[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <201103181659.46558.arnd@arndb.de>
Date: Fri, 18 Mar 2011 16:59:46 +0100
From: Arnd Bergmann <arnd@...db.de>
To: Tim Soderstrom <tim@...cowproductions.org>
Cc: Justin Piszcz <jpiszcz@...idpixels.com>, linux-usb@...r.kernel.org,
xfs@....sgi.com, linux-kernel@...r.kernel.org,
Alan Piszcz <ap@...arrain.com>
Subject: Re: 2.6.38: XFS/USB/HW issue, or failing USB stick?
On Friday 18 March 2011, Tim Soderstrom wrote:
> >
> > However, after some amount of time, the errors occur below, is this USB
> > stick failing? Since it has no SMART, is there any other way to verify
> > the 'health' of a USB stick?
>
> What prompted you to go with XFS over, say, ext2? The journal will generally
> cause quite a bit more writes onto your USB device. I use ext2 on my CF card
> in my NAS for that reason (the spinning media is on XFS of course). I know
> that's not an answer to your problem but thought I would add it as a suggestion :)
Using ext2 on flash media instead of ext3 or other file systems is
recommended a lot, but the situation is actually much more complex.
In https://lwn.net/Articles/428584/, I explain how these things work
under the cover. For a drive that can only have very few erase blocks
open, using a journaled file system will always mean thrashing, but
for drives with more open erase blocks, it's probably better to
use a journal than not.
I still need to do simulations to figure out how this exactly
ends up on various file systems, and I had not considered XFS
so far.
Getting back to the rogiinal question, I'd recommend testing the
stick by doing raw accesses instead of a file system. A simple
dd if=/dev/sdX of=/dev/zero iflag=direct bs=4M
will read the entire stick and report any errors. The corresponding
dd of=/dev/zero of=/dev/sdX oflag=direct bs=4M
writes the entire stick. Some media won't report errors on write,
though, so this might not help you at all.
I'm also interested in results from flashbench
(git://git.linaro.org/people/arnd/flashbench.git, e.g. like
http://lists.linaro.org/pipermail/flashbench-results/2011-March/000039.html)
That might help explain how the stick failed.
Arnd
Arnd
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists