On Mon, Dec 17, 2007 at 02:01:33PM +0100, Joerg Sonnenberger wrote:

> That adds a lot of complication for very little gain though.

It looks like a simple matter of programming to me.

Using a general purpose driver entrypoint that allows failure and a limited
form of versioning is one way to deal with new requirements like this.
Altering the established interfaces works, and so does using a lame hack -
that way lies the path of least resistance. I know that time is limited, but
the easy solutions do nothing to address interface churn and the old
complaint that "LKMs suck".

Andrew