stable MTD snapshot

[David Woodhouse]

On Wed, 2004-06-02 at 17:48 -0500, Zhang, Jian wrote:
> My reply to David's question has been bounced back twice. I still don't
> know how to configure outlook to reply in the way the list server likes.
> So I restart it.

Either use it in IMAP mode (you can do that even while it's talking to
the Exchange server, I'm told) or use a real, standards-compliant mail
program instead. Starting a new broken-thread without 'Re:' in the
subject line deliberately to get round the automatic filters will just
get you into the blacklist. :)

Talk to your IS department. Surely they have a responsibility to provide
you with a configuration which allows you to behave as a responsible
citizen in your interaction with the outside world?

> I have tried a few snapshots and encountered either compile failure or
> run-time errors. Can someone point me to a daily MTD snapshot that is
> known to pass compilation and do not have major issue with JFFS2 on
> NAND?

There's been quite a bit of work on the NAND drivers recently. Look at
the linux-mtd-cvs list archives and pick a date before the recent flurry
of activity, or try current CVS.

> 20040518:
> I was able to mount an empty NAND partition. The partition was
> erased/prepared by U-Boot. JFFS2 clean marker was programmed to OOB
> area. I could mount it cleanly. But I got message "jffs2: No clean,
> dirty _or_ erasable blocks to GC from! Where are they all" when I
> created a file. A 192K file was truncated after umount/re-mount. Files
> even disappeared after umount/re-mount.

If you can reproduce this today, it would be interesting to do so with
'CONFIG_JFFS2_FS_DEBUG=1' while you're writing and subsequently
remounting and trying to read the file. 

> 20040526:
> Kernel doesn't build because nand.c has compile error in
> single_erase_cmd()

Less interesting.

> 20040531:
> Kernel builds but I got I/O error when mounting a partition. The
> underlying MTD NAND driver works since I am able to read/write MTD
> nodes.

Can you be more precise about what happens? 

-- 
dwmw2