Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
Message-ID: <trinity-59bd4461-3429-4382-9754-90cf20287cad-1624624111998@3c-app-gmx-bs66>
From:   Frank Wunderlich <frank-w@public-files.de>
To:     Daniel Lezcano <daniel.lezcano@linaro.org>
Cc:     Eric Woudstra <ericwouds@gmail.com>,
        Rob Herring <robh+dt@kernel.org>,
        Matthias Brugger <matthias.bgg@gmail.com>,
        devicetree@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
        linux-mediatek@lists.infradead.org, linux-kernel@vger.kernel.org,
        Sean Wang <sean.wang@mediatek.com>,
        Ryder Lee <ryder.lee@mediatek.com>,
        Daniel Golle <daniel@makrotopia.org>
Subject: Aw: Re:  Re: Re: [PATCH] Fix mt7622.dtsi thermal cpu
Content-Type: text/plain; charset=UTF-8
Date:   Fri, 25 Jun 2021 14:28:32 +0200
Importance: normal
Sensitivity: Normal
In-Reply-To: <915b101a-6bea-ae96-78ed-d27fa5bfce3a@linaro.org>
References: <20210619121927.32699-1-ericwouds@gmail.com>
 <e30a2d01-a200-80cb-88d9-6aea62dd49f1@linaro.org>
 <56fb5540-fb86-4e6a-a596-1276026b37e5@gmail.com>
 <a4e41929-6ab4-fabb-741e-f25a5fd14e3b@linaro.org>
 <47261865-00e3-41eb-bb36-2b939f81f1e8@gmail.com>
 <fb633034-96e5-6165-b43f-290ae1a65cfd@linaro.org>
 <189b52d5-b103-43e1-a64f-1e627fbc75af@gmail.com>
 <173e6bab-9d21-eb28-9b91-a5f80c01fd03@linaro.org>
 <3dd22cf2-1186-4870-aa49-e5cddc18c6e9@gmail.com>
 <trinity-7580d955-3187-41e5-9297-1ac8f628a9d5-1624609003739@3c-app-gmx-bs66>
 <8b27246b-721e-fa0e-5c2b-b1b4b4d6fdd3@linaro.org>
 <trinity-2eb7c0ac-d9dc-446c-8907-69b5f4df6838-1624618996538@3c-app-gmx-bs66>
 <915b101a-6bea-ae96-78ed-d27fa5bfce3a@linaro.org>
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

> Gesendet: Freitag, 25=2E Juni 2021 um 13:47 Uhr
> Von: "Daniel Lezcano" <daniel=2Elezcano@linaro=2Eorg>

> > but we need to disable the passive trip as cpu-trotteling starts there=
=2E=2E=2Ethe higher temperature trips are currently not reached
>=20
> Sorry, can you rephrase it ? I'm not getting the point=2E

the problem currently is that passive is at 47degress Celsius and trottles=
 cpu, active (67=C2=B0C) and hot points are never reached this way=2E so at=
 least we need to change temperatures in dtsi, and maybe disable cpu-trotte=
ling on passive trip=2E imho fan will never start if it is in active and cp=
u is trottled before in passive

> > summary
> >=20
> > moving fan and cpu_thermal-override to bananapi-r64=2Edts
> >=20
> > passive-trip: cooling-device =3D <&cpu0/1 0 0> as in erics Patch
> > active trip: cooling-device =3D <&fan0 THERMAL_NO_LIMIT THERMAL_NO_LIM=
IT>;
> > the other 2 unchanged
> >=20
> > but i suggest changing the temperature points in mt7622 dtsi as this i=
s SoC specific
> >=20
> > so basicly:
> >=20
> > --- a/arch/arm64/boot/dts/mediatek/mt7622=2Edtsi
> > +++ b/arch/arm64/boot/dts/mediatek/mt7622=2Edtsi
> > @@ -143,13 +143,13 @@ cpu_thermal: cpu-thermal {
> >=20
> >                         trips {
> >                                 cpu_passive: cpu-passive {
> > -                                       temperature =3D <47000>;
> > +                                       temperature =3D <70000>;
>=20
> May be increase the passive temp to 75=C2=B0C=2E
>=20
> >                                         hysteresis =3D <2000>;
> >                                         type =3D "passive";
> >                                 };
> >=20
> >                                 cpu_active: cpu-active {
> > -                                       temperature =3D <67000>;
> > +                                       temperature =3D <80000>;
> >                                         hysteresis =3D <2000>;
> >                                         type =3D "active";
> >                                 };
>=20
> Move the active trip 'cpu_active' to mt7622-bananapi-bpi-r64=2Edts=2E an=
d
> set it to 70=C2=B0C in the mt7622-bananapi-bpi-r64=2Edts, so the fan wil=
l act
> before the cpu throttling=2E
>=20
> The behavior should be the following: The temperature reaches 70=C2=B0C,=
 the
> fan will start, if the temperature continues to increase, it will
> increase the speed=2E If the temperature reaches 75=C2=B0C, the fan is s=
till
> rotating at full speed but the cpu begins to be throttled=2E

passive to 75 and active lower to 70? is this as intended that active come=
s before passive?

mt7622-bananapi-bpi-r64=2Edts:

&cpu_thermal {
	trips {
		cpu_passive: cpu-passive {
			temperature =3D <75000>;
			hysteresis =3D <2000>;
			type =3D "passive";
		};

		cpu_active: cpu-active {
			temperature =3D <70000>;
			hysteresis =3D <2000>;
			type =3D "active";
		};
	};

	cooling-maps {
		map1 {
			trip =3D <&cpu_active>;
			cooling-device =3D <&fan0 THERMAL_NO_LIMIT THERMAL_NO_LIMIT>;
		};
	};
};


> AFAIU, it is a Cortex-A53 running @1=2E35GH, so except the board is in a
> black metal box under the sun, I don't see how we can reach this thermal
> limits=2E
>=20
> > @@ -170,8 +170,8 @@ cpu-crit {
> >                         cooling-maps {
> >                                 map0 {
> >                                         trip =3D <&cpu_passive>;
> > -                                       cooling-device =3D <&cpu0 THER=
MAL_NO_LIMIT THERMAL_NO_LIMIT>,
> > -                                                        <&cpu1 THERMA=
L_NO_LIMIT THERMAL_NO_LIMIT>;
> > +                                       cooling-device =3D <&cpu0 0 0>=
,
> > +                                                        <&cpu1 0 0>;
>=20
> You should keep it untouched=2E

then cpu is trottled at passive point (currently 47=C2=B0C) and imho fan d=
oes not start at active

> >                                 };