Re: [PATCH v6 11/21] dt-bindings: clock: Add StarFive JH7110 system clock and reset generator

[Hal Feng]

On Fri, 24 Mar 2023 10:39:36 +0100, Emil Renner Berthing wrote:
> On Thu, 23 Mar 2023 at 10:02, Conor Dooley <conor.dooley@xxxxxxxxxxxxx> wrote:
>>
>> Hal, Emil,
>>
>> On Thu, Mar 23, 2023 at 03:44:41PM +0800, Hal Feng wrote:
>> > On Wed, 22 Mar 2023 21:53:37 +0000, Conor Dooley wrote:
>> > > On Mon, Mar 20, 2023 at 06:37:40PM +0800, Hal Feng wrote:
>> > >> From: Emil Renner Berthing <kernel@xxxxxxxx>
>> > >>
>> > >> Add bindings for the system clock and reset generator (SYSCRG) on the
>> > >> JH7110 RISC-V SoC by StarFive Ltd.
>> > >>
>> > >> Reviewed-by: Conor Dooley <conor.dooley@xxxxxxxxxxxxx>
>> > >> Reviewed-by: Rob Herring <robh@xxxxxxxxxx>
>> > >> Signed-off-by: Emil Renner Berthing <kernel@xxxxxxxx>
>> > >> Signed-off-by: Hal Feng <hal.feng@xxxxxxxxxxxxxxxx>
>> > >> ---
>> > >>  .../clock/starfive,jh7110-syscrg.yaml         | 104 +++++++++
>> > >>  MAINTAINERS                                   |   8 +-
>> > >>  .../dt-bindings/clock/starfive,jh7110-crg.h   | 203 ++++++++++++++++++
>> > >>  .../dt-bindings/reset/starfive,jh7110-crg.h   | 142 ++++++++++++
>> > >>  4 files changed, 454 insertions(+), 3 deletions(-)
>> > >>  create mode 100644 Documentation/devicetree/bindings/clock/starfive,jh7110-syscrg.yaml
>> > >>  create mode 100644 include/dt-bindings/clock/starfive,jh7110-crg.h
>> > >>  create mode 100644 include/dt-bindings/reset/starfive,jh7110-crg.h
>> > >>
>> > >> diff --git a/Documentation/devicetree/bindings/clock/starfive,jh7110-syscrg.yaml b/Documentation/devicetree/bindings/clock/starfive,jh7110-syscrg.yaml
>> > >> new file mode 100644
>> > >> index 000000000000..84373ae31644
>> > >> --- /dev/null
>> > >> +++ b/Documentation/devicetree/bindings/clock/starfive,jh7110-syscrg.yaml
>> > >> @@ -0,0 +1,104 @@
>> > >> +# SPDX-License-Identifier: GPL-2.0-only OR BSD-2-Clause
>> > >> +%YAML 1.2
>> > >> +---
>> > >> +$id: http://devicetree.org/schemas/clock/starfive,jh7110-syscrg.yaml#
>> > >> +$schema: http://devicetree.org/meta-schemas/core.yaml#
>> > >> +
>> > >> +title: StarFive JH7110 System Clock and Reset Generator
>> > >> +
>> > >> +maintainers:
>> > >> +  - Emil Renner Berthing <kernel@xxxxxxxx>
>> > >> +
>> > >> +properties:
>> > >> +  compatible:
>> > >> +    const: starfive,jh7110-syscrg
>> > >> +
>> > >> +  reg:
>> > >> +    maxItems: 1
>> > >> +
>> > >> +  clocks:
>> > >> +    oneOf:
>> > >> +      - items:
>> > >> +          - description: Main Oscillator (24 MHz)
>> > >> +          - description: GMAC1 RMII reference or GMAC1 RGMII RX
>> > >> +          - description: External I2S TX bit clock
>> > >> +          - description: External I2S TX left/right channel clock
>> > >> +          - description: External I2S RX bit clock
>> > >> +          - description: External I2S RX left/right channel clock
>> > >> +          - description: External TDM clock
>> > >> +          - description: External audio master clock
>> > >> +
>> > >> +      - items:
>> > >> +          - description: Main Oscillator (24 MHz)
>> > >> +          - description: GMAC1 RMII reference
>> > >> +          - description: GMAC1 RGMII RX
>> > >> +          - description: External I2S TX bit clock
>> > >> +          - description: External I2S TX left/right channel clock
>> > >> +          - description: External I2S RX bit clock
>> > >> +          - description: External I2S RX left/right channel clock
>> > >> +          - description: External TDM clock
>> > >> +          - description: External audio master clock
>> > >> +
>> > >> +  clock-names:
>> > >> +    oneOf:
>> > >> +      - items:
>> > >> +          - const: osc
>> > >> +          - enum:
>> > >> +              - gmac1_rmii_refin
>> > >> +              - gmac1_rgmii_rxin
>> > >> +          - const: i2stx_bclk_ext
>> > >> +          - const: i2stx_lrck_ext
>> > >> +          - const: i2srx_bclk_ext
>> > >> +          - const: i2srx_lrck_ext
>> > >> +          - const: tdm_ext
>> > >> +          - const: mclk_ext
>> > >> +
>> > >> +      - items:
>> > >> +          - const: osc
>> > >> +          - const: gmac1_rmii_refin
>> > >> +          - const: gmac1_rgmii_rxin
>> > >> +          - const: i2stx_bclk_ext
>> > >> +          - const: i2stx_lrck_ext
>> > >> +          - const: i2srx_bclk_ext
>> > >> +          - const: i2srx_lrck_ext
>> > >> +          - const: tdm_ext
>> > >> +          - const: mclk_ext
>> > >
>> > > I'm sorry to be a bit of a bore about these bindings, but Emil mentioned
>> > > to me today that he had some doubts about whether any of these audio
>> > > clocks are actually required.
>> > > I've had a bit of a look at the driver, cos the TRM that I have doesn't
>> > > describe the clock tree (from what recall at least) and I think he is
>> > > right.
>> > > For example, the TDM clock:
>> > > +   JH71X0_GATE(JH7110_SYSCLK_TDM_AHB, "tdm_ahb", 0, JH7110_SYSCLK_AHB0),
>> > > +   JH71X0_GATE(JH7110_SYSCLK_TDM_APB, "tdm_apb", 0, JH7110_SYSCLK_APB0),
>> > > +   JH71X0_GDIV(JH7110_SYSCLK_TDM_INTERNAL, "tdm_internal", 0, 64, JH7110_SYSCLK_MCLK),
>> > > +   JH71X0__MUX(JH7110_SYSCLK_TDM_TDM, "tdm_tdm", 2,
>> > > +               JH7110_SYSCLK_TDM_INTERNAL,
>> > > +               JH7110_SYSCLK_TDM_EXT),
>> > >
>> > > Hopefully, I'm not making a balls of something here, but it looks like I
>> > > can choose an internal TDM clock, that is based on JH7110_SYSCLK_MCLK,
>> > > which in turn comes from either an internal or external source.
>> > > If I am following correctly, that'd be:
>> > > +   JH71X0__DIV(JH7110_SYSCLK_MCLK_INNER, "mclk_inner", 64, JH7110_SYSCLK_AUDIO_ROOT),
>> > >
>> > > Which in turn comes from:
>> > > +   JH71X0__DIV(JH7110_SYSCLK_AUDIO_ROOT, "audio_root", 8, JH7110_SYSCLK_PLL2_OUT),
>> > >
>> > > This leaves me wondering which clocks are *actually* required for a
>> > > functioning system - is it actually just osc and one of gmac1_rmii_refin
>> > > or gmac1_rgmii_rxin.
>> >
>> > As I had mentioned somewhere before, some audio clocks need to change their
>> > parents at different stages of work. I should explain in detail here.
>> >
>> > For the i2s*_ext clocks, we should use these external clocks as parents when
>> > the I2S module is working in the slave mode, while we should use the internal
>> > clocks as parents when the I2S module is working in the master mode.
> 
> Right, so what Hal is saying here is that the i2s*_ext clocks are only
> needed if the board is designed to have i2s modules in slave mode.

