linux-kernel - Re: [PATCH v3 3/4] OPP: Add support of "opp-microwatt" for advanced EM registration

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 for Android: free password hash cracker in your pocket

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

Message-ID: <9c8f15a3-563f-3bed-61a4-2f72b0c20e89@arm.com>
Date:   Thu, 24 Feb 2022 09:33:18 +0000
From:   Lukasz Luba <lukasz.luba@....com>
To:     Viresh Kumar <viresh.kumar@...aro.org>
Cc:     linux-kernel@...r.kernel.org, dietmar.eggemann@....com,
        rafael@...nel.org, daniel.lezcano@...aro.org, nm@...com,
        sboyd@...nel.org, mka@...omium.org, dianders@...omium.org,
        robh+dt@...nel.org, devicetree@...r.kernel.org,
        linux-pm@...r.kernel.org
Subject: Re: [PATCH v3 3/4] OPP: Add support of "opp-microwatt" for advanced
 EM registration



On 2/24/22 09:13, Viresh Kumar wrote:
> On 24-02-22, 08:11, Lukasz Luba wrote:
>> diff --git a/drivers/opp/of.c b/drivers/opp/of.c
>>   
>> +/*
>> + * Callback function provided to the Energy Model framework upon registration.
>> + * It provides the power used by @dev at @kHz if it is the frequency of an
>> + * existing OPP, or at the frequency of the first OPP above @kHz otherwise
>> + * (see dev_pm_opp_find_freq_ceil()). This function updates @kHz to the ceiled
>> + * frequency and @mW to the associated power.
>> + *
>> + * Returns 0 on success or a proper -EINVAL value in case of error.
>> + */
>> +static int __maybe_unused
>> +_get_opp_power(unsigned long *mW, unsigned long *kHz, struct device *dev)
> 
> Lets call it _get_dt_opp_power() or _get_dt_power() ?

OK, I'll return to _get_dt_power()

> 
>> +{
>> +	struct dev_pm_opp *opp;
>> +	unsigned long opp_freq, opp_power;
>> +
>> +	/* Find the right frequency and related OPP */
>> +	opp_freq = *kHz * 1000;
>> +	opp = dev_pm_opp_find_freq_ceil(dev, &opp_freq);
>> +	if (IS_ERR(opp))
>> +		return -EINVAL;
>> +
>> +	opp_power = dev_pm_opp_get_power(opp);
> 
> As I said in 2/4, this should really give the total instead.

make sense

> 
>> +	dev_pm_opp_put(opp);
>> +	if (!opp_power)
>> +		return -EINVAL;
>> +
>> +	*kHz = opp_freq / 1000;
>> +	*mW = opp_power / 1000;
>> +
>> +	return 0;
>> +}
>> +
>>   /*
>>    * Callback function provided to the Energy Model framework upon registration.
>>    * This computes the power estimated by @dev at @kHz if it is the frequency
>> @@ -1488,6 +1520,24 @@ static int __maybe_unused _get_power(unsigned long *mW, unsigned long *kHz,
>>   	return 0;
>>   }
>>   
>> +static bool _of_has_opp_microwatt_property(struct device *dev)
>> +{
>> +	unsigned long power, freq = 0;
>> +	struct dev_pm_opp *opp;
>> +
>> +	/* Check if at least one OPP has needed property */
>> +	opp = dev_pm_opp_find_freq_ceil(dev, &freq);
>> +	if (IS_ERR(opp))
>> +		return false;
>> +
>> +	power = dev_pm_opp_get_power(opp);
>> +	dev_pm_opp_put(opp);
>> +	if (!power)
> 
> What if this particular frequency has 0 power mentioned for some
> reason :)

If that power 0 would be allowed here, then in next step when EM
calls active_power() we check !power and report dev_err().
IPA design would also not accept the power=0

> 
> Instead of this heavy stuff, just pick the first OPP from the opp
> table and see its power-value.

It is the first opp: freq=0.
You mean by parsing the the DT node instead, like I had in v2 version?

> 
>> +		return false;
>> +
>> +	return true;
>> +}
>> +
>>   /**
>>    * dev_pm_opp_of_register_em() - Attempt to register an Energy Model
>>    * @dev		: Device for which an Energy Model has to be registered
>> @@ -1517,6 +1567,14 @@ int dev_pm_opp_of_register_em(struct device *dev, struct cpumask *cpus)
>>   		goto failed;
>>   	}
>>   
>> +	/* First, try to find more precised Energy Model in DT */
>> +	if (_of_has_opp_microwatt_property(dev)) {
>> +		struct em_data_callback em_dt_cb = EM_DATA_CB(_get_opp_power);
>> +
>> +		return em_dev_register_perf_domain(dev, nr_opp, &em_dt_cb,
>> +						   cpus, true);
>> +	}
>> +
> 
> We have another em_dev_register_perf_domain() down the line, lets keep
> only one such call and get it a callback that should be set in an
> if/else loop.

OK