Beagleboard (automatically) remounts as RO

Hello,

I've just got my BeagleBoard working. However, when working on the Board, I suddenly get this output (in minicom) :

    root@beagleboard:~# mmcblk0: error -110 transferring data
    end_request: I/O error, dev mmcblk0, sector 3657367
    end_request: I/O error, dev mmcblk0, sector 3657375
    end_request: I/O error, dev mmcblk0, sector 3657383
    end_request: I/O error, dev mmcblk0, sector 3657391
    end_request: I/O error, dev mmcblk0, sector 3657399
    end_request: I/O error, dev mmcblk0, sector 3657407
    end_request: I/O error, dev mmcblk0, sector 3657415
    end_request: I/O error, dev mmcblk0, sector 3657423
    end_request: I/O error, dev mmcblk0, sector 3657431
    end_request: I/O error, dev mmcblk0, sector 3657439
    Aborting journal on device mmcblk0p2.
                                                                                
    ext3_abort called.
    EXT3-fs error (device mmcblk0p2): ext3_journal_start_sb: Detected aborted journal
    Remounting filesystem read-only

Google was not very helpful and gave this link - http://groups.google.com/group/beagleboard/browse_thread/thread/60d79985a34047f5
There is a mention of a patch being used, but no link to it.

Is changing the filesystem to ext2 a solution? And, is it the only solution to this problem of remounting?

Thanking you
Regards
Kedar

    end_request: I/O error, dev mmcblk0, sector 3657439
    Aborting journal on device mmcblk0p2.

    ext3_abort called.
    EXT3-fs error (device mmcblk0p2): ext3_journal_start_sb: Detected aborted journal
    Remounting filesystem read-only

Your error log suggests hardware problems on the SD card.

When shutting down a Linux system, filesystems are either unmounted or temporarily remounted as read-only (so that no pending write operations remain out there).

A non-clean shutdown (for example: power fail) will leave lots of pending write operations (not only on explicit write operations by the user); this is the reason why a clean shutdown (for example, using "/sbin/halt") is always recommended.

Every time a Linux distribution boots, a file system check is performed before every mount, updating journals and files and directories whenever possible.

Sometimes this is insufficient and the user is requested to manually fix the filesystem (using the appropriate "fsck", "e2fsck" etc commands); for example, it happens when the "fixing" could wipe out files and directories.

Sometimes it is easier to use a Linux PC to fix the filesystems of the SD card. First disable the "hald" (hardware abstraction layer daemon), if present. I disabled it on my Ubuntu box using "/etc/init.d/hal stop" (root privileges required).

Then place the card in a card-reader, and use "dmesg" to see its naming (example: "/dev/sdb" the card, "/dev/sdb2" its second partition); the card will not be mounted due to "hald" absence, so you can manually invoke the appropriate "e2fsck", "e3fsck", etc.