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;
Subject: Re: [PATCH v3 1/3] ARM: tegra: Add device-tree for Ouya
To:     Peter Geis <pgwipeout@gmail.com>, Rob Herring <robh+dt@kernel.org>,
        Thierry Reding <thierry.reding@gmail.com>,
        Jonathan Hunter <jonathanh@nvidia.com>,
        Matt Merhar <mattmerhar@protonmail.com>,
        Stephen Warren <swarren@wwwdotorg.org>,
        Bob Ham <rah@settrans.net>,
        Leonardo Bras <leobras.c@gmail.com>,
        Michael Brougham <jusplainmike@gmail.com>
Cc:     linux-tegra@vger.kernel.org, devicetree@vger.kernel.org,
        linux-kernel@vger.kernel.org
References: <20201004133114.845230-1-pgwipeout@gmail.com>
 <20201004133114.845230-2-pgwipeout@gmail.com>
From:   Dmitry Osipenko <digetx@gmail.com>
Message-ID: <b3b99578-73c8-17e1-81ff-4b3fdf1b58f3@gmail.com>
Date:   Mon, 5 Oct 2020 17:02:00 +0300
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.10.0
MIME-Version: 1.0
In-Reply-To: <20201004133114.845230-2-pgwipeout@gmail.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 8bit
Precedence: bulk

04.10.2020 16:31, Peter Geis пишет:
> +	thermal-zones {
> +		cpu_thermal: cpu-thermal {
> +			polling-delay = <5000>;
> +			polling-delay-passive = <5000>;
> +
> +			thermal-sensors = <&cpu_temp 1>;
> +
> +			trips {
> +				cpu_alert0: cpu-alert0 {
> +					temperature = <50000>;
> +					hysteresis = <10000>;

Hello, Peter!

A day ago I was tuning thermal zones for Nexus 7 because found that the
current variant is a bit too unpractical. In particular temperature
hysteresis should be small, otherwise cpufreq could get throttled
enormously to the point that device becomes unusable. This is an
active-cooling zone, but it looks to me that hysteresis is a bit too
high and should make Ouya much noisier than it could be.

I suggest to try to set hysteresis to 0.2C here, i.e. hysteresis = <200>.

I also suggest to bump the temperature threshold to 55C in order to
ignore temporal temperature spikes because CPU temp should be about 40C
during idle and then it may raise quickly for a brief moments during
average usage.

> +					type = "active";
> +				};
> +				cpu_alert1: cpu-alert1 {
> +					temperature = <70000>;
> +					hysteresis = <5000>;
> +					type = "passive";
> +				};

And here to 0.2C as well.

> +				cpu_crit: cpu-crit {
> +					temperature = <90000>;
> +					hysteresis = <2000>;
> +					type = "critical";
> +				};

The critical zone perhaps should be fine as-is.