Distinguishing bitflips due to read-disturb or due to wear-out

[Artem Bityutskiy]

On Mon, 2012-03-19 at 08:02 -0400, Atlant Schmidt wrote:
>   What I *THINK* UBI is missing is the keeping of
>   appropriate statistics on which PEBs have had what
>   history of "soft" failures. UBI always lives "in
>   the here and now" and makes its decisions on the
>   1 bit of data it has at the instant (the latest
>   operation on the PEB succeeded or failed). Instead,
>   UBI should know that (e.g.) PEB 1234 ended up giving
>   it problems in 14 out of last 16 writes to that PEB
>   and maybe that PEB isn't worth using right now based
>   on the fill-ratio of the entire flash chip.

Yes, this makes sense. I think we still have few some unused bytes in
the EC header - those could be used for these purposed.

-- 
Best Regards,
Artem Bityutskiy
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part
URL: <http://lists.infradead.org/pipermail/linux-mtd/attachments/20120402/6c53333c/attachment.sig>