Re: [RFT][PATCH 0/3] cpufreq / PM: QoS: Introduce frequency QoS and use it in cpufreq

From: Sudeep Holla <sudeep.holla@arm.com>
To: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: Linux PM <linux-pm@vger.kernel.org>,
	Linux ACPI <linux-acpi@vger.kernel.org>,
	LKML <linux-kernel@vger.kernel.org>,
	Viresh Kumar <viresh.kumar@linaro.org>,
	Sudeep Holla <sudeep.holla@arm.com>,
	Dmitry Osipenko <digetx@gmail.com>
Subject: Re: [RFT][PATCH 0/3] cpufreq / PM: QoS: Introduce frequency QoS and use it in cpufreq
Date: Wed, 16 Oct 2019 15:23:43 +0100	[thread overview]
Message-ID: <20191016142343.GB5330@bogus> (raw)
In-Reply-To: <2811202.iOFZ6YHztY@kreacher>

On Wed, Oct 16, 2019 at 12:37:58PM +0200, Rafael J. Wysocki wrote:
> Hi All,
>
> The motivation for this series is to address the problem discussed here:
>
> https://lore.kernel.org/linux-pm/5ad2624194baa2f53acc1f1e627eb7684c577a19.1562210705.git.viresh.kumar@linaro.org/T/#md2d89e95906b8c91c15f582146173dce2e86e99f
>
> and also reported here:
>
> https://lore.kernel.org/linux-pm/20191015155735.GA29105@bogus/
>
> Plus, generally speaking, using the policy CPU as a proxy for the policy
> with respect to PM QoS does not feel particularly straightforward to me
> and adds extra complexity.
>
> Anyway, the first patch adds frequency QoS that is based on "raw" PM QoS (kind
> of in analogy with device PM QoS) and is just about min and max frequency
> requests (no direct relationship to devices).
>
> The second patch switches over cpufreq and its users to the new frequency QoS.
> [The Fixes: tag has been tentatively added to it.]
>
> The third one removes frequency request types from device PM QoS.
>
> Unfortunately, the patches are rather big, but also they are quite
> straightforward.
>
> I didn't have the time to test this series, so giving it a go would be much
> appreciated.

Thanks for the spinning these patches so quickly.

I did give it a spin, but unfortunately it doesn't fix the bug I reported.
So I looked at my bug report in detail and looks like the cpufreq_driver
variable is set to NULL at that point and it fails to dereference it
while trying to execute:
	ret = cpufreq_driver->verify(new_policy);
(Hint verify is at offset 0x1c/28)

So I suspect some race as this platform with bL switcher tries to
unregister and re-register the cpufreq driver during the boot.

I need to spend more time on this as reverting the initial PM QoS patch
to cpufreq.c makes the issue disappear.

--
Regards,
Sudeep