Re: [PATCH 13/15] sched: Use a static_key for sched_clock_stable

From: Dave Young
Date: Wed Jan 22 2014 - 20:54:05 EST


On 01/22/14 at 10:08pm, Peter Zijlstra wrote:
> >
> > I think its the right region to look through. My current suspect is the
> > linear continuity fit with the initial 'random' multiplier.
> >
> > That initial 'random' multiplier can get us quite high, and we'll fit
> > the function to match that but continue at a sane rate.
> >
> > I'll try and prod a little more later this evening as time permits.
>
> Does this cure things?

Peter, the odd timstamp still happens with this patch for me.

>
> ---
> arch/x86/kernel/tsc.c | 11 +++++++----
> 1 file changed, 7 insertions(+), 4 deletions(-)
>
> diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
> index a3acbac2ee72..bb04148c5fe0 100644
> --- a/arch/x86/kernel/tsc.c
> +++ b/arch/x86/kernel/tsc.c
> @@ -237,7 +237,7 @@ static inline unsigned long long cycles_2_ns(unsigned long long cyc)
> /* XXX surely we already have this someplace in the kernel?! */
> #define DIV_ROUND(n, d) (((n) + ((d) / 2)) / (d))
>
> -static void set_cyc2ns_scale(unsigned long cpu_khz, int cpu)
> +static void set_cyc2ns_scale(unsigned long cpu_khz, int cpu, bool origin)
> {
> unsigned long long tsc_now, ns_now;
> struct cyc2ns_data *data;
> @@ -252,7 +252,10 @@ static void set_cyc2ns_scale(unsigned long cpu_khz, int cpu)
> data = cyc2ns_write_begin(cpu);
>
> rdtscll(tsc_now);
> - ns_now = cycles_2_ns(tsc_now);
> + if (origin)
> + ns_now = 0;
> + else
> + ns_now = cycles_2_ns(tsc_now);
>
> /*
> * Compute a new multiplier as per the above comment and ensure our
> @@ -926,7 +929,7 @@ static int time_cpufreq_notifier(struct notifier_block *nb, unsigned long val,
> mark_tsc_unstable("cpufreq changes");
> }
>
> - set_cyc2ns_scale(tsc_khz, freq->cpu);
> + set_cyc2ns_scale(tsc_khz, freq->cpu, false);
>
> return 0;
> }
> @@ -1199,7 +1202,7 @@ void __init tsc_init(void)
> */
> for_each_possible_cpu(cpu) {
> cyc2ns_init(cpu);
> - set_cyc2ns_scale(cpu_khz, cpu);
> + set_cyc2ns_scale(cpu_khz, cpu, true);
> }
>
> if (tsc_disabled > 0)
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/