[PATCH 3/4] ARM: dts: qcom: apq8074-dragonboard: enable adsp and MSS

Dmitry Baryshkov dmitry.baryshkov at linaro.org
Mon May 8 23:38:26 UTC 2023


On Mon, 8 May 2023 at 21:01, Dmitry Baryshkov
<dmitry.baryshkov at linaro.org> wrote:
>
> On 08/05/2023 13:49, Dmitry Baryshkov wrote:
> > On 08/05/2023 13:38, Konrad Dybcio wrote:
> >>
> >>
> >> On 8.05.2023 12:33, Dmitry Baryshkov wrote:
> >>> On 08/05/2023 11:33, Konrad Dybcio wrote:
> >>>>
> >>>>
> >>>> On 7.05.2023 21:07, Dmitry Baryshkov wrote:
> >>>>> Enable ADSP and Modem DSPs on APQ8074 dragonboard. The MSS region
> >>>>> differs from the one defined in the msm8974, so it overriden locally.
> >>>>>
> >>>>> The modem is specified use mba.mbn instead of mbn.b00 (for the sake of
> >>>>> similarity with other platforms). This requires a patch for remoteproc
> >>>>> to be applied [1].
> >>>>>
> >>>>> [1]
> >>>>> https://lore.kernel.org/all/20230507172041.2320279-1-dmitry.baryshkov@linaro.org/
> >>>>>
> >>>>> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov at linaro.org>
> >>>>> ---
> >>>>>    .../arm/boot/dts/qcom-apq8074-dragonboard.dts | 28
> >>>>> +++++++++++++++++++
> >>>>>    1 file changed, 28 insertions(+)
> >>>>>
> >>>>> diff --git a/arch/arm/boot/dts/qcom-apq8074-dragonboard.dts
> >>>>> b/arch/arm/boot/dts/qcom-apq8074-dragonboard.dts
> >>>>> index 6b047c679370..c893afc00eb4 100644
> >>>>> --- a/arch/arm/boot/dts/qcom-apq8074-dragonboard.dts
> >>>>> +++ b/arch/arm/boot/dts/qcom-apq8074-dragonboard.dts
> >>>>> @@ -4,6 +4,8 @@
> >>>>>    #include "qcom-pm8841.dtsi"
> >>>>>    #include "qcom-pm8941.dtsi"
> >>>>>    +/delete-node/ &mpss_region;
> >>>>> +
> >>>>>    / {
> >>>>>        model = "Qualcomm APQ8074 Dragonboard";
> >>>>>        compatible = "qcom,apq8074-dragonboard", "qcom,apq8074";
> >>>>> @@ -17,6 +19,13 @@ aliases {
> >>>>>        chosen {
> >>>>>            stdout-path = "serial0:115200n8";
> >>>>>        };
> >>>>> +
> >>>>> +    reserved-memory {
> >>>>> +        mpss_region: mpss at ac00000 {
> >>>>> +            reg = <0x0ac00000 0x2500000>;
> >>>>> +            no-map;
> >>>>> +        };
> >>>>> +    };
> >>>>>    };
> >>>>>      &blsp1_uart2 {
> >>>>> @@ -39,6 +48,25 @@ eeprom: eeprom at 52 {
> >>>>>        };
> >>>>>    };
> >>>>>    +&remoteproc_adsp {
> >>>>> +    cx-supply = <&pm8841_s2>;
> >>>>> +
> >>>>> +    firmware-name = "qcom/apq8074/adsp.mbn";
> >>>>> +
> >>>>> +    status = "okay";
> >>>>> +};
> >>>>> +
> >>>>> +&remoteproc_mss {
> >>>>> +    cx-supply = <&pm8841_s2>;
> >>>>> +    mss-supply = <&pm8841_s3>;
> >>>>> +    mx-supply = <&pm8841_s1>;
> >>>>> +    pll-supply = <&pm8941_l12>;
> >>>> High time to move this to rpmpd!
> >>>> I won't object to adding this though, as it obviously works
> >>>> and is already used on other boards..
> >>>
> >>> I think the problem is that they are not level-voted on this
> >>> platform, so they are regulators, not PDs.
> >> They're corner-voted.
> >
> > Hmm. Indeed. In msm8974-regulators I see both voltage and corner entries
> > for these regulators.
>
> Checked. Only CX and GFX (yes, MX not included) are enabled as corners
> in vendor dtsi. So this probably doesn't gain us a lot.

I did a check. Implementing CX as a powerdomain here makes things
worse for now. We should first teach mss/pas/etc drivers to properly
handle the case when there is a single power-domain, which should be
unbound after staring the DSP Consider e.g. ADSP and CDSP on sm8150 or
CDSP on sm8250. But that's definitely a topic for a different patch
series. After that we can consider landing msm8974_genpd.

>
> >
> >>
> >> Konrad
> >>>
> >>>>
> >>>>> +
> >>>>> +    firmware-name = "qcom/apq8074/mba.mbn", "qcom/apq8074/modem.mbn";
> >>>> Could you please keep it one entry per line?
> >>>
> >>> Sure.
> >>>
> >>>>
> >>>> Otherwise,
> >>>>
> >>>> Reviewed-by: Konrad Dybcio <konrad.dybcio at linaro.org>
> >>>>
> >>>> Konrad
> >>>>> +
> >>>>> +    status = "okay";
> >>>>> +};
> >>>>> +
> >>>>>    &rpm_requests {
> >>>>>        regulators-0 {
> >>>>>            compatible = "qcom,rpm-pm8841-regulators";
> >>>
> >
>
> --
> With best wishes
> Dmitry
>


-- 
With best wishes
Dmitry


More information about the dri-devel mailing list