Hi,
I find the solutions here 
http://www.aleph1.co.uk/pipermail/yaffs/2005q3/001358.html.
Why this patch is not committed to CVS? Test for kernel 2.6 is wrong 
since every kernel
with latest MTD should have .eccbytes = 6 initialized

Regards,
Claudio Lanconelli

RCS file: /home/aleph1/cvs/yaffs2/yaffs_mtdif.c,v
retrieving revision 1.12
diff -u -r1.12 yaffs_mtdif.c
--- yaffs_mtdif.c    20 Sep 2005 23:14:14 -0000    1.12
+++ yaffs_mtdif.c    4 Nov 2005 13:03:06 -0000
@@ -28,9 +28,7 @@
 
 static struct nand_oobinfo yaffs_oobinfo = {
     .useecc = 1,
-#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,8))
     .eccbytes = 6,
-#endif
     .eccpos = {8, 9, 10, 13, 14, 15}
 };
 


Claudio Lanconelli wrote:

> Hi,
> Till last week I used yaffs (old yaffs1 module) without problems on my 
> Microblaze board
> with uClinux 2.4.31-uc0 and a Samsung K9F5608 standard 32MB Nand 
> flash. I wrote my own
> driver to interface the nand flash in MTD.
>
> Now I switched to latest CVS yaffs2 (in yaffs1 mode) + latest CVS MTD 
> patched for uClinux
> 2.4.31.
> If I configure YAFFS to use its own ECC calculations it seems to work 
> fine, but if I try to
> use MTD ECC calculations ( #undef CONFIG_YAFFS_DOES_ECC ) a lot of 
> error messages appears like
> this:
> **>>mtd ecc error unfixed on chunk 1:0
> **>>mtd ecc error unfixed on chunk 1:1
> **>> yaffs: getBlockInfo block 0 is not valid
> **>>Block 0 marked for retirement