On Mon, Sep 19, 2011 at 04:32:25PM +0100, Matthew Garrett wrote:
> On Wed, Sep 07, 2011 at 04:29:47PM -0700, Dmitry Torokhov wrote:
> > 
> > Initializing a device connected to a serport serio port is actually task
> > of inputattach utility.
> 
> For devices that present a unique pnp ID, we probably ought to support 
> doing this in-kernel rather than relying on inputattach. It makes it 
> easier to handle things like suspend/resume.

I'd prefer to keep all this logic in one place - inputattach. On resume
inputattack returns so we only need to restart it to get input device
back. Thankfully the rest of our infrastructure is ready to handle
hot-plugged devices by now.

Thanks.

-- 
Dmitry