Re: [PATCH] sched: Don't allow priority switch to realtime when the task doesn't belong to init_task_group and when CONFIG_RT_GROUP_SCHED isn't set

[Bharata B Rao <bharata@linux.vnet.ibm.com>]

On Sun, Nov 23, 2008 at 02:11:16AM +0100, Peter Zijlstra wrote:
> On Thu, 2008-11-20 at 11:48 +0530, Bharata B Rao wrote:
> > Applies on 2.6.28-rc5.
> > 
> > With CONFIG_RT_GROUP_SCHED not set, don't allow a task's priority switch
> > to realtime if the task isn't part of init_task_group.
> > 
> > A task belonging to a fair group could use sched_setscheduler/sched_setparam
> > to become a realtime task. If such a task belongs to one of the
> > child groups of init_task_group and if CONFIG_RT_GROUP_SCHED is not set,
> > then it ends up getting queued in init_task_group's runqueue.
> > So we have a situation where, a task belongs to one group (child)
> > but ends in the runqueue of another group (init_task_group).
> > This does not look correct.
> > 
> > Fix this by failing such priority change requests in sched_setscheduler()
> > and sched_setparam().
> 
> NAK
> 
> !RT_GROUP means the RT tasks should be fully invariant to any grouping
> configuration. This patch breaks that.
> 
> Furthermore your justification for this is plain wrong, you write as if
> an RT tasks can belong to any grouping (in the !RT_GROUP case), by the

Hmm... but that's I the impression I got when I started an RT
task in a subgroup and saw it listed in the tasks list of the cgroup.
So if RT tasks don't care about grouping (!RT_GROUP_SCHED), may be then
such tasks shoudn't be shown in the cgroup tasks list. And may be as Ingo
suggested they should be moved to init_task_group. Without that it makes one
think that a task is contributing to the cpu time of a group in which
it was started, but it in reality it is not.

Regards,
Bharata.