RE: [Xen-devel] [PATCH 05/12] xen/pvclock: add monotonicity check

[Dan Magenheimer <dan.magenheimer@oracle.com>]

> Other tsc-based clocksources add a monotonicity test to make 
> sure there's
> no regression in the returned cycles.
> 
> Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
> ---
>  arch/x86/xen/time.c |    7 ++++++-
>  1 files changed, 6 insertions(+), 1 deletions(-)
> 
> diff --git a/arch/x86/xen/time.c b/arch/x86/xen/time.c
> index 0a5aa44..00f06cc 100644
> --- a/arch/x86/xen/time.c
> +++ b/arch/x86/xen/time.c
> @@ -213,9 +213,14 @@ cycle_t xen_clocksource_read(void)
>  	return ret;
>  }
>  
> +static struct clocksource xen_clocksource;
> +
>  static cycle_t xen_clocksource_get_cycles(struct clocksource *cs)
>  {
> -	return xen_clocksource_read();
> +	cycle_t ret = xen_clocksource_read();
> +
> +	return ret >= xen_clocksource.cycle_last ?
> +		ret : xen_clocksource.cycle_last;

As long as we are going through the trouble of making
this monotonic, shouldn't it be monotonically increasing
(rather than just monotonically non-decreasing)?  The
rdtsc instruction and any suitably high-precision
hardware timer will never return the same value
on subsequent uses so this might be a reasonable
precedent to obey.  E.g.

+	return ret > xen_clocksource.cycle_last ?
+		ret : ++xen_clocksource.cycle_last;