Re: /proc/timer_list and weird behavior with dropbear

[Nathan Zimmer <nzimmer@sgi.com>]

On 07/19/2013 12:03 PM, Holger Hans Peter Freyther wrote:
> On Fri, Jul 19, 2013 at 10:45:15AM -0500, Nathan Zimmer wrote:
>
>> I hadn't noticed anything.
>> Let me try your program and see what I may have missed.
> Hi,
>
> I neither know the semantics of the timer_list nor how to use
> seq_file correctly. What happens is that timer_list_next will only
> be called once. This means that iter->cpu will never be increased.
>
> This just moves to the next CPU when stop is called (e.g. nothing
> was added once the print_tickdevice was printed). Do you think
> this could be correct?
>
>
>
> diff --git a/kernel/time/timer_list.c b/kernel/time/timer_list.c
> index 3bdf283..8d36a3d 100644
> --- a/kernel/time/timer_list.c
> +++ b/kernel/time/timer_list.c
> @@ -327,8 +327,10 @@ static void *timer_list_next(struct seq_file *file, void *v, loff_t *offset)
>   	return timer_list_start(file, offset);
>   }
>   
> -static void timer_list_stop(struct seq_file *seq, void *v)
> +static void timer_list_stop(struct seq_file *file, void *v)
>   {
> +	struct timer_list_iter *iter = file->private;
> +	iter->cpu = cpumask_next(iter->cpu, cpu_online_mask);
>   }
>   
>   static const struct seq_operations timer_list_sops = {

That certainly does make the issue go away.
I think a better solution would be to have an increment in the 
timer_list_start.