Have a read of http://users.actrix.co.nz/manningc/yaffs/HowYaffsWorks.pdf

Yaffs has knowledge  about bad blocks etc so that it knows which blocks to 
skip etc.

yaffs does not need mtd to store a bad block table, but some NAND parts need a 
bbt as part of their bad block handling strategy.

The MTD UBI layer also needs a BBT etc. If you run mtd simulation on top of 
UBI then you end up getting "perfect NAND" because UBI does all the fiddling 
to hide bad blocks etc. THis is overkill for yaffs, which does its own bad 
block management, and will hurt you performance wise.

-- Charles.


On Thursday 26 November 2009 08:09:50 Eddie Dawydiuk wrote:
> Hi all,
>
> The other day someone asked me a question regarding the interface between
> Yaffs2 and MTD regarding ECC and bad block handling. I know one has the
> option to let Yaffs do the ECC, or it can be handled by MTD. I also know
> Yaffs can retire blocks and keep track of what blocks are bad. Further I
> know MTD has a similar mechanism(Bad block table). So my question is if one
> configures MTD to handle ECC as well as to keep track of bad blocks(via bad
> block table), then does Yaffs have any feedback(or does Yaffs even care)
> regarding bad blocks, for instance when bad blocks are discovered and what
> blocks are marked bad? I'm really looking from a high level
> design/architecture point of view. Could someone point me in the right
> direction on this topic?
>
> Thanks in advance for any information.