Re: [PATCH 2/6] ARM: dts: r7s72100: Add SPIBSC clocks

[Geert Uytterhoeven <geert@xxxxxxxxxxxxxx>]

Hi Lee,

On Wed, Dec 4, 2019 at 9:38 AM Lee Jones <lee.jones@xxxxxxxxxx> wrote:
> On Tue, 03 Dec 2019, Geert Uytterhoeven wrote:
> > On Tue, Dec 3, 2019 at 7:58 PM Chris Brandt <Chris.Brandt@xxxxxxxxxxx> wrote:
> > > On Tue, Dec 3, 2019, Geert Uytterhoeven wrote:
> > > > > +                       reg = <0x3fefa000 0x100>, <0x18000000
> > > > > + 0x4000000>;
> > > >
> > > > The second region conflicts with the XIP flash@18000000 in
> > > > arch/arm/boot/dts/r7s72100-gr-peach.dts.
> > > > Yes, I know it is the same device ;-)
> > >
> > > Is that just an FYI?? Or do you have some suggestion??
> >
> > Can the flash subnode be compatible with "mtd-rom", even if the parent node
> > is kept disabled?
> >
> > > > > +                       clock-critical = <4>; /* spibsc0 */
> > > >
> > > > Iff we go this clock-critical route, I think this should be specified in the
> > > > board-specific .dts instead of in the SoC-specific .dtsi.
> > >
> > > OK, that's fine. It makes more sense to be in the .dts because it's a board
> > > design decision. I will remove it from the patch.
> > >
> > > The one 'tricky' thing is that the <4> is the index into the number of clocks.
> > >
> > > So in the Renesas BSP where it includes the VDC5 LCD controllers,
> > >
> > > clock-output-names = "i2c0", "i2c1", "i2c2", "i2c3", "vdc50", "vdc51", "spibsc0", "spibsc1";
> > >
> > > the <4> needs to become a <6>.
> >
> > Unless you pass "clkidx" instead of "i" to of_clk_detect_critical() in [1],
> > and use "clock-critical = <R7S72100_CLK_SPIBSC0>" in DT?
> >
> > Unfortunately the exact semantics of clock-critical were never documented.
> > Lee?
>
> /**
>  * of_clk_detect_critical() - set CLK_IS_CRITICAL flag from Device Tree
>  * @np: Device node pointer associated with clock provider
>  * @index: clock index
>  * @flags: pointer to top-level framework flags
>  *
>  * Detects if the clock-critical property exists and, if so, sets the
>  * corresponding CLK_IS_CRITICAL flag.
>  *
>  * Do not use this function. It exists only for legacy Device Tree
>  * bindings, such as the one-clock-per-node style that are outdated.
>  * Those bindings typically put all clock data into .dts and the Linux
>  * driver has no clock data, thus making it impossible to set this flag
>  * correctly from the driver. Only those drivers may call
>  * of_clk_detect_critical from their setup functions.
>  *
>  * Return: error code or zero on success
>  */
>
> If this meets the criteria, the API is pretty simple/self
> explanatory.  What are you having trouble with?

What exactly is the "index" parameter?  This value is matched against
the values of the "clock-critical" (array) property, but it is described
nowhere in the DT bindings.
stih407-clock.dtsi seems to assume this value is an index into the
clock-output-names array?
Can we use it as a value of "clock-indices" instead?

> > Thanks!
> >
> > [1] "[PATCH 1/6] clk: renesas: mstp: Add critical clock from device
> > tree support"
> >     https://lore.kernel.org/linux-renesas-soc/20191203034519.5640-2-chris.brandt@xxxxxxxxxxx/

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