Re: [RFC] Extend mwait idle to optimize away IPIs when possible

[Yong Zhang <yong.zhang0@gmail.com>]

On Fri, Feb 10, 2012 at 06:11:58PM -0800, Venki Pallipadi wrote:
> On Fri, Feb 10, 2012 at 11:19 AM, Peter Zijlstra <peterz@infradead.org> wrote:
> > On Mon, 2012-02-06 at 12:42 -0800, Venkatesh Pallipadi wrote:
> >> smp_call_function_single and ttwu_queue_remote sends unconditional IPI
> >> to target CPU. However, if the target CPU is in mwait based idle, we can
> >> do IPI-less wakeups using the magical powers of monitor-mwait.
> >
> > So I was thinking, why not change native_smp_send_reschedule() and
> > native_send_call_func_single_ipi() and keep the change entirely inside
> > the arch?
> >
> > Ideally its only APIC/idle that know about this detail, the scheduler
> > (or other consumers) really don't care about how the other cpu comes to
> > run the callback.
> >
> 
> OK. Moving most of this into arch code will be cleaner. But, Yong
> mentioned in this thread that he was looking to do something similar
> on MIPS. So, we may end up with some code duplication though..

Yeah, most of the things could be shared. Such as the added logic when
return from idle and the hooks into ttwu_queue_remote() and
generic_exec_single().

BTW, I think we should also let the arch provide a wrapper for
TIF_IPI_PENDING, such as tsk_ipi_pending().

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

-- 
Only stand for myself