Message-ID: <YSUNSmgeGgc+x5xA@ripper>
Date:   Tue, 24 Aug 2021 08:16:26 -0700
From:   Bjorn Andersson <bjorn.andersson@...aro.org>
To:     AngeloGioacchino Del Regno 
        <angelogioacchino.delregno@...ainline.org>
Cc:     Thara Gopinath <thara.gopinath@...aro.org>,
        Konrad Dybcio <konrad.dybcio@...ainline.org>,
        ~postmarketos/upstreaming@...ts.sr.ht, martin.botka@...ainline.org,
        marijn.suijten@...ainline.org, jamipkettunen@...ainline.org,
        Andy Gross <agross@...nel.org>,
        Amit Kucheria <amitk@...nel.org>,
        Zhang Rui <rui.zhang@...el.com>,
        Daniel Lezcano <daniel.lezcano@...aro.org>,
        Rob Herring <robh+dt@...nel.org>,
        linux-arm-msm@...r.kernel.org, linux-pm@...r.kernel.org,
        devicetree@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 14/39] arm64: dts: qcom: sdm630: Add TSENS node

On Thu 29 Jul 06:48 PDT 2021, AngeloGioacchino Del Regno wrote:

> Il 29/07/21 13:14, Thara Gopinath ha scritto:
> > 
> > 
> > On 7/29/21 6:55 AM, Konrad Dybcio wrote:
> > > 
> > > On 29.07.2021 12:54, Thara Gopinath wrote:
> > > > 
> > > > 
> > > > On 7/29/21 6:52 AM, Konrad Dybcio wrote:
> > > > > 
> > > > > On 29.07.2021 12:50, Thara Gopinath wrote:
> > > > > > Hi Konrad,
> > > > > > 
> > > > > > On 7/28/21 6:25 PM, Konrad Dybcio wrote:
> > > > > > > This will enable temperature reporting for various SoC
> > > > > > > components.
> > > > > > > 
> > > > > > > Signed-off-by: AngeloGioacchino Del Regno
> > > > > > > <angelogioacchino.delregno@...ainline.org>
> > > > > > > Signed-off-by: Konrad Dybcio <konrad.dybcio@...ainline.org>
> > > > > > > ---
> > > > > > >     .../devicetree/bindings/thermal/qcom-tsens.yaml       |  1 +
> > > > > > >     arch/arm64/boot/dts/qcom/sdm630.dtsi                  | 11 +++++++++++
> > > > > > >     2 files changed, 12 insertions(+)
> > > > > > > 
> > > > > > > diff --git
> > > > > > > a/Documentation/devicetree/bindings/thermal/qcom-tsens.yaml
> > > > > > > b/Documentation/devicetree/bindings/thermal/qcom-tsens.yaml
> > > > > > > index 4a2eaf28e3fd..d3b9e9b600a2 100644
> > > > > > > --- a/Documentation/devicetree/bindings/thermal/qcom-tsens.yaml
> > > > > > > +++ b/Documentation/devicetree/bindings/thermal/qcom-tsens.yaml
> > > > > > > @@ -48,6 +48,7 @@ properties:
> > > > > > >                   - qcom,sc7180-tsens
> > > > > > >                   - qcom,sc7280-tsens
> > > > > > >                   - qcom,sc8180x-tsens
> > > > > > > +              - qcom,sdm630-tsens
> > > > > > >                   - qcom,sdm845-tsens
> > > > > > >                   - qcom,sm8150-tsens
> > > > > > >                   - qcom,sm8250-tsens
> > > > > > > diff --git a/arch/arm64/boot/dts/qcom/sdm630.dtsi
> > > > > > > b/arch/arm64/boot/dts/qcom/sdm630.dtsi
> > > > > > > index 1e54828817d5..7e9c80e35fba 100644
> > > > > > > --- a/arch/arm64/boot/dts/qcom/sdm630.dtsi
> > > > > > > +++ b/arch/arm64/boot/dts/qcom/sdm630.dtsi
> > > > > > > @@ -627,6 +627,17 @@ mnoc: interconnect@...5000 {
> > > > > > >                      <&mmcc AHB_CLK_SRC>;
> > > > > > >             };
> > > > > > >     +        tsens: thermal-sensor@...e000 {
> > > > > > > +            compatible = "qcom,sdm630-tsens", "qcom,tsens-v2";
> > > > > > > +            reg = <0x010ae000 0x1000>, /* TM */
> > > > > > > +                  <0x010ad000 0x1000>; /* SROT */
> > > > > > > +            #qcom,sensors = <12>;
> > > > > > 
> > > > > > Are all 12 sensors used ? I see that in a later patch
> > > > > > "arm64: dts: qcom: sdm630: Add thermal-zones
> > > > > > configuration" only 9 are used.
> > > > > 
> > > > > Hi,
> > > > > 
> > > > > if I recall correctly, they all give output but not all of
> > > > > the mappings were documented in the downstream sources and
> > > > > we have no documentation whatsoever :(
> > > > 
> > > > Right. In that case, why not change #qcom,sensors to 9 and add
> > > > rest of the sensors if and when needed ?
> > > > 
> > > I don't think it makes sense to describe the hardware incorrectly,
> > > even if some of it is unused.
> > 
> > My thinking was more along the lines of don't expose unused h/w bits.
> > 
> 
> You're right about not exposing unused HW bits, but even PC x86 motherboards
> (I mean the smbus/i2c drivers for the big holy management/sensors chips) do
> have such a "base" configuration, where some lines are read as 0 because they
> are effectively not connected by hardware.
> 
> In order to avoid confusion to other developers, in my personal opinion, it would
> be good go for the current value of 12 (which isn't incorrect, as that's what the
> SoC supports)... I don't think that anyone would be confused by seeing zero
> readings on some sensors (if their device don't support such sensor), as I think
> that everyone is used to that anyway, even if that's in other circumstances...
> 
> In any case, luckily that's also safe, because there's no firmware that restricts
> the readings to a subset of sensors in this domain (nobody is going to get a
> hypervisor fault for that).
> 
> I would also, in case, propose to see how things go: I would expect other
> developers to push device trees for many SDM630/636/660 devices, including but
> not limited to smartphones and SBCs.. so perhaps if we find out that really
> nobody uses the 12 sensors, or if the very vast majority uses a different amount,
> perhaps we may just transfer the value to device-specific configurations in one
> go, as to avoid unnecessary noise... I think :)))
> 

If the SoC has 12 sensors I think it makes sense to define that, similar
to how a SoC might have 200 GPIOs, even though only a handful is
actually used.

Regards,
Bjorn

Powered by blists - more mailing lists