lkml.org 
[lkml]   [2009]   [Dec]   [2]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [patch] f83f9ac causes tasks running at MAX_PRIO
From
Date
On Wed, 2009-12-02 at 12:46 +0100, Peter Zijlstra wrote:

> init_idle() doing:
> idle->prio = idle->normal_prio = MAX_PRIO;
>
> Which will propagate... like reported.
>
> Now, since the idle-threads usually run on &idle_sched_class, nobody
> will actually look at their ->prio, so having that out-of-range might
> make sense.
>
> Just needs to get fixed up when we fork a normal thread, which would be
> in sched_fork(), now your call to normal_prio() fixes this by setting
> everything to ->static_prio for SCHED_OTHER tasks, however
>
> migration_call()
> CPU_DEAD:
> rq->idle->static_prio = MAX_PRIO;
>
> spoils that too..

Darn.

> Ingo, any particular reason we set idle threads at MAX_PRIO? Can't we
> simply do something like below and be done with it?

Hysterical reasons? That might have been a doorstop conversion kit for
O(1), but boots fine with CFS, and prio 40 tasks are history.

> ---
> kernel/sched.c | 2 --
> 1 files changed, 0 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/sched.c b/kernel/sched.c
> index c0e4e9d..5ad5a66 100644
> --- a/kernel/sched.c
> +++ b/kernel/sched.c
> @@ -6963,7 +6963,6 @@ void __cpuinit init_idle(struct task_struct *idle,
> int cpu)
> __sched_fork(idle);
> idle->se.exec_start = sched_clock();
>
> - idle->prio = idle->normal_prio = MAX_PRIO;
> cpumask_copy(&idle->cpus_allowed, cpumask_of(cpu));
> __set_task_cpu(idle, cpu);
>
> @@ -7667,7 +7666,6 @@ migration_call(struct notifier_block *nfb,
> unsigned long action, void *hcpu)
> spin_lock_irq(&rq->lock);
> update_rq_clock(rq);
> deactivate_task(rq, rq->idle, 0);
> - rq->idle->static_prio = MAX_PRIO;
> __setscheduler(rq, rq->idle, SCHED_NORMAL, 0);
> rq->idle->sched_class = &idle_sched_class;
> migrate_dead_tasks(cpu);
>



\
 
 \ /
  Last update: 2009-12-02 13:51    [W:0.163 / U:0.060 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site