I don't think we want to describe the dt-bindings of clock controllers according
to the board design.
As I said in v5, IMO, the "optional" here means whether the clock is optional
in the sense that we are able to run all designed functions on the SoC.

> 
>> > For the tdm_ext clock, we use it as the clock source for an accurate playback
>> > rate. If we use the internal clock as clock source, the TDM can't work
>> > normally, because it can't get a required rate from the internal divider.
>> > By the way, note that we need to use the internal clock as clock source when
>> > we try to reset the tdm clock, otherwise, the reset will fail.
>> >
>> > For the mclk_ext clock, which is 12.288MHz, it's used as the clock source
>> > through all the running time, otherwise, the daughter clocks can't get the
>> > required rate from the internal PLL2 clock (1188MHz) through dividers.
> 
> Right, so PLL2 is 1188MHz on the VisionFive 2.
> 
> Hal: But is it not possible to program the PLL2 to run at a multiple
> of 12.288MHz in some other configuration?

Yeah, we can definitely set the PLL2 at a multiple of 12.288MHz as what we did in
the v1. We changed it to 1188MHz for supporting dual-screen display.

Best regards,
Hal

> 
>> > So all these audio external clocks (i2s*_ext / tdm_ext / mclk_ext) are
>> > actually required.
>>
>> Okay. I think I am okay with leaving the binding as-is then, and if
>> someone needs to omit the entire audio subsystem on the SoC, they can
>> follow Stephen's suggestion.
>>
>> @Emil, is that okay with you?
> 
> Conor: I'm fine with the bindings like this. I just want to make sure
> that we all have the same idea of what is "optional" and should be
> marked as such in the bindings.
> 
> /Emil