Re: [PATCH]CPU hotplug breaks wake_up_new_task

[Ashok Raj]

On Tue, May 31, 2005 at 12:35:09AM -0700, Shaohua Li wrote:
> 
>    Hi,
>    There is a race condition at wake_up_new_task at CPU hotplug case.
>    Say do_fork
>             copy_process  (which  sets  new  forked  task's  current cpu,
>    cpu_allowed)
>                    <-------- the new forked task's current cpu is offline
>            wake_up_new_task
>    wake_up_new_task will put the forked task into a dead cpu.
> 
>    Thanks,
>    Shaohua
> 
>    Signed-off-by: Shaohua Li<shaohua.li@xxxxxxxxx>
>    ---
> 
>     linux-2.6.11-rc5-mm1-root/kernel/sched.c            |              25
.... Deleted....
>     /*
>       *  wake_up_new_task  -  wake  up a newly created task for the first
>    time.
>      *
>    @@ -1426,9 +1430,20 @@ void fastcall wake_up_new_task(task_t *
>            runqueue_t *rq, *this_rq;
> 
>            rq = task_rq_lock(p, &flags);
>    -       BUG_ON(p->state != TASK_RUNNING);
>            this_cpu = smp_processor_id();
>            cpu = task_cpu(p);
>    +#ifdef CONFIG_HOTPLUG_CPU
>    +       while (!cpu_online(cpu)) {
>    +               cpu = task_select_online_cpu(cpu, p);
>    +               set_task_cpu(p, cpu);
>    +               task_rq_unlock(rq, &flags);
>    +               /* CPU hotplug might occur here */
>    +               rq = task_rq_lock(p, &flags);
>    +               this_cpu = smp_processor_id();
>    +               cpu = task_cpu(p);
>    +       }
>    +#endif

The while() loop doesnt look pretty here.. could you try to 
disable preempt, and see the problem goes away? or use 
get_cpu()/put_cpu() combo when you get this_cpu?

Just wondering if the code would be a little more simpler in this
case.

Nick should have something to say ...

Cheers,
ashok
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/