Re: [PATCH v3] serio: add support for PS2Mult multiplexer protocol

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

On Mon, Oct 18, 2010 at 08:56:27AM -0700, Dmitry Torokhov wrote:
> On Mon, Oct 18, 2010 at 03:24:51PM +0400, Dmitry Eremin-Solenikov wrote:
> > Hello,
> > 
> > On 14.10.2010 18:23, Dmitry Torokhov wrote:
> > >On Fri, Oct 08, 2010 at 12:50:34PM +0400, Dmitry Eremin-Solenikov wrote:
> > >>Hello,
> > >>
> > >>On Thu, Oct 7, 2010 at 8:36 PM, Dmitry Torokhov
> > >><dmitry.torokhov@gmail.com>  wrote:
> > >>>On Thu, Oct 07, 2010 at 07:19:57PM +0400, Dmitry Eremin-Solenikov wrote:
> > >>>>Hello,
> > >>>>
> > >>>>On Thu, Sep 30, 2010 at 10:25 AM, Dmitry Torokhov
> > >>>><dmitry.torokhov@gmail.com>  wrote:
> > >>>>>On Wed, Sep 29, 2010 at 04:45:53PM +0400, Dmitry Eremin-Solenikov wrote:
> > >>>>>>On Thu, Sep 23, 2010 at 8:44 PM, Dmitry Eremin-Solenikov
> > >>>>>><dbaryshkov@gmail.com>  wrote:
> > >>>>>>>PS2Mult is a simple serial protocol used for multiplexing several PS/2 streams
> > >>>>>>>into one serial data stream. It's used e.g. on TQM85xx serie of boards.
> > >>>>>>>
> > >>>>>>>Signed-off-by: Dmitry Eremin-Solenikov<dbaryshkov@gmail.com>
> > >>>>>>>---
> > >>>>>>>
> > >>>>>>>It actually depends on "serio: multiple children" patch. I'm not resending it
> > >>>>>>>as you were the originator of the latest version of the patch.
> > >>>>>>
> > >>>>>>So, what about this version of patch?
> > >>>>>>
> > >>>>>
> > >>>>>Looks better but I think you also need ->start() to make sure you do not
> > >>>>>try to deliver events too early. Does the following still work for you?
> > >>>>
> > >>>>Sorry for the delay. Crashes w/o the attached patch.
> > >>>>
> > >>>
> > >>>Ah, I see, however what I actually wanted is to create ports before hand
> > >>>and handle any errors that might arise and then enable the device and
> > >>>register child ports.
> > >>>
> > >>>If you apply the patch below instead of yours does it still work?
> > >>
> > >>It won't work, as we don't set psm->ports[i].serio before ps2mult_serio_start()
> > >>
> > >
> > >Indeed. I guess we need to set the pointer immediately then and also add
> > >a flag to the port structure telling whether it has been registered.
> > >Does the following work for you by any chance?
> > 
> > Yeah, works w/o any glitches. Thank you.
> > 
> 
> Thanks Dmitry.
> 

BTW, is there corresponding patches to inputattach?

-- 
Dmitry