Charles,

Would you consider taking these patches upstream?  I wanted to make sure you
at least saw them.

The first two are just correctness fixes, and don't change the output.  So if
you wanted to be conservative in the changes, consider at least those two.

Thanks,
Chuck


On Fri, Aug 27, 2010 at 11:13:14AM -0600, Chuck Coffing wrote:
>
>I realize that mkyaffs2image is not the recommended way to create yaffs2
>images, due to the variations of OOB orderings and ECC.  Even so, I've
>found it very useful for automated sitations, such as on a nightly build
>server.
>
>But to trust it as part of the build process, I wanted to make
>mkyaffs2image more robust, and make its output a little more in line with
>the actual filesystem.  So here are 4 patches to do that:
>
>mkyaffs2image.correctness.diff:
>    Fix a variety of bugs and non-deterministic output.
>mkyaffs2image.error.diff:
>    Tighten up the error codes.
>mkyaffs2image.pad-block.diff:
>    Pad the resulting image to the block size.
>mkyaffs2image.noroot.diff:
>    Do not create the root directory entry in the image.
>
>With these changes, mkyaffs2image now runs cleanly under valgrind, and I
>have also tested the resulting images on a device with good success.
>
>I welcome any feedback, and would like to get these upstream.
>
>Thanks,
>Chuck
>
>PS - these changes probably also apply (at least conceptually) against
>mkyaffsimage.c, but I haven't tried that yet.

-- 
Chuck   /   The history of intellectual progress is written
       /   in the lives of infidels.  -- Robert G. Ingersoll