Messages in this thread | | | Date | Mon, 09 Sep 2013 10:18:02 +0200 | From | Ronald Wahl <> | Subject | Re: RFC: [PATCH] clocksource: tcb: fix min_delta calculation |
| |
On 20.06.2013 17:00, Marc Kleine-Budde wrote: > Hello Thomas, > > On 04/25/2013 04:53 PM, Marc Kleine-Budde wrote: >> On 04/25/2013 04:21 PM, Thomas Gleixner wrote: >>> On Tue, 23 Apr 2013, Marc Kleine-Budde wrote: >>>> The commit >>>> >>>> 77cc982 clocksource: use clockevents_config_and_register() where possible >>>> >>>> switches from manually calculating min_delta_ns (and others) and >>>> clockevents_register_device() to automatic calculation via >>>> clockevents_config_and_register(). During this conversation the "+ 1" in >>>> >>>> min_delta_ns = clockevent_delta2ns(1, &clkevt.clkevt) + 1; >>>> >>>> was lost. This leads to problems with schedule_delayed_work() with a delay of >>>> "1". Resulting in the work not scheduled in time. >>> >>> Errm. How is schedule_delayed_work() related to this?
I also stumbled over this issue now after upgrading from 3.4.57 to 3.10.10. Even sleep() is unreliable - sleeping for 2 seconds often sleeps just more than 3 seconds.
The now lost +1 has actually been added in 2007 to solve a rounding issue:
http://permalink.gmane.org/gmane.linux.kernel/549744
So I think this should either be fixed by always rounding up in clockevent_delta2ns() or by adding +1 to the min_delta_ns in the clockevents_config_and_register() function. Dunno if rounding up will cause problems with the max_delta_ns as it might be too big then so probably the second approach is better.
Opinions?
- ron
-- Ronald Wahl - ronald.wahl@raritan.com - Phone +49 375271349-0 Fax -99 Raritan Deutschland GmbH, Kornmarkt 7, 08056 Zwickau, Germany USt-IdNr. DE813094160, Steuer-Nr. 227/117/01749 Amtsgericht Chemnitz HRB 23605 Geschäftsführung: Stuart Hopper, Ralf Ploenes -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |