Hi!
I got 32GB kingston SD card, and am using it with ext2 for storing git
trees etc.
Unfortunately, every time I run fsck, I get rather nasty corruption.
I switched it to ext3 now, but I believe I have seen corruption even
on volume marked clean, which should be impossible from user error.
If I suspect wrong block device, what are useful tests to run there?
Or maybe I should do some test on filesystem level? (So far I try
compiling kernels, but that does not seem to provoke the corruption.)
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
On Tue, Nov 4, 2008 at 12:02 PM, Pavel Machek <[email protected]> wrote:
> If I suspect wrong block device, what are useful tests to run there?
Tools like xdd allow to write certain data patterns to a block device
and to verify the written data.
Bart.
On Tue, 4 November 2008 12:02:25 +0100, Pavel Machek wrote:
>
> I got 32GB kingston SD card, and am using it with ext2 for storing git
> trees etc.
>
> Unfortunately, every time I run fsck, I get rather nasty corruption.
> I switched it to ext3 now, but I believe I have seen corruption even
> on volume marked clean, which should be impossible from user error.
>
> If I suspect wrong block device, what are useful tests to run there?
Not likely in your case, but a number of counterfeited devices are on
the market. They contain a much smaller chip inside than is advertised
plus some logic to return 0x00 when reading from non-existent memory.
To test for this, simply write 0xff to the complete device and read it
back. 'hd' is useful, as it compressed the output into four lines for a
good device and a bit more when you bought crap.
Jörn
--
Man darf nicht das, was uns unwahrscheinlich und unnatürlich erscheint,
mit dem verwechseln, was absolut unmöglich ist.
-- Carl Friedrich Gauß
> On Tue, 4 November 2008 12:02:25 +0100, Pavel Machek wrote:
> >
> > I got 32GB kingston SD card, and am using it with ext2 for storing git
> > trees etc.
> >
> > Unfortunately, every time I run fsck, I get rather nasty corruption.
> > I switched it to ext3 now, but I believe I have seen corruption even
> > on volume marked clean, which should be impossible from user error.
> >
> > If I suspect wrong block device, what are useful tests to run there?
>
> Not likely in your case, but a number of counterfeited devices are on
> the market. They contain a much smaller chip inside than is advertised
> plus some logic to return 0x00 when reading from non-existent memory.
>
> To test for this, simply write 0xff to the complete device and read it
> back. 'hd' is useful, as it compressed the output into four lines for a
> good device and a bit more when you bought crap.
Ok, I switched to ext3 and the card seems to behave now. It seems to
hold the data, so it is probably not fake :-).
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html