Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:6 as permitted sender) client-ip=2620:137:e000::3:6;
Date:   Fri, 8 Dec 2023 15:21:08 +0530
From:   Viresh Kumar <viresh.kumar@linaro.org>
To:     David Dai <davidai@google.com>
Cc:     "Rafael J. Wysocki" <rafael@kernel.org>,
        Rob Herring <robh+dt@kernel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        Conor Dooley <conor+dt@kernel.org>,
        Sudeep Holla <sudeep.holla@arm.com>,
        Saravana Kannan <saravanak@google.com>,
        Quentin Perret <qperret@google.com>,
        Masami Hiramatsu <mhiramat@google.com>,
        Will Deacon <will@kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Vincent Guittot <vincent.guittot@linaro.org>,
        Marc Zyngier <maz@kernel.org>,
        Oliver Upton <oliver.upton@linux.dev>,
        Dietmar Eggemann <dietmar.eggemann@arm.com>,
        Pavan Kondeti <quic_pkondeti@quicinc.com>,
        Gupta Pankaj <pankaj.gupta@amd.com>,
        Mel Gorman <mgorman@suse.de>, kernel-team@android.com,
        linux-pm@vger.kernel.org, devicetree@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH v4 2/2] cpufreq: add virtual-cpufreq driver
Message-ID: <20231208095108.mg5bz7yjvla6syk2@vireshk-i7>
References: <20231111014933.1934562-1-davidai@google.com>
 <20231111014933.1934562-3-davidai@google.com>
 <20231115062932.vz2tyg6wgux5lx6t@vireshk-i7>
 <CABN1KCKfCWB6fVAuMSN9AdJOe-zueNMPFUdDnKLcq-uetz2ZFQ@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <CABN1KCKfCWB6fVAuMSN9AdJOe-zueNMPFUdDnKLcq-uetz2ZFQ@mail.gmail.com>
Precedence: bulk

On 08-12-23, 10:18, David Dai wrote:
> Hi Viresh,
> 
> Apologies for the late reply,
> 
> On Wed, Nov 15, 2023 at 3:29 PM Viresh Kumar <viresh.kumar@linaro.org> wrote:
> >
> > On 10-11-23, 17:49, David Dai wrote:
> > > diff --git a/drivers/cpufreq/virtual-cpufreq.c b/drivers/cpufreq/virtual-cpufreq.c
> > > +static unsigned int virt_cpufreq_set_perf(struct cpufreq_policy *policy)
> > > +{
> > > +     writel_relaxed(policy->cached_target_freq,
> >
> > Drivers shouldn't be using the cached_target_freq directly. Use the target freq
> > or index passed from cpufreq core.
> 
> We were trying to avoid rounding to frequency table entries to provide
> more accurate frequency requests. However, we didn't find any
> significant power or performance regressions using the frequencies
> from the table, so I'll send another patch series using your
> suggestion.
> 
> >
> > > +static int virt_cpufreq_cpu_exit(struct cpufreq_policy *policy)
> > > +{
> > > +     topology_clear_scale_freq_source(SCALE_FREQ_SOURCE_VIRT, policy->related_cpus);
> > > +     kfree(policy->freq_table);

This becomes a dangling pointer for a very short amount of time. There may or
may not be a actual race here and so I said the ordering must be just the
opposite anyway.

> > > +     policy->freq_table = NULL;

And I thought this isn't required too since the core is going the free the
policy structure right after returning from here. But maybe it is not a
guarantee that the core provides (the code can change in future) and so be
better to unset it anyway.

> > No need of doing this. Also the order of above two calls is wrong anyway.
> 
> Can you clarify this point a bit more? Are you suggesting to just
> remove setting policy->freq_table to NULL and swap the ordering
> freeing the freq_table vs clearing the topology source? I can
> alternatively use dev_pm_opp_free_cpufreq_table to mirror the init.

That would be better actually, to let a single piece of code manage this :)

-- 
viresh