Re: [PATCH 1/3] media: dt-bindings: add bindings for Toshiba TC358746

[Marco Felsch <m.felsch@pengutronix.de>]

Hi Rob,

I think you didn't followed the discussion in detail so I will ask you
personal. In short the tc358746 can act as parallel-in -> csi-out or as
csi->in -> parallel-out device. The phyiscal pins are always the same
only the internal timings are different. So we have two ports with two
endpoints.

Now the question is how we determine the mode. We have two approaches:
1)
  port@0 -> input port
  port@1 -> output port

  pro:
  + no extra vendor specific binding is needed to determine the mode

  con:
  - input/output endpoint can be parallel or mipi-csi2.

2)
  port@0 -> parallel port
  port@1 -> mipi-csi2 port

  pro:
  + input/output endpoint are fixed to parallel or mipi

  con:
  - vendor specific binding is needed to determine the mode

Thanks for your comments :)

Regards,
Marco

On 19-03-05 09:49, Jacopo Mondi wrote:
> Hi Sakari, Marco,
> 
> On Mon, Mar 04, 2019 at 08:17:48PM +0200, Sakari Ailus wrote:
> > Hi Marco,
> >
> > On Mon, Mar 04, 2019 at 05:55:28PM +0100, Marco Felsch wrote:
> > > > > (more device specific)
> > > > > tc358746,default-mode = <CSI-Tx> /* Parallel-in -> CSI-out */
> > > > > tc358746,default-mode = <CSI-Rx> /* CSI-in -> Parallel-out */
> > > > >
> > > > > or
> > > > >
> > > > > (more generic)
> > > > > tc358746,default-dir = <PARALLEL_TO_CSI2>
> > > > > tc358746,default-dir = <CSI2_TO_PARALLEL>
> > > >
> > > > The prefix for Toshiba is "toshiba". What would you think of
> > > > "toshiba,csi2-direction" with values of either "rx" or "tx"? Or
> > > > "toshiba,csi2-mode" with either "master" or "slave", which would be a
> > > > little bit more generic, but could be slightly more probable to get wrong
> > > > as well.
> > >
> > > You're right mixed the prefix with the device.. If we need to introduce
> > > a property I would prefer the "toshiba,csi2-direction" one. I said if
> > > because as Jacopo mentioned we can avoid the property by define port@0
> > > as input and port@1 as output. I tink that's the best solution, since we
> > > can avoid device specific bindings and it's common to use the last port
> > > as output (e.g. video-mux).
> >
> > The ports represent hardware and I think I would avoid reordering them. I
> > wonder what would the DT folks prefer.
> >
> 
> I might have missed why you mention re-ordering? :)
> 
> > The device specific property is to the point at least: it describes an
> > orthogonal part of the device configuration. That's why I'd pick that if I
> > were to choose. But I'll let Rob to comment on this.
> 
> That's true indeed. Let's wait for inputs from DT people, I'm fine
> with both approaches.
> 
> Thanks
>    j
> 
> >
> > --
> > Regards,
> >
> > Sakari Ailus
> > sakari.ailus@linux.intel.com



-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |