Re: [PATCH 0/3] sched: idle: Provide the basis to integrate cpuidle

["Rafael J. Wysocki" <rjw@rjwysocki.net>]

On Thursday, April 24, 2014 02:24:48 PM Daniel Lezcano wrote:
> This patchset provides three patches for the basis to integrate cpuidle with
> the scheduler.
> 
> The first patch is a cleanup.
> The second one adds the sched balance option as requested by Ingo.
> The third one stores the idle state a cpu is and adds a rcu_barrier() to
>  prevent races when using the pointed object.
> 
> This patchset is based on top of v3.15-rc2.
> 
> This patchset does not modify the behavior of the scheduler.
> 
> Taking into account the cpuidle information from the scheduler will be
> posted in a separate patchset in order to keep focused on the right decisions
> the scheduler should take regarding the policy vs idle parameters.
> 
> Daniel Lezcano (3):
>   sched: idle: Encapsulate the code to compile it out
>   sched: idle: Add sched balance option
>   sched: idle: Store the idle state the cpu is

Please rebase patch [1/3] on top of https://patchwork.kernel.org/patch/4021831/

I agree with the Amit's comment regarding the power supply assumption in
patch [2/3].

>  drivers/cpuidle/cpuidle.c    |    6 ++
>  include/linux/sched/sysctl.h |   14 ++++
>  kernel/sched/fair.c          |   92 ++++++++++++++++++++++-
>  kernel/sched/idle.c          |  169 +++++++++++++++++++++++-------------------
>  kernel/sched/sched.h         |    5 ++
>  kernel/sysctl.c              |   11 +++
>  6 files changed, 220 insertions(+), 77 deletions(-)
> 
> 

-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.