Re: [patch] CFS scheduler, -v15

[Srivatsa Vaddagiri <vatsa@in.ibm.com>]

On Wed, Jun 06, 2007 at 04:07:01PM +0530, Balbir Singh wrote:
> Dmitry Adamushko wrote:
> > On 06/06/07, Srivatsa Vaddagiri <vatsa@in.ibm.com> wrote:
> >> On Wed, Jun 06, 2007 at 09:01:43AM +0200, Ingo Molnar wrote:
> >> > [...] and my tree already contains the fixes for rt task's
> >> > exec_start.
> >>
> >> Can I have this snapshot pls? I have to deal with the same issue when
> >> the current task switches groups and I was planning to fix it by
> >> introducing a set_curr_task() method in fair_sched_class which
> >> initializes exec_start and other fields for that task.
> > 
> > Hum.. what about accounting 'exec_time' and updating 'exec_start' in
> > rt_sched :: dequeue_task_rt() instead (like update_curr() does it in
> > dequeue_task_fair())?

[For some reason Dimitry's mail has not made it to my inbox yet ..]

> > This way, on RT -> NORMAL transition.. some 'delta_exec' ( between
> > deactivate_task() ---> activate_task() ) will be accounted later as if
> > the task was 'sched_fair_class' during this time.. which I think makes
> > some sense. What do you think?
> > 
> 
> Why not do it explicitly in __setscheduler() if the new policy is SCHED_NORMAL
> or SCHED_BATCH. 

Yes this is the approach I prefer, because we burden the fast/normal
path less that way (RT->NORMAL transition is not common). That's why I
was considering a set_curr_task() method in fair_sched_class which will
be invoked in __setscheduler() if the new policy of currently running
task happens to be SCHED_NORMAL/BATCH. Alternately if the new policy of
currently running task happens to be SCHED_FIFO (and its old policy was
SCHED_NORMAL) we need to invoke put_prev_task() method (so that
fair_clock etc is updated based on outgoing task's execution time in
SCHED_NORMAL class).

>You could also add the smarts to deactivate_task()


-- 
Regards,
vatsa