Re: [Yaffs] YAFFS2 - sequence number overflow?

On Monday 28 August 2006 19:01, snuggl3s wrote:
> Hello all,
>
> currently I'm trying to familiarise oneself with the YAFFS2 concepts,
> especially
> the idea, how to accomplish the sequential write restriction of future
> FLASH chips.
>
> YAFFS2 uses a 4 byte sequence number to differ older chunks from the latest
> one.
> This really allows a lot of Blocks to be written, but how does YAFFS handle
> a sequence number overflow?

Currently it doesn't.

>
> On a 512MB FLASH memory this will surely happen
> before reaching the FLASH limitations of about 100 000 Block rewrite
> operations,
> even with perfect wear leveling.
>
> Is there a mechanism to prevent a sequence number overflow?
Not at present, but one could be devised.

However.... let's put this number in perspective.

We can write 2^32 blocks before this mechanism breaks down. Each block will be
128k bytes or greater.

For 128k blocks that works out to be approx 1Mbyte/sec sustained for almost 18
years. That's quite a bit when compared to the expected lifetime of most
devices that would use yaffs and I would suggest this is not going to be the
constraining factor in any real products lifetime.

If that starts to seem marginal, then you can make bigger blocks and extend
the mechanism's life.

-- Charles