Hello Volker
On Tue, Jan 30, 2018 at 9:38 PM, Volker Floeder <
vf@floeder.com> wrote:
> Hello Charles,
>
> we are working on an automotive project where we are in contact with the
> German department of Micron.
>
> They did an evaluation of the project in their lab simulating automotive
> environment, i.e. temperature. The result was a data retention time of
> about 3 years, which simply isn't enough, so refreshing is needed in any
> case. We have several r/w partitions and some r/o as well.
>
> So yes, a dedicated mounting option like the one you described would be
> great.
>
> However, I think, we will need to implement a "read all nand pages" daemon
> as well.
>
I am curious about what you mean by a "read all nand pages daemon". Would
that just be something that ensures that all pages are occasionally read to
prevent bit-rot going undetected?
There is already something a bit like that for r/w partitions since Yaffs
will, occasionally, rewrite the oldest block and thus force it to refresh.
It sounds like what is needed here are:
1) A bit more active searching for bad pages.
2) Mount flags to allow fix-up for read-only.
>
> Regards,
> Volker
>
>
> ----- Ursprüngliche Mail -----
> Von: "Charles Manning" <cdhmanning@gmail.com>
> An: "Volker Floeder" <vf@floeder.com>
> CC: "Bean Huo (beanhuo)" <beanhuo@micron.com>, "yaffs" <
> yaffs@stoneboat.aleph1.co.uk>
> Gesendet: Montag, 29. Januar 2018 19:06:10
> Betreff: Re: [Yaffs] Read-Only mounted YAFFS2 - refreshing of NAND pages
>
> Hello Volker
>
>
> Read-only mounting prevents any writes to Yaffs, including any refresh.
>
> It is true that NAND does have a read disturb issue were reading a page
> can cause corruption of bits on OTHER pages in the same block. However this
> typically takes many hundreds of thousands of read operations.
>
> It is also possible that there is leakage with time.
>
> The best way to assess the impact for your design is to discuss the read
> disturb prevalence with your NAND manufacturer.
>
> Perhaps it would be good if Yaffs had a read-only mount mode that was read
> only except for doing refresh. Does that sound like something useful?
>
>
> Regards
>
> Charles
>
>
>
>
> On Mon, Jan 29, 2018 at 10:48 PM, Volker Floeder < vf@floeder.com > wrote:
>
>
> Ok, understood. Thanks!
>
> However, all pages that are used by yaffs should be read when the
> file-system is mounted, correct?
>
> So at least after a boot, there should be a read-access across all valid
> data. Or do I need some kind of "scrubbing" to be sure?
>
>
> ----- Ursprüngliche Mail -----
> Von: "Bean Huo (beanhuo)" < beanhuo@micron.com >
> An: vf@floeder.com
> CC: yaffs@stoneboat.aleph1.co.uk
> Gesendet: Montag, 29. Januar 2018 10:04:22
> Betreff: RE: [Yaffs] Read-Only mounted YAFFS2 - refreshing of NAND pages
>
> >
> >Hi,
> >
> >I am wondering if the page-refresh works on a read-only mounted filesystem
> >as well. However, if it would not, then some flash blocks might get
> >uncorrectable read-errors after some period of time, so I'd expect that to
> >work - but on the other hand, it might be a bad idea to mount read-only at
> >all ...
> >
> >Does anybody know? Thanks!
> >
> >- Volker
> >
> >
>
> According to my understanding, if there is read action on this read-only
> mounted
> Partition, page-refresh will work. Even for the read-write mounted
> filesystem, if no read action,
> page-refresh doesn't work.
> mtd layer will report bitflips to file system layer, rather than file
> system itself.
>
> >
> >
> >
> >------------------------------
> >
> >Subject: Digest Footer
> >
> >_______________________________________________
> >yaffs mailing list
> > yaffs@stoneboat.aleph1.co.uk
> > http://stoneboat.aleph1.co.uk/cgi-bin/mailman/listinfo/yaffs
> >
> >
> >------------------------------
> >
> >End of yaffs Digest, Vol 150, Issue 2
> >*************************************
>
> _______________________________________________
> yaffs mailing list
> yaffs@stoneboat.aleph1.co.uk
> http://stoneboat.aleph1.co.uk/cgi-bin/mailman/listinfo/yaffs
>
>
>
>
>
(text/plain)