Re: [PATCH 1/4] dt-bindings: thermal: mediatek: Add AP domain to LVTS thermal controllers for mt8195

From: Rob Herring <robh@kernel.org>
To: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Chen-Yu Tsai <wenst@chromium.org>,
	bchihi@baylibre.com, angelogioacchino.delregno@collabora.com,
	rafael@kernel.org, amitk@kernel.org, rui.zhang@intel.com,
	matthias.bgg@gmail.com, krzysztof.kozlowski+dt@linaro.org,
	rdunlap@infradead.org, ye.xingchen@zte.com.cn,
	p.zabel@pengutronix.de, linux-pm@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org,
	linux-mediatek@lists.infradead.org, devicetree@vger.kernel.org,
	khilman@baylibre.com, james.lo@mediatek.com,
	rex-bc.chen@mediatek.com
Subject: Re: [PATCH 1/4] dt-bindings: thermal: mediatek: Add AP domain to LVTS thermal controllers for mt8195
Date: Thu, 16 Mar 2023 17:35:43 -0500	[thread overview]
Message-ID: <20230316223543.GA4008428-robh@kernel.org> (raw)
In-Reply-To: <e5959cb5-af8c-9410-9530-b3e19e9b647a@linaro.org>

On Thu, Mar 09, 2023 at 11:39:13AM +0100, Daniel Lezcano wrote:
> On 09/03/2023 05:40, Chen-Yu Tsai wrote:
> > On Wed, Mar 8, 2023 at 12:46 AM <bchihi@baylibre.com> wrote:
> > > 
> > > From: Balsam CHIHI <bchihi@baylibre.com>
> > > 
> > > Add AP Domain to LVTS thermal controllers dt-binding definition for mt8195.
> > > 
> > > Signed-off-by: Balsam CHIHI <bchihi@baylibre.com>
> > > ---
> > >   include/dt-bindings/thermal/mediatek,lvts-thermal.h | 10 ++++++++++
> > >   1 file changed, 10 insertions(+)
> > > 
> > > diff --git a/include/dt-bindings/thermal/mediatek,lvts-thermal.h b/include/dt-bindings/thermal/mediatek,lvts-thermal.h
> > > index c09398920468..8fa5a46675c4 100644
> > > --- a/include/dt-bindings/thermal/mediatek,lvts-thermal.h
> > > +++ b/include/dt-bindings/thermal/mediatek,lvts-thermal.h
> > > @@ -16,4 +16,14 @@
> > >   #define MT8195_MCU_LITTLE_CPU2  6
> > >   #define MT8195_MCU_LITTLE_CPU3  7
> > > 
> > > +#define MT8195_AP_VPU0  8
> > 
> > Can't this start from 0? This is a different hardware block. The index
> > namespace is separate. Same question for MT8192.
> 
> The ID is used to differentiate the thermal zone identifier in the device
> tree from the driver.
> 
> +		vpu0-thermal {
> +			polling-delay = <0>;
> +			polling-delay-passive = <0>;
> +			thermal-sensors = <&lvts_ap MT8195_AP_VPU0>;
> +
> +			trips {
> +				vpu0_crit: trip-crit {
> +					temperature = <100000>;
> +					hysteresis = <2000>;
> +					type = "critical";
> +				};
> +			};
> +		};
> 
> If MT8195_AP_VPU0 is 0, then the code won't be able to differentiate
> MT8195_AP_VPU0 and MT8195_MCU_BIG_CPU0
> 
> The LVTS driver will call devm_thermal_of_zone_register() with the sensor
> id. If MT8195_MCU_BIG_CPU0 and MT8195_AP_VPU0 have the same id, then at the
> moment of registering the MT8195_AP_VPU0, the underlying OF thermal
> framework code will use MT8195_MCU_BIG_CPU0 description instead because it
> will be the first to be find in the DT.
> 
> If MT8195_AP_VPU0 is described in DT before, then the same will happen when
> registering MT8195_MCU_BIG_CPU0, MT8195_AP_VPU0 will be registered instead.
> 
> IOW all ids must be different.

That's broken for how producer/consumer phandle+args bindings work.

Rob