linux-kernel - Re: [PATCH v4 3/8] PM / devfreq: Use the available min/max frequency

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] [day] [month] [year] [list]

Date:   Wed, 18 Oct 2017 11:12:42 +0900
From:   Chanwoo Choi <cw00.choi@...sung.com>
To:     myungjoo.ham@...il.com
Cc:     Kyungmin Park <kyungmin.park@...sung.com>,
        "Rafael J . Wysocki" <rafael.j.wysocki@...el.com>,
        chanwoo@...nel.org,
        대인기 <inki.dae@...sung.com>,
        LKML <linux-kernel@...r.kernel.org>,
        Linux PM list <linux-pm@...r.kernel.org>
Subject: Re: [PATCH v4 3/8] PM / devfreq: Use the available min/max
 frequency

On 2017년 10월 18일 10:31, Chanwoo Choi wrote:
> Hi,
> 
> On 2017년 10월 17일 23:43, MyungJoo Ham wrote:
>> On Fri, Oct 13, 2017 at 4:48 PM, Chanwoo Choi <cw00.choi@...sung.com> wrote:
>>> The commit a76caf55e5b35 ("thermal: Add devfreq cooling") is able
>>> to disable OPP as a cooling device. In result, both update_devfreq()
>>> and {min|max}_freq_show() have to consider the 'opp->available'
>>> status of each OPP.
>>>
>>> So, this patch adds the 'scaling_{min|max}_freq' to struct devfreq
>>> in order to indicate the available mininum and maximum frequency
>>> by adjusting OPP interface such as dev_pm_opp_{disable|enable}().
>>> The 'scaling_{min|max}_freq' are used for on both update_devfreq()
>>> and {min|max}_freq_show().
>>>
>>> Signed-off-by: Chanwoo Choi <cw00.choi@...sung.com>
>>> ---
>>>  drivers/devfreq/devfreq.c | 36 ++++++++++++++++++++++++++++--------
>>>  include/linux/devfreq.h   |  4 ++++
>>>  2 files changed, 32 insertions(+), 8 deletions(-)
>>>
>>> diff --git a/drivers/devfreq/devfreq.c b/drivers/devfreq/devfreq.c
>>> index b6ba24e5db0d..9de013ffeb67 100644
>>> --- a/drivers/devfreq/devfreq.c
>>> +++ b/drivers/devfreq/devfreq.c
>> []
>>> @@ -494,6 +499,19 @@ static int devfreq_notifier_call(struct notifier_block *nb, unsigned long type,
>>>         int ret;
>>>
>>>         mutex_lock(&devfreq->lock);
>>> +
>>> +       devfreq->scaling_min_freq = find_available_min_freq(devfreq);
>>> +       if (!devfreq->scaling_min_freq) {
>>> +               mutex_unlock(&devfreq->lock);
>>> +               return -EINVAL;
>>> +       }
>>> +
>>> +       devfreq->scaling_max_freq = find_available_max_freq(devfreq);
>>> +       if (!devfreq->max_freq) {
>>
>> 1. s/max_freq/scaling/max_freq/ ??
> 
> My mistake. The scaling_max_freq is right. I'll fix it.
> 
>>
>> 2. what if thermal is not active or has never triggered any event and
>> the user has never stated max/min? (making scaling_*_freq zero)
> 
> 
> The devfreq-cooling.c of tmu uses the OPP interface
> and then OPP interface affect the scaling_min/max_freq of devfreq
> through dev_pm_opp_disable/enable(). So, even if 'thermal is not active
> or has never triggered any event', devfreq will use the OPP interface
> as a mandatory.
> 
> In result, I think that devfreq should maintain the correct frequency
> of scaling_min/max_freq indicating the 'limit minimum/maximum frequency
> requested by OPP interface' instead of zero.
> 
> So, I'll change the description of scaling_min/max_freq as following:
> (by devfreq-cooling -> by OPP interface)
> On v4:
> + * @scaling_min_freq:  Limit minimum frequency requested by devfreq-cooling
> + * @scaling_max_freq:  Limit maximum frequency requested by devfreq-cooling
> 
> On v5:
> + * @scaling_min_freq:  Limit minimum frequency requested by OPP interface
> + * @scaling_max_freq:  Limit maximum frequency requested by OPP interface
> 
> 
> And, this patch showed the wrong value of min/max_freq_show() by my mistake.
> I showed the 'min/max_freq' directly through min/max_freq_show()
> without comparing with scaling_min/max_freq. So, I'll fix this issue as following:
> ---------------
> On v5:
> static ssize_t min_freq_show(struct device *dev, struct device_attribute *attr,
>                              char *buf)
>  {
> -       return sprintf(buf, "%lu\n", to_devfreq(dev)->min_freq);
> +       struct devfreq *df = to_devfreq(dev);
> +
> +       return sprintf(buf, "%lu\n", MAX(df->scaling_min_freq, df->min_freq));
>  }
> 
>  static ssize_t max_freq_store(struct device *dev, struct device_attribute *attr,
> @@ -1161,7 +1183,9 @@ static ssize_t max_freq_store(struct device *dev, struct device_attribute *attr,
>  static ssize_t max_freq_show(struct device *dev, struct device_attribute *attr,
>                              char *buf)
>  {
> -       return sprintf(buf, "%lu\n", to_devfreq(dev)->max_freq);
> +       struct devfreq *df = to_devfreq(dev);
> +
> +       return sprintf(buf, "%lu\n", MIN(df->scaling_max_freq, df->max_freq));
> ---------------

If you agree my opinion, I'll send v5 patchset right now
because if patch3 gets the review, everything is done without patch8.
As I replied, I'll drop the patch8 from this patchset.

> 
> 
>>
>>> +               mutex_unlock(&devfreq->lock);
>>> +               return -EINVAL;
>>> +       }
>>> +
>>>         ret = update_devfreq(devfreq);
>>>         mutex_unlock(&devfreq->lock);
>>>
>>
>>
>>
> 
> 


-- 
Best Regards,
Chanwoo Choi
Samsung Electronics