On Wednesday 16 November 2011 14:26:21 Chris wrote:
> Linux 2.6.30
> yaffs_guts.c,v 1.93 2009/11/07 02:06:58 charles
> I have 2 : 32GiB Micron NAND chips. 4096 blocks, 128 blocks/page
>
> Here is what i have:
> i have an mtdblock6 partition which is ~8 GiB. I mount using the flags "-t
> yaffs -sync". So my mount point is /mnt/mtdblock6. I used "dd" to create a
> file ~4GiB called "/mnt/mtdblock6/public". This file is formatted as fat32.
> Everything seemed to work fine until i began stressing its limits. I have a
> script file which does the following steps 3 times:
> 1) write files
> 2) compare files
> 3) delete files
> The first two iterations passed with no problems, the third iteration
> experienced some problems. Main problem being the "public" file contains a
> bunch of "FF" and "00" where the master boot record for fat32 should be. It
> seems the first two iteration would have probably touched nearly every
> chuck in yaffs, leaving the third to do a bunch of garbage collection. Does
> this seem resonable, if so where could i start to debug?
>

Single yaffs files of 4GiB are probably a problem. 

See if the problem goes away with a yaffs file size of <2G.

-- Charles