Re: [PATCH v1 1/5] dt-bindings: cpufreq: Add binding for NVIDIA Tegra20/30

[Rob Herring]

On Tue, Sep 25, 2018 at 12:29 PM Dmitry Osipenko <digetx@xxxxxxxxx> wrote:
>
> On 9/25/18 7:58 PM, Rob Herring wrote:
> > On Thu, Aug 30, 2018 at 10:43:52PM +0300, Dmitry Osipenko wrote:
> >> Add device-tree binding that describes CPU frequency-scaling hardware
> >> found on NVIDIA Tegra20/30 SoC's.
> >>
> >> Signed-off-by: Dmitry Osipenko <digetx@xxxxxxxxx>
> >> ---
> >>  .../cpufreq/nvidia,tegra20-cpufreq.txt        | 38 +++++++++++++++++++
> >>  1 file changed, 38 insertions(+)
> >>  create mode 100644 Documentation/devicetree/bindings/cpufreq/nvidia,tegra20-cpufreq.txt
> >>
> >> diff --git a/Documentation/devicetree/bindings/cpufreq/nvidia,tegra20-cpufreq.txt b/Documentation/devicetree/bindings/cpufreq/nvidia,tegra20-cpufreq.txt
> >> new file mode 100644
> >> index 000000000000..2c51f676e958
> >> --- /dev/null
> >> +++ b/Documentation/devicetree/bindings/cpufreq/nvidia,tegra20-cpufreq.txt
> >> @@ -0,0 +1,38 @@
> >> +Binding for NVIDIA Tegra20 CPUFreq
> >> +==================================
> >> +
> >> +Required properties:
> >> +- clocks: Must contain an entry for each entry in clock-names.
> >> +  See ../clocks/clock-bindings.txt for details.
> >> +- clock-names: Must include the following entries:
> >> +  - pll_x: main-parent for CPU clock, must be the first entry
> >> +  - backup: intermediate-parent for CPU clock
> >> +  - cpu: the CPU clock
> >
> > The Cortex A9 has CLK, PERIPHCLK, and PERIPHCLKEN clocks and only CLK
> > is used for the cpu core. You can't just define your own clocks that
> > you happen to want access to.
> >
> > Otherwise, you're not defining anything new here, so a binding document
> > isn't required.
>
> PERIPHCLK is a different thing.

Right, that's what I meant. Only CLK is used.

> Here we are defining the CPU clock and
> its *parent* sources, the PLLX (main) and backup (intermediate clock
> that is used while PLLX changes its rate). These are not some random
> clocks "that you happen to want access to", they are essential for the
> Tegra CPUFreq driver, CPU is running off them.

assigned-clocks is generally how we get parent clocks in this
situation. "clocks" is for what physical clocks are attached to a
given block. ARM very clearly documents the clocks for their IP
blocks.

> I assume that PERIPHCLK and other clocks are derived from the "CPU"
> clock and their configuration is hardwired. Probably Peter knows how
> it's implemented in HW.

Yes, because PERIPHCLK must be synchronous. In any case, it is
irrelevant to cpu nodes and applies to timer, SCU, and GIC nodes.

> I'm now working on v2 that will include more Tegra-specific stuff in the
> binding, like custom "opp-supported-hw" property and probably some more.