Re: [PATCH 1/2] dt-bindings: display: panel: Add Visionox R66451 AMOLED DSI panel bindings

[Neil Armstrong]

On 22/05/2023 16:51, Marijn Suijten wrote:
> On 2023-05-22 11:05:38, Neil Armstrong wrote:
> > On 21/05/2023 12:30, Marijn Suijten wrote:
> > > On 2023-05-16 13:20:30, Jessica Zhang wrote:
> > > > Document the 1080x2340 Visionox R66451 AMOLED DSI panel bindings
> > > >
> > > > Signed-off-by: Jessica Zhang <quic_jesszhan@xxxxxxxxxxx>
> > > > ---
> > > >    .../bindings/display/panel/visionox,r66451.yaml    | 59 ++++++++++++++++++++++
> > > >    1 file changed, 59 insertions(+)
> > > >
> > > > diff --git a/Documentation/devicetree/bindings/display/panel/visionox,r66451.yaml b/Documentation/devicetree/bindings/display/panel/visionox,r66451.yaml
> > > > new file mode 100644
> > > > index 000000000000..6ba323683921
> > > > --- /dev/null
> > > > +++ b/Documentation/devicetree/bindings/display/panel/visionox,r66451.yaml
> > > > @@ -0,0 +1,59 @@
> > > > +# SPDX-License-Identifier: GPL-2.0-only or BSD-2-Clause
> > > > +%YAML 1.2
> > > > +---
> > > > +$id: http://devicetree.org/schemas/display/panel/visionox,r66451.yaml#
> > > > +$schema: http://devicetree.org/meta-schemas/core.yaml#
> > > > +
> > > > +title: Visionox R66451 AMOLED DSI Panel
> > > > +
> > > > +maintainers:
> > > > +  - Jessica Zhang <quic_jesszhan@xxxxxxxxxxx>
> > > > +
> > > > +allOf:
> > > > +  - $ref: panel-common.yaml#
> > > > +
> > > > +properties:
> > > > +  compatible:
> > > > +    const: visionox,r66451
> > > > +
> > > > +  reg:
> > > > +    maxItems: 1
> > > > +    description: DSI virtual channel
> > > > +
> > > > +  vddio-supply: true
> > > > +  vdd-supply: true
> > > > +  port: true
> > > > +  reset-gpios: true
> > >
> > > Normally for cmd-mode panels there is also a `disp-te` pin which is
> > > optionally registered in dsi_host.c as GPIOD_IN, but on **ALL** my Sony
> > > phones this breaks vsync (as in: mdp5 stops receiving the interrupt, but
> > > we can see disp-te in /proc/interrupts then).
> >
> > Describing it as a gpio is wrong, it should be described as a pinctrl state instead.
>
> We defined both in our DTS, what weirdness does it cause when then
> requested using GPIOD_IN?  It'd still be beneficial to see the vsync
> interrupt raise in /proc/interrupts (but it's just a waste of CPU cycles
> OTOH, this is all handled in the MDP hardware after all, so it's not
> something I'd like to enable by default).

Sure, but it's a sw hack, the pin has a TE function which directly goes to
the DSI logic, claiming it as a GPIO will set it as GPIO function.

On some platforms, PINMUX is only on output and input is always directed
to all HW blocks, seems it's not the case here !

> Anyway, this is what we ended up doing to "fix" the bug (only bias the
> pin via pinctrl, omit the disp-te DTS property).  Thanks for confirming!
>
> - Marijn
>
> > Neil
>
> <snip>