Re: [PATCH] sched/rt: fix rt timer activation/deactivation

[Kirill Tkhai <tkhai@yandex.ru>]

26.02.2014, 13:35, "Kirill Tkhai" <tkhai@yandex.ru>:
> 26.02.2014, 13:07, "Peter Zijlstra" <peterz@infradead.org>:
>
>>  On Wed, Feb 26, 2014 at 03:37:38AM +0100, Mike Galbraith wrote:
>>>   BTW, I noticed you can no longer turn the turn the noisy thing off since
>>>   we grew DL.  I added an old SGI boot parameter to tell it to go away.
>>  You're talking about the rt badnwidth timer, right?
>>
>>   1) why won't it go away with DL added?
>
> We know nothing about rt_time and rt_runtime values, when we are doing
> update_curr_dl().
>
> rt_time maybe not zero, so we increment rt_time.
>
> rt_rq->rt_runtime maybe less than maximum available for RT(and for dl too),
> because someone could borrow a part of rt_runtime during previous
> balance_runtime().
>
> If dl task uses all available runtime, we always will have "idle = 0"
>
>         if (rt_rq->rt_time || rt_rq->rt_nr_running)
>                 idle = 0;
>
> in do_sched_rt_period_timer(). Timer will be restarted.
>
> This is boundary case, of course.

I wrote this in assumtion, that patch like Juri's is applied
[https://lkml.org/lkml/2014/2/25/307].

In current tip.git state above is always.


>>   2) it should never appear when !rt_bandwidth_enabled(), so if you set
>>   sysctl_sched_rt_runtime to -1 all this should go away already, no extra
>>   patches required.
>
> --
> 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/