linux-kernel - Re: [PATCH] [RFC] cpufreq: can't raise max frequency with cpu

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]

Date:	Tue, 18 Dec 2012 21:45:32 -0800
From:	Doug Anderson <dianders@...omium.org>
To:	amit daniel kachhap <amit.daniel@...sung.com>
Cc:	Sonny Rao <sonnyrao@...omium.org>, linux-pm@...r.kernel.org,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	Zhang Rui <rui.zhang@...el.com>,
	Sameer Nanda <snanda@...omium.org>
Subject: Re: [PATCH] [RFC] cpufreq: can't raise max frequency with cpu_thermal

Amit,

On Tue, Dec 18, 2012 at 8:17 PM, amit daniel kachhap
<amit.daniel@...sung.com> wrote:
> On Tue, Dec 18, 2012 at 12:29 AM, Sonny Rao <sonnyrao@...omium.org> wrote:
>> The cpu_thermal generic thermal management code has a bug where once
>> max cpu frequency has been lowered in sysfs (scaling_max_freq) it is
>> not possible to raise the max back up later.  The bug is that the
>> notifer gets called by __cpufreq_set_policy() before the user policy
>> max is raised, and is incorrectly trying to enforce the max frequency
>> policy even when we are trying to change the policy.  It is also not
>> clear why this driver is looking at the user policy since it is
>> primarily supposed to enforce thermal policy, not user set policy.
>
> Hi Sunny,
>
> I am not sure if this change is needed.

Do you have a machine that's running with your code?  Can you go into
sysfs (/sys/devices/system/cpu/cpu0/cpufreq/) and try lowering then
raising the max frequency by doing something like this (assumes that
you can scale down to 200MHz):

  cd /sys/devices/system/cpu/cpu0/cpufreq/
  OLD_VAL=$(cat scaling_max_freq)
  cat scaling_min_freq > scaling_max_freq
  echo ${OLD_VAL} > scaling_max_freq

  echo "$(cat scaling_max_freq) should be ${OLD_VAL}.  Is it?"

...when I run the above without Sonny's patch on my system I see:
  200000 should be 1700000. Is it?

...after Sonny's patch then the above works.

> There is a check in cpufreq_thermal_notifier function to return 0 if
> notify_device == NOTIFY_INVALID. So the user will be always able to
> change the max frequency in normal situation. Did you tested this for
> some corner cases?
> The reason behind putting this check is that I don't want to override
> the user constraints.
>
> Thanks,
> Amit Daniel
>
>>
>> Signed-off-by: Sonny Rao <sonnyrao@...omium.org>
>> ---
>>  drivers/thermal/cpu_cooling.c |    4 ----
>>  1 files changed, 0 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/thermal/cpu_cooling.c b/drivers/thermal/cpu_cooling.c
>> index 836828e..63bc708 100644
>> --- a/drivers/thermal/cpu_cooling.c
>> +++ b/drivers/thermal/cpu_cooling.c
>> @@ -219,10 +219,6 @@ static int cpufreq_thermal_notifier(struct notifier_block *nb,
>>         if (cpumask_test_cpu(policy->cpu, &notify_device->allowed_cpus))
>>                 max_freq = notify_device->cpufreq_val;
>>
>> -       /* Never exceed user_policy.max*/
>> -       if (max_freq > policy->user_policy.max)
>> -               max_freq = policy->user_policy.max;
>> -
>>         if (policy->max != max_freq)
>>                 cpufreq_verify_within_limits(policy, 0, max_freq);
>>
>> --
>> 1.7.7.3
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
>> the body of a message to majordomo@...r.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> Please read the FAQ at  http://www.tux.org/lkml/

-Doug
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/