Re: [PATCH] Wait to remove active timer when rescheduling hrtimer

[Thomas Gleixner <tglx@linutronix.de>]

On Fri, 21 Jan 2011, Vernon Mauery wrote:

> In the current hrtimer implementation, when explicitly
> removing a timer, the hrtimer_cancel function waits
> until the timer is not currently running before actually
> calling remove_hrtimer.  But in the __hrtimer_start_range_ns
> function (called by hrtimer_start, among others), it
> calls remove_hrtimer without checking to see if the
> timer is currently running.  This doesn't seem to cause
> much trouble in the vanilla kernel, but in the -RT
> kernel, this causes a silent hang when exercising the
> sch_htb module on a 10GbE interface.
> 
> Signed-off-by: Vernon Mauery <vernux@us.ibm.com>
> ---
>  kernel/hrtimer.c |    5 +++++
>  1 files changed, 5 insertions(+), 0 deletions(-)
> 
> diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c
> index 0c8d7c0..c7e8ba0 100644
> --- a/kernel/hrtimer.c
> +++ b/kernel/hrtimer.c
> @@ -935,6 +935,11 @@ int __hrtimer_start_range_ns(struct hrtimer *timer, ktime_t tim,
>  	base = lock_hrtimer_base(timer, &flags);
>  
>  	/* Remove an active timer from the queue: */
> +	while (hrtimer_callback_running(timer)) {
> +		unlock_hrtimer_base(timer, &flags);
> +		hrtimer_wait_for_timer(timer);
> +		base = lock_hrtimer_base(timer, &flags);
> +	}
>  	ret = remove_hrtimer(timer, base);
>  
>  	/* Switch the timer base, if necessary: */

Hmm, why does it hang ?

Thanks,

	tglx