Re: [PATCH, v9 3/3] cgroups: introduce timer slack controller

[Kirill A. Shutemov]

On Mon, Mar 14, 2011 at 04:46:52PM -0700, Andrew Morton wrote:
> On Mon, 14 Mar 2011 16:05:24 +0200
> Kirill A. Shutemov" <kirill@xxxxxxxxxxxxx> wrote:
> 
> > +Overview
> > +--------
> > +
> > +Every task_struct has timer_slack_ns value. This value uses to round up
> > +poll() and select() timeout values. This feature can be useful in
> > +mobile environment where combined wakeups are desired.
> > +
> > +Originally, prctl() was the only way to change timer slack value of
> > +a process. So you was not able change timer slack value of another
> > +process.
> > +
> > +cgroup subsys "timer_slack" implements timer slack controller. It
> > +provides a way to set minimal timer slack value for a group of tasks.
> > +If a task belongs to a cgroup with minimal timer slack value higher than
> > +task's value, cgroup's value will be applied.
> > +
> > +Timer slack controller allows to implement setting timer slack value of
> > +a process based on a policy. For example, you can create foreground and
> > +background cgroups and move tasks between them based on system state.
> 
> (quoting myself from last time)
> 
> Why do we need a cgroup for this as opposed to (say) inheritance over
> fork(), or a system-wide knob, or a per-process/threadgroup knob, or
> just leaving the existing code as-is?  Presumably you felt that a
> cgroup approach is better for manageability, but you didn't tell us
> about this and you didn't explore alternative ways of solving the
> problem-which-you-didn't-describe.

The main goal is reducing wakeups without affecting user experience.
Only userspace knows which tasks are important for UX. So we needed a
mechanism to define a policy in userspace.

Currently timer slack value inheritance over fork().

System-wide/threadgroup knob doesn't allow to change timer slack value of
a particular process. It's needed for foreground/background use-case.

With per-process know we have to reinvent grouping of tasks in userspace.

cgroups is a good solution for it, from my point of view.

> 
> I'm still having trouble seeing why we should merge this.  Who will use
> it, and for what reason and what benefits will they see?  Quantified
> benefits, if possible!
> 

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