Re: data corruption with nvidia chipsets and IDE/SATA drives // memory hole mapping related bug?!

[Christoph Anton Mitterer <calestyo@scientia.net>]

Chris Wedgwood wrote:
> Heh, I see this also with an Tyan S2866 (nforce4 chipset).  I've been
> aware something is a miss for a while because if I transfer about 40GB
> of data from one machine to another there are checksum mismatches and
> some files have to be transfered again.
>   
It seems that this may be occur on _all_ nvidia chipsets (of course I'm
talking about mainboard chips ;) )
Which harddisk types to you use (vendor an interface type)

> I've kept quite about it so far because it's not been clear what the
> cause is and because i can mostly ignore it now that I checksum all my
> data and check after xfer that it's sane (so I have 2+ copies of all
> this stuff everywhere).
>   
I assume that a large number of users actually experience this error,..
but as it's so rare only few correctly identify it.
Most of them might think that its filesystem related or so.



>> The corrupted data is not one single completely wrong block of data
>> or so,.. but if you look at the area of the file where differences
>> are found,.. than some bytes are ok,.. some are wrong,.. and so on
>> (seems to be randomly).
>>     
>
> For me it seems that a single block in the file would be bad and the
> rest OK --- we I'm talking about 2 random blocks in 30BG or so.
>   
Did you check this with an hex editor? I did it an while the errors were
restricted to one "region" of a file.... it was not so that that region
was completely corrupted but only some single bytes.
Actually it was that mostly one bit was wrong,..


Chris.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/