I do
	flash_eraseall /dev/mtd12
	mount -t yaffs2 /dev/mtdblock12 /mnt

	then

	cd /mnt
	echo "this is a test" > test

	ls
	garbage-entry-skdafd;jasdf;jasdf;ljasdf;ljasdf;ljasdflkjasdf;lajsdf;lajsdfjalsdfadfsj
	test

If I then umount and remount, it fails as the directory has been corrupted.

The test you propose works sometimes. At other times, files 'a', 'b' etc just aren't there on the remount or still other times, the mount will fail.

Why do you use the oob so heavily for yaffs2? I have hardware ecc that controls data in this area. Example:

	_ReadChunkWithTagsFromNAND

calls part_read_oob right off.

JFFS2 runs fine with the mtd I have (2.6.27.4).

Does yaffs write directories from a specific area in the code I could track?

Cal Page

On Wednesday 05 November 2008 03:00:39 Cal Page wrote:

> > Hi, I'm trying to debug why my yaffs2 clobbers directories. Is there any
> > architecture document for yaffs2?
>   
A reasonably up to date doc can be found at:
http://users.actrix.co.nz/manningc/yaffs/HowYaffsWorks.pdf

To try to separate the various things that might be going wrong, I suggest you 
start off with a very simple test. Something like:

mount

touch /yaffs/a
mkdir /yaffs/b
touch /yaffs/b/a

umount
mount

Does it pass simple tests and only fail on more complex tests?
It also helps if you describe your system better, including the state 
of /proc/yaffs.

-- Charles


_______________________________________________
yaffs mailing list
yaffs@lists.aleph1.co.uk
http://lists.aleph1.co.uk/cgi-bin/mailman/listinfo/yaffs