Current mtd tree is broken

Josh Boyer jwboyer at gmail.com
Thu Jun 22 07:54:16 EDT 2006


On 6/22/06, Artem B. Bityutskiy <dedekind at infradead.org> wrote:
> On Wed, 2006-06-21 at 17:35 -0500, Josh Boyer wrote:
> > As panto pointed out on IRC today, the current MTD git tree is
> > completely broken for quite a few chip types.  Anything older than
> > Intel Sibley or P30 is broken since writesize is not set to something
> > other than 0 in the other cases.  This causes a BUG to trigger in
> > add_mtd_device because someone decided writesize == 0 was not a sane
> > thing.
>
> How about this?
>
> writesize is a new feature and many drivers may not initialize it. It
> might be harmless, depending on the usage of the driver. So don't be
> that difficult and issue warning instead of getting into panic.

I don't like that very much.  It's laziness.  An interface was added
and working.  Someone changed the semantics of it, but didn't fix all
the things effected.

If it's just changed to a WARN, it will take a really long time for
everything to get fixed and it will quickly become an FAQ item.  "My
$random driver used to work and now it gets a warn on, what do I do?"
Just fix it now.

josh




More information about the linux-mtd mailing list