On Thursday 07 April 2011 19:51:54 Boaz Ben-David wrote:
> Hi,
>
> This indeed seems like a problem in our kernel, but this brings me to
> another question.
> It seems that YAFFS is marking a lot of blocks as bad, the board itself
> is very new so this couldn't be a bad NAND chip and it happens on all of
> our boards.
>
> Is there any debugging option that can help me find out why YAFFS is
> marking blocks as bad?
>
> Also, could this be somehow related to the fact I am using YAFFS on a
> MLC device? Is there some kernel option I need to enable/disable?

MLC per se should not be a problem. YAFFS ships on millions of MLC devices.

This is almost guaranteed to be an issue in the driver getting stuff messed 
up.

Have a snoop around Linux MTD. I think there are some mtd test tools that 
might help.

The other thing to consider is to force yaffs into doing readback verification 
and checking if that makes the problem more clear.

-- CHarles