Re: [PATCH v3 4/5] cpufreq: apple-soc: Add new driver to control Apple SoC CPU P-states

From: Ulf Hansson <ulf.hansson@linaro.org>
To: Hector Martin <marcan@marcan.st>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>,
	Viresh Kumar <viresh.kumar@linaro.org>,
	Matthias Brugger <matthias.bgg@gmail.com>,
	Sven Peter <sven@svenpeter.dev>,
	Alyssa Rosenzweig <alyssa@rosenzweig.io>,
	Rob Herring <robh+dt@kernel.org>,
	Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
	Stephen Boyd <sboyd@kernel.org>, Marc Zyngier <maz@kernel.org>,
	Mark Kettenis <mark.kettenis@xs4all.nl>,
	asahi@lists.linux.dev, linux-arm-kernel@lists.infradead.org,
	linux-pm@vger.kernel.org, devicetree@vger.kernel.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH v3 4/5] cpufreq: apple-soc: Add new driver to control Apple SoC CPU P-states
Date: Mon, 14 Nov 2022 20:49:54 +0100	[thread overview]
Message-ID: <CAPDyKFpnzmRkgRHsRoRRSpzFGPT7oMx0u+d1g5Rmu_mCA_nC2A@mail.gmail.com> (raw)
In-Reply-To: <c791b776-c90f-4ee5-b025-79352937bf62@marcan.st>

On Tue, 1 Nov 2022 at 19:17, Hector Martin <marcan@marcan.st> wrote:
>
> On 02/11/2022 00.16, Ulf Hansson wrote:
> > On Mon, 24 Oct 2022 at 06:40, Hector Martin <marcan@marcan.st> wrote:
> >>
> >> This driver implements CPU frequency scaling for Apple Silicon SoCs,
> >> including M1 (t8103), M1 Max/Pro/Ultra (t600x), and M2 (t8112).
> >>
> >> Each CPU cluster has its own register set, and frequency management is
> >> fully automated by the hardware; the driver only has to write one
> >> register. There is boost frequency support, but the hardware will only
> >> allow their use if only a subset of cores in a cluster are in
> >> non-deep-idle. Since we don't support deep idle yet, these frequencies
> >> are not achievable, but the driver supports them. They will remain
> >> disabled in the device tree until deep idle is implemented, to avoid
> >> confusing users.
> >
> > Out of curiosity, may I ask if this implies the need of a
> > synchronization mechanism on the Linux side? Or is the boost frequency
> > dynamically managed solely by HW/FW?
>
> It's managed by hardware - Linux gets to request whatever frequency it
> wants, and the hardware will limit it to what is achievable given the
> current idle states within the cluster (and it will change automatically
> with them). So if Linux asks for 3.2 GHz but there are no deep idle
> cores in the cluster, you get 3.0. If there's one deep idle core, you
> get 3.1 (I think). Three, 3.2. So this driver doesn't have to do
> anything (and will report the correct current-frequency as long as the
> per-SoC compatible is matched; without that this feature is disabled and
> it just reports the requested frequency). We could enable the boost
> states today just fine, it's just that they would never actually be
> reached by the hardware.

Thanks for sharing these details. It's always nice to know a bit more
about how the HW works!

From the reviewing point of view, I don't have more to add at this point!

Kind regards
Uffe