tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: NiLFS(2) import

Dear Toru,

On Wed, Jul 15, 2009 at 08:17:00PM +0900, Toru Nishimura wrote:
> >It is said to perform
> >very well on flash media since it is not overwriting pieces apart from a
> >incidental update of the superblock, but that might change. It is 
> >accompanied
> >by a cleaner to clean up the segments and recover lost space.
> That's a fatastic news.  How well the on-disc format copes with bare NAND
> chip parametrics; say 512B small NAND case and/or 2048B large NAND
> case?

Good question. I think it would do pretty good but for the superblock writeout
since that is written out at a standard location, not very handy for NAND.
However this could be tailored to only be written out on unmount or when the
cleaner would mess up the line to the last partial segment.

> It's fairly easy and pretty economical to use bare NAND chip for embeded
> NetBSD rootfs.  The design issue is wear-leveling, asymmetric nature of
> erase/write size and bad block management.  How is NiLFS approaching?

Wear leveling shouldn't be difficult. As of the asymetric nature of
erase/write size, a segment can be made as big as a erase block since a
segment is only sequentially written and reset by the cleaner wich can thus
also do the erasing.

As of bad block management... hmm... maybe a `dont use this blocks' list can
be added or else a segment can allways be disabled. If the segments are not
that big the only disadvantage would be a slightly bigger overhead on disc.

With regards,

Home | Main Index | Thread Index | Old Index