MIME-Version: 1.0
References: <20230125113418.455089-1-krzysztof.kozlowski@linaro.org> <ac3ed1da-fcd1-d8e9-7770-8a047f0e392d@linaro.org>
In-Reply-To: <ac3ed1da-fcd1-d8e9-7770-8a047f0e392d@linaro.org>
From:   Ulf Hansson <ulf.hansson@linaro.org>
Date:   Thu, 26 Jan 2023 13:45:12 +0100
Message-ID: <CAPDyKFr9ePUvi-h_jJByrfsBKp+Mu+oyn=qcgp=Ta_jjygEeLw@mail.gmail.com>
Subject: Re: [PATCH v4] cpuidle: psci: Do not suspend topology CPUs on PREEMPT_RT
To:     Daniel Lezcano <daniel.lezcano@linaro.org>
Cc:     Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>,
        "Rafael J. Wysocki" <rafael@kernel.org>,
        Lorenzo Pieralisi <lpieralisi@kernel.org>,
        Sudeep Holla <sudeep.holla@arm.com>, linux-pm@vger.kernel.org,
        linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
        Adrien Thierry <athierry@redhat.com>,
        Brian Masney <bmasney@redhat.com>,
        linux-rt-users@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, 25 Jan 2023 at 17:46, Daniel Lezcano <daniel.lezcano@linaro.org> wrote:
>
>
> Hi Krzysztof,
>
>
> On 25/01/2023 12:34, Krzysztof Kozlowski wrote:
> > The runtime Power Management of CPU topology is not compatible with
> > PREEMPT_RT:
> > 1. Core cpuidle path disables IRQs.
> > 2. Core cpuidle calls cpuidle-psci.
> > 3. cpuidle-psci in __psci_enter_domain_idle_state() calls
> >     pm_runtime_put_sync_suspend() and pm_runtime_get_sync() which use
> >     spinlocks (which are sleeping on PREEMPT_RT).
> >
> > Deep sleep modes are not a priority of Realtime kernels because the
> > latencies might become unpredictable.  On the other hand the PSCI CPU
> > idle power domain is a parent of other devices and power domain
> > controllers, thus it cannot be simply skipped (e.g. on Qualcomm SM8250).
> >
> > Disable the idle callbacks in cpuidle-psci and mark the domain as
> > always on.  This is a trade-off between making PREEMPT_RT working and
> > still having a proper power domain hierarchy in the system.
>
> Wouldn't make sense to rely on the latency constraint framework ?

The main problem is that for runtime PM there is a per device spinlock
being used, which becomes a sleepable lock on PREEMPT_RT.

In other words, the only simple solution is to avoid the calls to
runtime PM in the idle path.

[...]

Kind regards
Uffe