Messages in this thread Patch in this message | | | Date | Tue, 22 May 2012 23:31:43 +0200 (CEST) | From | Thomas Gleixner <> | Subject | Re: [RFC][PATCH 00/32] Nohz cpusets v2 (adaptive tickless kernel) |
| |
On Tue, 27 Mar 2012, Peter Zijlstra wrote:
> On Tue, 2012-03-27 at 17:02 +0200, Gilad Ben-Yossef wrote: > > > > In my case, I also had to disable the clocksource watchdog, but only > > because TSC is not stable on my VM. > > This is really not a nohz/cpuset problem. > > No but that thing is annoying, I ran afoul of it too the other day. > > Thomas, would you object to a means of turning that thing off? And if > not, do you have a preference as to what particular means > (sysctl/sysfs/etc..) ?
We have the commandline option "tsc=reliable" already. That disables the stupid watchdog. Handle with care.
To take it a level further, use the patch below :)
Thanks,
tglx
------------> x86: Add tsc=perfect option which enforces sched_clock_stable The sched_clock dance with updating local clocks can be avoided even if the CPU does not have all the magic features. Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 590900c..dc8ecc3 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -109,6 +109,10 @@ static int __init tsc_setup(char *str) { if (!strcmp(str, "reliable")) tsc_clocksource_reliable = 1; + if (!strcmp(str, "perfect")) { + tsc_clocksource_reliable = 1; + sched_clock_stable = 1; + } if (!strncmp(str, "noirqtime", 9)) no_sched_irq_time = 1; return 1;
| |