Re: [PATCH] Documentation/cpu-freq: add intel-pstate.txt

[Dirk Brandewie <dirk.brandewie@gmail.com>]

On 01/03/2014 04:09 AM, Rafael J. Wysocki wrote:
> On Friday, January 03, 2014 05:01:13 PM Ramkumar Ramachandra wrote:
>> The Intel P-state driver is currently undocumented. Add some
>> documentation based on the cover-letter sent with the original series.
>>
>> Cc: Dirk Brandewie <dirk.brandewie@gmail.com>
>
> Dirk, what do you think?
>

This is accurate, along with the wording changes suggested by you and
Randy,
>> Cc: Rafael J. Wysocki <rjw@rjwysocki.net>
>> Cc: Viresh Kumar <viresh.kumar@linaro.org>
>> Signed-off-by: Ramkumar Ramachandra <artagnon@gmail.com>
>> ---
>>   I was looking for documentation on my cpufreq driver, intel_pstate,
>>   and found nothing: so, here's a small start. Although the original
>>   message said SandyBridge only, I noticed that
>>   /sys/devices/system/cpu/cpu0/cpufreq/scaling_driver says intel_pstate
>>   on my IvyBridge; so I figured subsequent patches introduced support
>>   for SandyBridge+.
>>
>>   I noticed two "policies" in the code: CPUFREQ_POLICY_PERFORMANCE and
>>   CPUFREQ_POLICY_POWERSAVE, but I have no idea how to switch from one
>>   to the other. The userspace tool cpupower seems to be too tightly
>>   tied to the acpi-cpufreq driver, and I'm not able to find any others.
>>
>>   Also, how well is the driver performing? I seem to be getting
>>   terrible battery life on my 2013 MBP.
>>
>>   Thanks.
>>
>>   Documentation/cpu-freq/intel-pstate.txt | 40 +++++++++++++++++++++++++++++++++
>>   1 file changed, 40 insertions(+)
>>   create mode 100644 Documentation/cpu-freq/intel-pstate.txt
>>
>> diff --git a/Documentation/cpu-freq/intel-pstate.txt b/Documentation/cpu-freq/intel-pstate.txt
>> new file mode 100644
>> index 0000000..fda74ab
>> --- /dev/null
>> +++ b/Documentation/cpu-freq/intel-pstate.txt
>> @@ -0,0 +1,40 @@
>> +Intel P-state driver
>> +--------------------
>> +
>> +This driver implements a scaling driver with an internal governor for
>> +Intel Core processors.  The driver follows the same model as the
>> +Transmeta scaling driver (longrun.c) and implements the setpolicy()
>> +instead of target().  Scaling drivers that implement setpolicy() are
>> +assmuned to implement internal governors by the cpufreq core. All the
>> +logic for selecting the current P state is contained within the driver
>> +no external governor is used by the cpufreq core.
>> +
>> +Only Intel SandyBridge+ processors are supported.
>
> That's not the case any more.

SandyBridge+ is accurate "only SandyBridge" is not

>
>> +
>> +New sysfs files for controlling P state selection have been added to
>> +/sys/devices/system/cpu/intel_pstate/
>> +
>> +      max_perf_pct: limits the maximum P state that will be requested by
>> +      the driver stated as a percentage of the avail performance.
>> +
>> +      min_perf_pct: limits the minimum P state that will be  requested by
>> +      the driver stated as a percentage of the avail performance.
>> +
>> +      no_turbo: limits the driver to selecting P states below the turbo
>> +      frequency range.
>> +
>
> It would be good to document them in Documentation/ABI/testing/ as well.
>
>> +The units for these for these files are purposely abstract and stated
>> +in terms of available performance and not frequency.  In idea that
>
> I would say "For contemporary Intel processor the frequency is controlled
> by the processor itself and the P-states exposed to software are related to
> performance levels" instead.
>
>> +frequency can be set to a single frequency is a fiction for Intel Core
>> +processors. Even if the scaling driver selects a single P state the
>> +actual frequency the processor will run at is selected by the
>> +processor itself.
>> +
>> +New debugfs files have also been added to /sys/kernel/debug/pstate_snb/
>> +
>> +      deadband
>> +      d_gain_pct
>> +      i_gain_pct
>> +      p_gain_pct
>> +      sample_rate_ms
>> +      setpoint
>>
>
> Thanks!
>