Re: [PATCH V7 1/4] Documentation/devicetree/bindings: b850v3_lvds_dp

[Laurent Pinchart]

Hi Peter,

On Thursday 19 Jan 2017 09:12:14 Peter Senna Tschudin wrote:
> On Wed, Jan 18, 2017 at 11:10:58PM +0200, Laurent Pinchart wrote:
> > On Monday 16 Jan 2017 09:37:11 Peter Senna Tschudin wrote:
> >> On Tue, Jan 10, 2017 at 11:04:58PM +0200, Laurent Pinchart wrote:
> >>> On Saturday 07 Jan 2017 01:29:52 Peter Senna Tschudin wrote:
> >>>> On 04 January, 2017 21:39 CET, Rob Herring wrote:
> >>>>> On Tue, Jan 3, 2017 at 5:34 PM, Peter Senna Tschudin wrote:
> >>>>>> On 03 January, 2017 23:51 CET, Rob Herring <robh@xxxxxxxxxx> wrote:
> >>>>>>> On Sun, Jan 01, 2017 at 09:24:29PM +0100, Peter Senna Tschudin 
wrote:
> >>>>>>>> Devicetree bindings documentation for the GE B850v3 LVDS/DP++
> >>>>>>>> display bridge.
> >>>>>>>> 
> >>>>>>>> Cc: Martyn Welch <martyn.welch@xxxxxxxxxxxxxxx>
> >>>>>>>> Cc: Martin Donnelly <martin.donnelly@xxxxxx>
> >>>>>>>> Cc: Javier Martinez Canillas <javier@xxxxxxxxxxxx>
> >>>>>>>> Cc: Enric Balletbo i Serra <enric.balletbo@xxxxxxxxxxxxx>
> >>>>>>>> Cc: Philipp Zabel <p.zabel@xxxxxxxxxxxxxx>
> >>>>>>>> Cc: Rob Herring <robh@xxxxxxxxxx>
> >>>>>>>> Cc: Fabio Estevam <fabio.estevam@xxxxxxx>
> >>>>>>>> Signed-off-by: Peter Senna Tschudin <peter.senna@xxxxxxxxxxxxx>
> >>>>>>>> ---
> >>>>>>>> There was an Acked-by from Rob Herring <robh@xxxxxxxxxx> for V6,
> >>>>>>>> but I changed the bindings to use i2c_new_secondary_device() so I
> >>>>>>>> removed it from the commit message.
> 
> [...]
> 
> >>>>>>>>  .../devicetree/bindings/ge/b850v3-lvds-dp.txt      | 39 +++++++++
> >>>>>>> 
> >>>>>>> Isn't '-lvds-dp' redundant? The part# should be enough.
> >>>>>> 
> >>>>>> b850v3 is the name of the product, this is why the proposed name.
> >>>>>> What about, b850v3-dp2 dp2 indicating the second DP output?
> >>>>> 
> >>>>> Humm, b850v3 is the board name? This node should be the name of the
> >>>>> bridge chip.
> >>>> 
> >>>> From the cover letter:
> >>>> 
> >>>> -- // --
> >>>> There are two physical bridges on the video signal pipeline: a
> >>>> STDP4028(LVDS to DP) and a STDP2690(DP to DP++).  The hardware and
> >>>> firmware made it complicated for this binding to comprise two device
> >>>> tree nodes, as the design goal is to configure both bridges based on
> >>>> the LVDS signal, which leave the driver powerless to control the
> >>>> video processing pipeline. The two bridges behaves as a single bridge,
> >>>> and the driver is only needed for telling the host about EDID / HPD,
> >>>> and for giving the host powers to ack interrupts. The video signal
> >>>> pipeline
> >>>> 
> >>>> is as follows:
> >>>>   Host -> LVDS|--(STDP4028)--|DP -> DP|--(STDP2690)--|DP++ -> Video
> >>>>   output
> >>>> 
> >>>> -- // --
> >>> 
> >>> You forgot to prefix your patch series with [HACK] ;-)
> >>> 
> >>> How about fixing the issues that make the two DT nodes solution
> >>> difficult ? What are they ?
> >> 
> >> The Firmware and the hardware design. Both bridges, with stock firmware,
> >> are fully capable of providig EDID information and handling interrupts.
> >> But on this specific design, with this specific firmware, I need to read
> >> EDID from one bridge, and handle interrupts on the other.
> > 
> > Which firmware are you talking about ? Firmware running on the bridges, or
> > somewhere else ?
> 
> Each bridge has it's own external flash containing a binary firmware.
> The goal of the firmware is to configure the output end based on the
> input end. This is part of what makes handling EDID and HPD challenging.
> 
> >> Back when I was starting the development I could not come up with a
> >> proper way to split EDID and interrupts between two bridges in a way
> >> that would result in a fully functional connector. Did I miss something?
> > 
> > You didn't, we did :-) I've been telling for quite some time now that we
> > must decouple bridges from connectors, and this is another example of why
> > we have such a need. Bridges should expose additional functions needed to
> > implement connector operations, and the connector should be instantiated
> > by the display driver with the help of bridge operations. You could then
> > create a connector that relies on one bridge to read the EDID and on the
> > other bridge to handle HPD.
> 
> Ah thanks. So for now the single DT node approach is acceptable, right?
> The problem is that even if the driver is getting better on each
> iteration, the single DT node for two chips issue comes back often and I
> believe is _the_ issue preventing the driver from getting upstream. V1
> was sent ~ 8 months ago...
> 
> Can I have some blessing on the single DT node approach for now?

With the "DT as an ABI" approach, I'm afraid not. Temporary hacks are 
acceptable on the driver side, but you need two nodes in DT.

> I'm one of the 3 proposed maintainers for the driver, and I'm willing to
> maintain the driver on the long run, as is the same with the other two
> proposed maintainers. So when the time to split the node in two comes,
> we will be around, and willing to do it ourselves.

How about putting that team of 3 maintainers to work on fixing the problem in 
the bridge API ? :-)

-- 
Regards,

Laurent Pinchart