State of read-only filesystems in NAND / MTD bad blocks handling when reading

[Ricard Wanderlof]

On Wed, 2 May 2012, Thilo Fromm wrote:

>> Finally Ricard Wanderlof share his experience about bit flipping
>> during reading and concluded that "At least for this chip [ST 256
>> Mbit flash], it seemed that if a block has only been written a few
>> times, then there seems to be virtually no limit to how many times it
>> can be read without bit flips occurring ".
>
> So the bit-rot-after-many-reads-issue is maybe a non-issue?

It wasn't much of an issue when I ran tests on 256 Mbit flashes, but they 
are quite old by know, and starting to be EOL'd by manufacturers. Modern 
flashes have smaller on-chip geometries with physically smaller bit cells, 
and the chances of bits flipping is much greater, if nothing indicated by 
increased requirements in ECC strength. Sometimes the requirement changes 
have been subtle. For instance, for 256 Mbit flashes, most manufacturers 
specified the first block to be always good. For 1 Gbit flashes, this was 
changed to be a 'valid block up to 1k program/erase cycles with 1bit/512 
byte ECC', although the general ECC requirement was the same. Larger 
flashes have 4 bit ECC requirements instead of the old 1 bit ECC, and as 
chip foundries migrate to small geometries older chips get reworked to 
suit modern manufacturing methods, and as a result they also get higher 
bitflip probabilities and consequently severer ECC requirements.

So, I would consider 'read disturb' as the phenomen is normally described 
to be an issue which must be dealt with these days.

/Ricard
-- 
Ricard Wolf Wanderlöf                           ricardw(at)axis.com
Axis Communications AB, Lund, Sweden            www.axis.com
Phone +46 46 272 2016                           Fax +46 46 13 61 30