Re: [PATCH] spi: sh-msiof: Enforce fixed DTDL for R-Car H3

[Geert Uytterhoeven <geert@xxxxxxxxxxxxxx>]

Hi Wolfram,

On Thu, Jan 26, 2023 at 2:07 PM Wolfram Sang
<wsa+renesas@xxxxxxxxxxxxxxxxxxxx> wrote:
> > > Documentation says only DTDL of 200 is allowed for this SoC.
> >
> > Do you have a pointer to that?
>
> Yes: Gen3 docs Rev.2.30 from Aug 2021, Section 59.2.1, Bits 22-20:
>
> "[R-Car H3, R-Car H3-N]
> 010: 2-clock-cycle delay
> Other than above: Setting prohibited"

Oops, I looked at the 59.2.4, which describes the receive equivalent,
and which does not have this limitations.

> > We already have "renesas,dtdl" to configure this from DT.
> > Iff this is really needed, perhaps it should be added to r8a77951.dtsi?
>
> I have to disagree here. The docs say that other values are prohibited.
> IMO the driver should take care of valid values then. We should not rely
> on user provided input.

Then we should make sure the user cannot override to an invalid value
through "renesas,dtdl" either?

> > I suspect this is a leftover in the BSP from attempts to get MSIOF
> > working on R-Car H3 ES1.0 (which it never did for me, as CLK starts
> > and stops too soon, compared to MOSI/MISO).
> > On R-Car H3 ES2.0, everything works fine, without touching DTDL.
>
> The BSP originally has this patch for ES3 only. I extended to ES2 as
> well because that is what the docs say.

This limitation appeared in Hardware User's Manual rev. 1.50, which
is also the first version to document R-Car H3-N.
So I wouldn't be surprised if this applies to R-Car H3(-N) ES3.0 only.
Remember that rev.0.53 was the switching point from R-Car H3
ES1.0 to ES2.0.
To be clarified with Renesas?

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds