[RFC PATCH v3 3/6] sched: pack small tasks

[Vincent Guittot]

On 27 March 2013 09:46, Peter Zijlstra <peterz at infradead.org> wrote:
> On Tue, 2013-03-26 at 08:29 -0700, Arjan van de Ven wrote:
>> > Isn't this basically related to picking the NO_HZ cpu; if the system
>> > isn't fully symmetric with its power gates you want the NO_HZ cpu to be
>> > the 'special' cpu. If it is symmetric we really don't care which core
>> > is left 'running' and we can even select a new pack cpu from the idle
>> > cores once the old one is fully utilized.
>>
>> you don't really care much sure, but there's some advantages for sorting "all the way left",
>> e.g. to linux cpu 0.
>> Some tasks only run there, and interrupts tend to be favored to that cpu as well on x86.
>
> Right, and I suspect all the big-little nonsense will have the little
> cores on low numbers as well (is this architected or can a creative
> licensee screw us over?)

It's not mandatory to have little cores on low numbers even if it's advised

>
> So find_new_ilb() already does cpumask_first(), so it has a strong
> leftmost preference. We just need to make sure it indeed does the right
> thing and doesn't have some unintended side effect.
>