel that the info I had was sufficient - I figured that the description above wouldn't be enough information to work with (though I did mention it in passing in a thread on the other forum).


[/not quite on-topic]

 

[mindlord]

Have you tried saving your data, reformat the card and put the data back?

Numerous times. I've tried it with ext2, ext3 journaled, and ext3 non-journaled. I tested the card with a read/write verify tool on a windows machine after formatting with the Panasonic tool and it came back clean, and again Ubuntu has no issues with the card.


Edit: Oh, forgot to mention that it happens in either slot - to eliminate hardware failure.

 

[Tor]

Could it be caused by an MMC power management problem maybe? E.g. like a problem occurring with a kernel upgrade for the N800 back in 2007 (description of problem and fix below).


https://bugs.maemo.org/show_bug.cgi?id=1204#c120

 

[mindlord]

Could it be caused by an MMC power management problem maybe?

This issue seems very close or identical to the symptoms I am experiencing. I am going to try the card in a USB SDHC card reader next to see if the corruption occurs outside of SDIO.


Update: Previously, I would copy 4GB of random songs from my PC to the Music folder on the SD card (formatted as EXT2). I'd move the card over to slot 2 on the Pandora and run a script on it that would append a random string of characters to the beginning of the song names to randomize the play order. Up until HF4 this would work fine. Now it corrupts the card 100% the time. I repeated this process, except this time using a USB SDHC card reader and the card formatted to EXT2 and no corruption occured.

 

[mindlord]

A small update to this issue. I have since formatted the very same card to FAT32 and have experienced no corruption issues at all, even under extremely heavy load and when copying files under high wifi activity, or from the first slot to the second. I suspect there is something not quite right with the EXT2/3 driver in regards to certain cards.

 

[Prometheus]

^ What type of card did you experience it with?


The one of mine that got corrupted (as mentioned previously, this was way back with Hotfix 3) was a Class 6 32GB card from Verbatim (designed for HD camcorders, according to its packaging). Like you, I've experienced no problems since switching it to FAT32, either.

 

[mindlord]

It's a 16GB aData (Yeah, I know not the best brand), but this is the only time I've had problems out of it. On my Ubuntu machine it works just fine formatted as EXT2. It only exhibits this odd behavior when in the Pandora and formatted as EXT2 or 3, not NTFS or FAT32.

 

[Prometheus]

Sounds about the same as my experience. I had no problems with mine under Ubuntu (well, Kubuntu and then Xubuntu, but same thing ), either, when it was ext2 (I've not tried ext3). The issue's only on the Pandora.


Even though I know it's unlikely, I wonder if it's anything to do with speedier cards? I mean, we both used Class 6 ones and experienced the same, no?


EDIT: This reminds me of something. I'm just going to try to find it.


EDIT 2: I found it, but it wasn't what I'd thought it was. (There was a mention of the SD driver "getting confused sometimes" here, but it doesn't seem to be related to this, as far as I can tell. Putting a link here *just in case*, though.)

 

[joseLuís]

These days I experimented the same (I/O errors, remounting as read only, etc) on a sandisk 32GB card). Mainly while trying to compile the kernel, and/or copying many files simultaneously. The errors could only be 'fixed' from my laptop using the manual mode of fsck. I thought maybe the errors were caused by the card so I'm currently reinstalling everything in a new one... sigh


I must admit most of all those I/O errors heavily frustrates my intentions of having a somewhat stable dev environment in my pandi...

 

[joseLuís]

I thought those IO erros wouldn't happen when working from the card plugged to a usb card reader, man I was wrong. After going on compiling gcc for an hour the system remounted the card as readonly because of the many errors it suffered. I had to fsck it again and I've never seen more errors together. Now /lost+found/folder is full of stuff . I don't know what to do except to try to install debian or gentoo or another distribution, and see if it suffers from this IO problem.