Re: [PATCH v3 2/3] dt-bindings: Add header for the ingenic-drm driver bindings

[Paul Cercueil]

Hi,

Le sam. 20 avril 2019 à 14:23, Ezequiel Garcia 
<ezequiel@xxxxxxxxxxxxxxxxxxxx> a écrit :
> On Wed, 17 Apr 2019 at 10:48, Rob Herring <robh+dt@xxxxxxxxxx> wrote:
> >  On Sun, Apr 14, 2019 at 3:08 PM Paul Cercueil 
> > <paul@xxxxxxxxxxxxxxx> wrote:
> >  >
> >  > Add macros that can be used with the ingenic,lcd-mode property in 
> > the
> >  > devicetree node that corresponds to the ingenic-drm driver.
> >
> >  DRM is a Linuxism.
> >
> >  >
> >  > Signed-off-by: Paul Cercueil <paul@xxxxxxxxxxxxxxx>
> >  > Tested-by: Artur Rojek <contact@xxxxxxxxxxxxxx>
> >  > ---
> >  >
> >  > Notes:
> >  >     v2: No change
> >  >
> >  >     v3: s/_DRM//
> >  >
> >  >  include/dt-bindings/display/ingenic,drm.h | 28 
> > ++++++++++++++++++++++++++++
> >
> >  DRM is a Linuxism...
> >
> >  >  1 file changed, 28 insertions(+)
> >  >  create mode 100644 include/dt-bindings/display/ingenic,drm.h
> >  >
> >  > diff --git a/include/dt-bindings/display/ingenic,drm.h 
> > b/include/dt-bindings/display/ingenic,drm.h
> >  > new file mode 100644
> >  > index 000000000000..c749b8c346fc
> >  > --- /dev/null
> >  > +++ b/include/dt-bindings/display/ingenic,drm.h
> >  > @@ -0,0 +1,28 @@
> >  > +/* SPDX-License-Identifier: GPL-2.0 */
> >  > +/*
> >  > + *  Ingenic JZ47xx KMS driver
> >
> >  Pretty sure this file is not a KMS driver.
> >
> >  > + *
> >  > + *  Copyright (C) 2019, Paul Cercueil <paul@xxxxxxxxxxxxxxx>
> >  > + */
> >  > +#ifndef __INCLUDE_DT_BINDINGS_DISPLAY_INGENIC_DRM_H__
> >  > +#define __INCLUDE_DT_BINDINGS_DISPLAY_INGENIC_DRM_H__
> >  > +
> >  > +#define JZ_LCD_GENERIC_16BIT                   0
> >  > +#define JZ_LCD_GENERIC_18BIT                   16
> >
> >  We have some generic properties for defining the bus width already.
> >  Plus, in many cases this can be implied by the panel compatible. The
> >  exceptions are cases such as when both the ctrlr and panel support
> >  multiple modes.
>
> In the same direction as Rob's comment, I'd like to see these 
> decoupled
> from the register value.
>
> Ideally, we'd like to reuse this driver and its bindings for JZ4780,
> and these values are already different.

Well they are not any different on the JZ4780, I just checked on the 
manual.
But yes, I don't mind getting rid of that devicetree property and its
corresponding bindings file.

The major problem is that we don't have a way to map some of the values 
there
to DRM macros. I sumitted a PR to add a MEDIA_BUS_FMT_RGB888_3X8_BE bus 
format
(that I could detect to support 8-bit serial TFT panels, which is what 
the
Ben Nanonote uses), but I miss a way to detect the special* TFT panels.
I will leave these aside for now, but I want to support them eventually.

Thanks,
-Paul

* "special" TFT panels have extra pins (CLS, SPL, REV, PS) and seem to 
be a
thing from Sharp. Ingenic SoCs (as well as i.MX SoCs) have the required
hardware to drive these panels.

> Thanks!
> --
> Ezequiel García, VanguardiaSur
> www.vanguardiasur.com.ar