by Andreas Kemnade

[permalink] [raw]

Subject: Re: [RFC PATCH 1/6] dt-bindings: display: imx: Add EPDC

On Sat, 12 Mar 2022 20:23:48 +0100
Jonathan Neuschäfer <[email protected]> wrote:

> Hello Andreas,
>
> Sorry for the delay, I finally got around to having a look at the
> patchset.
>
> Some comments from skimming the patches below, and in my other replies.
>
>
> On Sun, Feb 06, 2022 at 09:00:11AM +0100, Andreas Kemnade wrote:
> > Add a binding for the Electrophoretic Display Controller found at least
> > in the i.MX6.
> > The timing subnode is directly here to avoid having display parameters
> > spread all over the plate.
> >
> > Supplies are organized the same way as in the fbdev driver in the
> > NXP/Freescale kernel forks. The regulators used for that purpose,
> > like the TPS65185, the SY7636A and MAX17135 have typically a single bit to
> > start a bunch of regulators of higher or negative voltage with a
> > well-defined timing. VCOM can be handled separately, but can also be
> > incorporated into that single bit.
> >
> > Signed-off-by: Andreas Kemnade <[email protected]>
> > ---
> > .../bindings/display/imx/fsl,mxc-epdc.yaml | 159 ++++++++++++++++++
> > 1 file changed, 159 insertions(+)
> > create mode 100644 Documentation/devicetree/bindings/display/imx/fsl,mxc-epdc.yaml
> >
> > diff --git a/Documentation/devicetree/bindings/display/imx/fsl,mxc-epdc.yaml b/Documentation/devicetree/bindings/display/imx/fsl,mxc-epdc.yaml
> > new file mode 100644
> > index 000000000000..7e0795cc3f70
> > --- /dev/null
> > +++ b/Documentation/devicetree/bindings/display/imx/fsl,mxc-epdc.yaml
> > @@ -0,0 +1,159 @@
> > +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
> > +%YAML 1.2
> > +---
> [...]
> > + - vscan-holdoff
> > + - sdoed-width
> > + - sdoed-delay
> > + - sdoez-width
> > + - sdoez-delay
> > + - gdclk-hp-offs
> > + - gdsp-offs
> > + - gdoe-offs
> > + - gdclk-offs
> > + - num-ce
>
> These parameters should perhaps have sane defaults in the driver, and be
> optional in the DT.
>
First of all I think I should document them better (as said in an
earlier review mail)

I doubt there are sane defaults, in vendor kernels, there is typically a
definition of these parameters and a video mode per display.

>
> > +
> > +additionalProperties: false
> > +
> > +examples:
> > + - |
> > + #include <dt-bindings/clock/imx6sl-clock.h>
> > + #include <dt-bindings/interrupt-controller/arm-gic.h>
> > +
> > + epdc: epdc@20f4000 {
> [...]
> > +
> > + timing {
> > + clock-frequency = <80000000>;
> > + hactive = <1448>;
> > + hback-porch = <16>;
> > + hfront-porch = <102>;
> > + hsync-len = <28>;
> > + vactive = <1072>;
> > + vback-porch = <4>;
> > + vfront-porch = <4>;
> > + vsync-len = <2>;
> > + };
> > + };
>
> The way you did it here, the timing parameters are directly under the
> EPDC node in the DT, but I wonder if it would be better to have a
> separate node for the display panel, which can then provide the timing
> parameters either in the DT or in the panel driver (selected by compatible
> string of the panel).
>
IMHO it makes sense to store these timing parameters together with the
timing parameters from above. If that all somehow comes from a panel
driver, we need to design an interface for it. So for simplicity I
added the stuff just to the EPDC node.

Vendor kernel has this:
struct imx_epdc_fb_mode {
struct fb_videomode *vmode;
int vscan_holdoff;
int sdoed_width;
int sdoed_delay;
int sdoez_width;
int sdoez_delay;
int gdclk_hp_offs;
int gdsp_offs;
int gdoe_offs;
int gdclk_offs;
int num_ce;
};

So things are basically combined here.

Regards,
Andreas