From mboxrd@z Thu Jan 1 00:00:00 1970 From: Wei Ni Subject: Re: [PATCH v2 4/9] of: Add bindings of hw throttle for soctherm Date: Thu, 28 Apr 2016 14:48:41 +0800 Message-ID: <5721B249.3010101@nvidia.com> References: <1461727554-15065-1-git-send-email-wni@nvidia.com> <1461727554-15065-5-git-send-email-wni@nvidia.com> <20160427233054.GA26451@localhost.localdomain> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: In-Reply-To: <20160427233054.GA26451@localhost.localdomain> Sender: linux-pm-owner@vger.kernel.org To: Eduardo Valentin Cc: thierry.reding@gmail.com, robh+dt@kernel.org, rui.zhang@intel.com, MLongnecker@nvidia.com, swarren@wwwdotorg.org, mikko.perttunen@kapsi.fi, linux-tegra@vger.kernel.org, linux-pm@vger.kernel.org, devicetree@vger.kernel.org, linux-kernel@vger.kernel.org List-Id: linux-tegra@vger.kernel.org On 2016=E5=B9=B404=E6=9C=8828=E6=97=A5 07:30, Eduardo Valentin wrote: > From: Eduardo Valentin > To: Wei Ni > Cc: thierry.reding@gmail.com, robh+dt@kernel.org, rui.zhang@intel.com= , > MLongnecker@nvidia.com, swarren@wwwdotorg.org, > mikko.perttunen@kapsi.fi, linux-tegra@vger.kernel.org, > linux-pm@vger.kernel.org, devicetree@vger.kernel.org, > linux-kernel@vger.kernel.org > Bcc:=20 > Subject: Re: [PATCH v2 4/9] of: Add bindings of hw throttle for socth= erm > Reply-To:=20 > In-Reply-To: <1461727554-15065-5-git-send-email-wni@nvidia.com> >=20 > The patch title must say something about the fact that this is specif= ic > to nvidia thermal driver. Yes, it's my mistake, will fix it in next series. >=20 > On Wed, Apr 27, 2016 at 11:25:49AM +0800, Wei Ni wrote: >> Add HW throttle configuration sub-node for soctherm, which >> is used to describe the throttle event, and worked as a >> cooling device. The "hot" type trip in thermal zone can >> be bound to this cooling device, and trigger the throttle >> function. >> >> Signed-off-by: Wei Ni >> --- >> .../bindings/thermal/nvidia,tegra124-soctherm.txt | 89 +++++++++++= ++++++++++- >> 1 file changed, 87 insertions(+), 2 deletions(-) >> >> diff --git a/Documentation/devicetree/bindings/thermal/nvidia,tegra1= 24-soctherm.txt b/Documentation/devicetree/bindings/thermal/nvidia,tegr= a124-soctherm.txt >> index edebfa0a985e..dc337d139f49 100644 >> --- a/Documentation/devicetree/bindings/thermal/nvidia,tegra124-soct= herm.txt >> +++ b/Documentation/devicetree/bindings/thermal/nvidia,tegra124-soct= herm.txt >> @@ -10,8 +10,14 @@ Required properties : >> - compatible : For Tegra124, must contain "nvidia,tegra124-soctherm= ". >> For Tegra132, must contain "nvidia,tegra132-soctherm". >> For Tegra210, must contain "nvidia,tegra210-soctherm". >> -- reg : Should contain 1 entry: >> +- reg : Should contain at least 2 entries for each entry in reg-nam= es: >> - SOCTHERM register set >> + - Tegra CAR register set: Required for Tegra124 and Tegra210. >> + - CCROC register set: Required for Tegra132. >> +- reg-names : Should contain at least 2 entries: >> + - soctherm-reg >> + - car-reg >> + - ccroc-reg >> - interrupts : Defines the interrupt used by SOCTHERM >> - clocks : Must contain an entry for each entry in clock-names. >> See ../clocks/clock-bindings.txt for details. >> @@ -25,17 +31,44 @@ Required properties : >> - #thermal-sensor-cells : Should be 1. See ./thermal.txt for a desc= ription >> of this property. See = for a >> list of valid values when referring to thermal sensors. >> +- throttle-cfgs: A sub-node which is a container of configuration f= or each >> + hardware throttle events. These events can be set as cooling de= vices. >> + * throttle events: Sub-nodes must be named as "light" or "heavy". >> + Properties: >> + - priority: Each throttles has its own throttle settings, so = the SW need >> + to set priorities for various throttle, the HW arbiter can = select the >> + final throttle settings. >> + Bigger value indicates higher priority, In general, higher = priority >> + translates to lower target frequency. SW needs to ensure th= at critical >> + thermal alarms are given higher priority, and ensure that t= here is >> + no race if priority of two vectors is set to the same value= =2E >> + - cpu-throt-depth: This property is for Tegra124 and Tegra21= 0. It is >> + the throttling depth of pulse skippers, it's the percentage >> + throttling. >> + - cpu-throt-level: This property is only for Tegra132, it is = the level >> + of pulse skippers, which used to throttle clock frequencies= =2E It >> + indicates cpu clock throttling depth, and the depth can be = programmed. >> + Must set as following values: >> + TEGRA_SOCTHERM_THROT_LEVEL_LOW, TEGRA_SOCTHERM_THROT_LEVEL_= MED >> + TEGRA_SOCTHERM_THROT_LEVEL_HIGH, TEGRA_SOCTHERM_THROT_LEVEL= _NONE >=20 > These properties are not generic properties. My understanding is that > you must have vendor prefix in such case. Same applies to the new nod= es. Ok, will do it. >=20 >> + - #cooling-cells: Should be 1. This cooling device only suppo= rt on/off state. >> + See ./thermal.txt for a description of this property. >> =20 >> Note: >> - the "critical" type trip points will be set to SOC_THERM hardware= as the >> shut down temperature. Once the temperature of this thermal zone is= higher >> than it, the system will be shutdown or reset by hardware. >> +- the "hot" type trip points will be set to SOC_THERM hardware as t= he throttle >> +temperature. Once the the temperature of this thermal zone is highe= r >> +than it, it will trigger the HW throttle event. >> =20 >> Example : >> =20 >> soctherm@700e2000 { >> compatible =3D "nvidia,tegra124-soctherm"; >> - reg =3D <0x0 0x700e2000 0x0 0x1000>; >> + reg =3D <0x0 0x700e2000 0x0 0x600 /* SOC_THERM reg_base */ >> + 0x0 0x60006000 0x0 0x400 /* CAR reg_base */ >> + reg-names =3D "soctherm-reg", "car-reg"; >> interrupts =3D ; >> clocks =3D <&tegra_car TEGRA124_CLK_TSENSOR>, >> <&tegra_car TEGRA124_CLK_SOC_THERM>; >> @@ -44,6 +77,45 @@ Example : >> reset-names =3D "soctherm"; >> =20 >> #thermal-sensor-cells =3D <1>; >> + >> + throttle-cfgs { >> + throttle_heavy: heavy { >> + priority =3D <100>; >> + cpu-throt-depth =3D <85>; >> + >> + #cooling-cells =3D <1>; >> + }; >> + throttle_light: light { >> + priority =3D <80>; >> + cpu-throt-depth =3D <50>; >> + >> + #cooling-cells =3D <1>; >> + }; >> + }; >=20 > This is a sensor, which at the same time, has sub nodes that can be > cooling devices. Is my understanding correct of what you are trying t= o > do? Yes, we set this sub nodes as cooling devices. Let me explain it. The Tegra's soctherm has the HW throttle function, it will be triggered= when the temperatures reach to the "hot" type trip. These throttle events can be configured to different levels "heavy" and "light" on different sensors= =2E So it something like cooling devices which can be bound with thermal zones, a= nd in this way, we can use the linux of-thermal framework to handle it. If no= t, we have to add more codes to parse these configuration and set them for th= ose sensors. I had used this method to handle the shutdown function in Tegr= a's soctherm driver in my previous series, however Rob didn't like it, he p= referred to use linux framework to handle it. So I think it's better to set these throttle-cfgs as cooling devices, s= o they can be bound to the cpu/gpu thermal zones easily. >=20 >=20 >> + }; >> + >> +Example: referring to Tegra132's "reg", "reg-names" and "throttle-c= fgs" : >> + >> + soctherm@0,700e2000 { >> + compatible =3D "nvidia,tegra132-soctherm"; >> + reg =3D <0x0 0x700e2000 0x0 0x600 /* SOC_THERM reg_base */ >> + 0x0 0x70040000 0x0 0x200>; /* CCROC reg_base */; >> + reg-names =3D "soctherm-reg", "ccroc-reg"; >> + >> + throttle-cfgs { >> + throttle_heavy: heavy { >> + priority =3D <100>; >> + cpu-throt-level =3D ; >> + >> + #cooling-cells =3D <1>; >> + }; >> + throttle_light: light { >> + priority =3D <80>; >> + cpu-throt-level =3D ; >> + >> + #cooling-cells =3D <1>; >=20 > Could you please describe a little here what to expect of the above > setup? Ok, let me describe it. When the "heavy" cooling device triggered, the hardware will skip cpu c= lock's pulse in HIGH level, and if the "light" triggered, the hardware will sk= ip cpu clock's pulse in MED level. If these two devices are triggered in same time, the HW throttle arbite= r will select the highest priority as the final throttle settings to skip cpu = pulse. >=20 >=20 >> + }; >> + }; >> }; >> =20 >> Example: referring to thermal sensors : >> @@ -62,6 +134,19 @@ Example: referring to thermal sensors : >> hysteresis =3D <1000>; >> type =3D "critical"; >> }; >> + >> + cpu_throttle_trip: throttle-trip { >> + temperature =3D <100000>; >> + hysteresis =3D <1000>; >> + type =3D "hot"; >> + }; >> + }; >> + >> + cooling-maps { >> + map0 { >> + trip =3D <&cpu_throttle_trip>; >> + cooling-device =3D <&throttle_heavy 1 1>; >> + }; >> }; >> }; >> }; >> --=20 >> 1.9.1 >> > -- > To unsubscribe from this list: send the line "unsubscribe linux-pm" i= n > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html >=20