Description of problem:

As can be seen in kernel/time/tick-oneshot.c:tick_dev_program_event(), clock_event_device:min_delta_ns, which represents the granularity of the clockevent timer increments, can grow till overflow and can never be reduced.

One possible observable consequence of that is, if it ever overflows, the loop in this function becomes endless, because

      expires = ktime_add_ns(now, dev->min_delta_ns);

gives either negative expires or expires less than now, either of which resulting in error return from clockevents_program_event() which causes the loop to start over.


Version-Release number of selected component (if applicable):

2.6.32-504.3.3.el6.x86_64


How reproducible:

The endless loop on one of the CPUs was seen once in a virtual machine in Parallels Server.  The exact details are still investigated.


Additional info:

These problems have been addressed by the following commits in the mainline kernel:

commit 80a05b9ffa7dc13f6693902dd8999a2b61a3a0d7
Author: Thomas Gleixner <tglx@linutronix.de>
Date:   Fri Mar 12 17:34:14 2010 +0100

    clockevents: Sanitize min_delta_ns adjustment and prevent overflows
    
    The current logic which handles clock events programming failures can
    increase min_delta_ns unlimited and even can cause overflows.
    
    Sanitize it by:
     - prevent zero increase when min_delta_ns == 1
     - limiting min_delta_ns to a jiffie
     - bail out if the jiffie limit is hit
     - add retries stats for /proc/timer_list so we can gather data
    
    Reported-by: Uwe Kleine-Koenig <u.kleine-koenig@pengutronix.de>
    Signed-off-by: Thomas Gleixner <tglx@linutronix.de>


commit d1748302f70be7469809809283fe164156a34231
Author: Martin Schwidefsky <schwidefsky@de.ibm.com>
Date:   Tue Aug 23 15:29:42 2011 +0200

    clockevents: Make minimum delay adjustments configurable
    
    The automatic increase of the min_delta_ns of a clockevents device
    should be done in the clockevents code as the minimum delay is an
    attribute of the clockevents device.
    
    In addition not all architectures want the automatic adjustment, on a
    massively virtualized system it can happen that the programming of a
    clock event fails several times in a row because the virtual cpu has
    been rescheduled quickly enough. In that case the minimum delay will
    erroneously be increased with no way back. The new config symbol
    GENERIC_CLOCKEVENTS_MIN_ADJUST is used to enable the automatic
    adjustment. The config option is selected only for x86.
    
    Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
    Cc: john stultz <johnstul@us.ibm.com>
    Link: http://lkml.kernel.org/r/20110823133142.494157493@de.ibm.com
    Signed-off-by: Thomas Gleixner <tglx@linutronix.de>