Data corrupted

Artem B. Bityuckiy dedekind at infradead.org
Thu Mar 3 08:29:17 EST 2005


Bobzin,Heiko wrote:
> Hi Artem,
> it still happens with your patch.
Since your flash is plain NOR it should, I see now. It shoul help for 
ECC'd NOR and another flashes which celanmarker size != 0 and c->wbuf != 
NULL, though.

> 
> I first get
> 
> Node header CRC failed at 00510760. But it must have been OK earlier.
> Node was: { 0000, 2000, 00000000, 00000000 }
> 
> The bad node was in the 'inetd', which core-dumped then.
> 
> After a second power-cycle I get another one:
> 
> Node header CRC failed at 005d3a94. But it must have been OK earlier.
> Node was: { 0000, 2000, 00000000, 00000000 }
> 
> The bad node is in a 1.5M application file.
> 
> After the third power-cycle I get 3 of those, then 1, and later
> the test application itself gets corrupted.
> 

Well, I have no ideas why this can happen.

Is it possible to reproduce the error if JFFS2 debugging level 1 is on ?

--
Best Regards,
Artem B. Bityuckiy,
St.-Petersburg, Russia.




More information about the linux-mtd mailing list