Re: Response to a Write Error

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, Apr 14, 2010 at 12:51 PM, Ryusuke Konishi <ryusuke@xxxxxxxx> wrote:

> I have no idea why the corruption occured since the 2.6.31-20 Ubuntu
> kernel is almost the same with the 2.6.31.y stable kernel.

I believe that the error was caused by accumulated defects on the
DVD-RAM disk, not by NILFS2.  I interpret the error messages to mean
that the DVD drive detected a hardware error when it tried to read
sector 8778160 and then had a write error at sector 8781280 that could
not be correct by the drive because there were no more spare
defect-correction sectors on that particular DVD-RAM disk.

> I'd appreciate it if you could help to test whether the same problem
> arises with nilfs in 2.6.32.y or 2.6.33.y.

The problem does not seem to be repeatable, even on 2.6.31.  I
detected no errors when I previously ran the same general test many
times using both that DVD-RAM and a loop-back file system on a hard
drive.

I will try again when I upgrade to 2.6.32 with Ubuntu 10.04.

> The following are the steps to make&install a newer nilfs in the
> 2.6.31 kernel.
>
>  $ sudo apt-get install linux-headers-`uname -r`
>  $ cd /usr/src
>  $ wget http://www.kernel.org/pub/linux/kernel/v2.6/linux-2.6.32.11.tar.bz2
>  $ tar jxf linux-2.6.32.11.tar.bz2
>  $ cd linux-2.6.32.11/fs/nilfs2
>  $ make -C /lib/modules/`uname -r`/build SUBDIRS=$PWD BUILD_DIR=$PWD modules
>  $ su
>  # rmmod nilfs2
>  # mv /lib/modules/`uname -r`/kernel/fs/nilfs2/nilfs2.ko{,.bak}
>  # install -m 644 -c nilfs2.ko /lib/modules/`uname -r`/kernel/fs/nilfs2/
>  # /sbin/depmod -A
>  # mount -t nilfs2 <device> <mount-dir>

Thank you for clarifying that process.  It will be very useful in
keeping the kernel module up-to-date.

> As for the write errors, the following two patches may work:
>
> "nilfs2: use GFP_NOIO for bio_alloc instead of GFP_NOWAIT" (merged to v2.6.32)
> "nilfs2: do not return io error for bio allocation failure" (merged to v2.6.33)
>
> It's hard to track the cause from the raw block data, so you don't
> have to keep the DVD-RAM data.  (dumpseg log for in-use segments is
> enough to me)

I will send a segment dump off-list.

If my analysis is correct, I have a DVD-RAM disk that has no more
sectors with which my DVD drive can correct defects.  Would you like
for me to save that disk for future error handling experiments?

Ed
--
To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[Index of Archives]     [Linux Filesystem Development]     [Linux BTRFS]     [Linux CIFS]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux SCSI]

  Powered by Linux