[PATCH/RFC] Linux MTD striping middle layer

[Artem B. Bityutskiy]

On Tue, 2006-03-21 at 15:36 +0300, Belyakov, Alexander wrote:
> Hello,
> 
> attached diff file is a patch to be applied on MTD snapshot 20060315
> introducing striping feature for Linux MTD. Despite striping
> is well known feature is was not implemented in MTD for some reason.
> We did it and ready to share with community. Hope, striping will find
> its
> place in Linux MTD.

More questions about how do you handle bad blocks. I do not look to the
source code so far. Actually I believe you should write a short text
file with some documentation...

1. In case of NAND concatenation, what do you do with bad eraseblocks?
Say, you stripe 2 NAND flashes, and the fist one has bad eraseblock 0,
and the second one has bad eraseblock 1? Am I right that in this case
you'll just waste eraseblock 0 of chip 1 and eraseblock 1 of chip 0?

2. Suppose we have a stripped device mtd2 which stripes mtd0 and mtd1.
Suppose user calls mtd2->block_mark_bad(N) (the block_mark_bad(N) method
of the mtd2 device). Your actions? Will you mark eraseblock N of both
mtd0 and mtd1 as bad physically? Note, actually only one of them became
bad...


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