>Hi Stephane
>I tried to reproduce this but could not.
>I wrote a test which should do what you are reporting. Can you please have a look at:
>http://yaffs.net/gitweb?p=yaffs2.git;a=shortlog;h=refs/heads/stephanetest
>and see if that represents your test.

Hi, thanks for having a look.
This is not exactly what I tested (unmount instead of 'power failure'),
but I still have the problem here on the last step.

I pretty sure my low-level NAND functions (or file simulation) work correctly, so this must be a bad configuration.
I attached the output of the test with debug, and my yaffs adaptation layer.