[i2c] [PATCH] i2c: Clean up core locking

[Jean Delvare]

On Fri, 07 Dec 2007 10:04:46 -0800, David Brownell wrote:
> > Hmm, that's right, good catch. Damn legacy drivers :( What a pain.
> >
> > I've been thinking about it for 2 hours now, and it turns out that I
> > simply don't understand it all good enough to make safe changes in this
> > area. It's too complex for me it seems
> 
> Wouldn't it suffice to just add a flag to the driver saying whether
> an attach_adapter() call would be valid?  You can prevent the loop
> fairly directly, rather than by code flow.  Solving this puzzler by
> codeflow morphing is probably a losing game.

With what benefit? I thought that we could clean up the locking easily,
but I was wrong. Unless you see something wrong with the current code,
I'd leave it as is. There are so many other areas that need work, I
don't really want to spend more time on this without a good reason.

-- 
Jean Delvare