>
> On Wednesday 29 August 2012 04:47:19 Jeffrey Johnston wrote:
> > I have a client that is currently using a version of YAFFS from about
> mid
> > 2007. They would like to upgrade to the latest version of YAFFS.
> > All of their equipment in the field typically gets its load updated every
> > year with about 10 to 100 Meg of database stored in the filesystem. I
> > have tested upgrading the file system from the old Yaffs to the latest
> > YAFFS and have not seen any problems.
> > Are there any pitfalls to upgrading the filesystem in the field (with
> > existing contents) that I should be aware of?
> > I have already verified that downgrades will not work.
> > Jeff Johnston
>
> Upgrades should just work.
>
> Downgrades should also work if you disable a few things. The older code
> does
> not have support for block summaries for instance. Checkpoints are visioned
> and should be handled properly (ie. non-,atching versions are ignored).
>
> What did you find that did not work in the downgrade?
>
>
>
During the downgrade YAFFS would issue a message complaining about a
Partially written block being detected and marking it for retirement. The
message was issued from yaffs_ScanBackwards. The checkpoint data was
always invalidated because the YAFFS_CHECKPOINT_VERSION changed from 3 to 4.
I did not determine why the sequence number did not match the block that
was the current allocation block.
Jeff Johnston
(text/plain)