linux-kernel - Re: [PATCH 02/17] PM: EM: Find first CPU online while updating OPP efficiency

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <0cda1fc9-2e99-66a2-b833-fe5be676d815@arm.com>
Date:   Wed, 10 May 2023 08:08:03 +0100
From:   Lukasz Luba <lukasz.luba@....com>
To:     Pierre Gondois <pierre.gondois@....com>
Cc:     dietmar.eggemann@....com, rui.zhang@...el.com, rafael@...nel.org,
        linux-pm@...r.kernel.org, linux-kernel@...r.kernel.org,
        amit.kucheria@...durent.com, amit.kachhap@...il.com,
        daniel.lezcano@...aro.org, viresh.kumar@...aro.org,
        len.brown@...el.com, pavel@....cz, ionela.voinescu@....com,
        rostedt@...dmis.org, mhiramat@...nel.org
Subject: Re: [PATCH 02/17] PM: EM: Find first CPU online while updating OPP
 efficiency



On 4/11/23 16:40, Pierre Gondois wrote:
> Hello Lukasz,
> 
> On 3/14/23 11:33, Lukasz Luba wrote:
>> The Energy Model might be updated at runtime and the energy efficiency
>> for each OPP may change. Thus, there is a need to update also the
>> cpufreq framework and make it aligned to the new values. In order to
>> do that, use a first online CPU from the Performance Domain.
>>
>> Signed-off-by: Lukasz Luba <lukasz.luba@....com>
>> ---
>>   kernel/power/energy_model.c | 11 +++++++++--
>>   1 file changed, 9 insertions(+), 2 deletions(-)
>>
>> diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
>> index 265d51a948d4..3d8d1fad00ac 100644
>> --- a/kernel/power/energy_model.c
>> +++ b/kernel/power/energy_model.c
>> @@ -246,12 +246,19 @@ em_cpufreq_update_efficiencies(struct device 
>> *dev, struct em_perf_state *table)
>>       struct em_perf_domain *pd = dev->em_pd;
>>       struct cpufreq_policy *policy;
>>       int found = 0;
>> -    int i;
>> +    int i, cpu;
>>       if (!_is_cpu_device(dev) || !pd)
>>           return;
> 
> Since dev is a CPU, I think it shouldn be possible to get the cpu id via 
> 'dev->id'.
> If so the code below should not be necessary anymore.

When you look at the code it does two things.
It tries to get the CPU id - this might be similar to what you
have proposed with the 'dev->id' but it's also looking at CPUs
which are 'active'. The 'dev' that we have might come from
some place, e.g. thermal cooling, which had a first CPU in
the domain stored somewhere. That CPU might be sometimes
not active, but the rest of the CPUs in the domain might be
running. We have to find an active CPU id and then we get the
'policy'.

> 
>> -    policy = cpufreq_cpu_get(cpumask_first(em_span_cpus(pd)));
>> +    /* Try to get a CPU which is online and in this PD */
>> +    cpu = cpumask_first_and(em_span_cpus(pd), cpu_active_mask);
>> +    if (cpu >= nr_cpu_ids) {
>> +        dev_warn(dev, "EM: No online CPU for CPUFreq policy\n");
>> +        return;
>> +    }
>> +
>> +    policy = cpufreq_cpu_get(cpu);
>>       if (!policy) {
>>           dev_warn(dev, "EM: Access to CPUFreq policy failed");
>>           return;
> 
> Regards,
> Pierre