[PATCH] CPUIdle: Reevaluate C-states under CPU load to favor deeper C-states

[Kevin Hilman]

melwyn lobo <linux.melwyn at gmail.com> writes:

> Hey Kevin,
> I would like to try out this patch in my platform see the benefits
> that you are reporting. But there is one issue in this patch. You have
> not initialized "hrtimer_timeout" variable.
> This will always be 0 right ?.

Correct.

The generic code defaults to zero so that the default behavior with this
patch is unchanged from previous behavior.  In order to use this
feature, your platform-specific code which creates your C-states sets
the per-C-state timer values.

Kevin

> Thanks,
> -M
>
> On Wed, Nov 9, 2011 at 11:36 PM, Chalhoub, Nicole <n-chalhoub at ti.com> wrote:
>> Hi Deepthi,
>>
>>>
>> Texas Instruments France SA, 821 Avenue Jack Kilby, 06270 Villeneuve Loubet. 036 420 040 R.C.S Antibes. Capital de EUR 753.920
>>
>> -----Original Message-----
>>> From: Deepthi Dharwar [mailto:deepthi at linux.vnet.ibm.com]
>>> Sent: Wednesday, November 09, 2011 12:13 PM
>>> To: Hilman, Kevin
>>> Cc: linux-kernel at vger.kernel.org; Arjan van de Ven; linux-arm-
>>> kernel at lists.infradead.org; linux-omap at vger.kernel.org; linux-
>>> pm at lists.linux-foundation.org; Chalhoub, Nicole
>>> Subject: Re: [PATCH] CPUIdle: Reevaluate C-states under CPU load to favor
>>> deeper C-states
>>
>> [...]
>>> By setting timers when we enter non-deepest C-state possible, such that
>>> when it fires we
>>> can re-evaluate and try moving into deeper and deeper C-states enhancing
>>> the
>>> power savings is a good feature to have.
>>>
>>> Looking at the current implementation, is it possible to have it as
>>> configurable option
>>> where one can enable/disable this functionality through the backhand
>>> driver ?
>>
>> The timeout values of the c state timers are set in the backhand driver.
>> By setting the timeout to 0 the timers will not fire so you'll not have this functionality enabled
>>
>>> Also I am thinking, instead of having them in governor
>>> wouldnt it be a good idea to have it implemented in
>>> the backhand driver itself ?
>>> --Deepthi
>>
>>
>> In fact each C-state had its own configurable timer, so it is a parameter characterizing a C-state as it is for the exit_latency and target_residency parameters.
>> And we wanted the timer to fire only when we do not go in deep Cstate due to a high load. This decision is made in the CPU idle governor. So the functionality should be seen from the governor..
>>
>> Thanks and Regards
>> Nicole
>>
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
>> the body of a message to majordomo at vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> Please read the FAQ at  http://www.tux.org/lkml/
>
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel