timekeeping/ntp: Don't align NTP frequency adjustments to ticks
authorMiroslav Lichvar <mlichvar@redhat.com>
Fri, 9 Mar 2018 18:42:47 +0000 (10:42 -0800)
committerIngo Molnar <mingo@kernel.org>
Sat, 10 Mar 2018 08:12:41 +0000 (09:12 +0100)
When the timekeeping multiplier is changed, the NTP error is updated to
correct the clock for the delay between the tick and the update of the
clock. This error is corrected in later updates and the clock appears as
if the frequency was changed exactly on the tick.

Remove this correction to keep the point where the frequency is
effectively changed at the time of the update. This removes a major
source of the NTP error.

Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Stephen Boyd <stephen.boyd@linaro.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1520620971-9567-2-git-send-email-john.stultz@linaro.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/time/timekeeping.c

index cd03317..c1a0ac1 100644 (file)
@@ -1860,8 +1860,6 @@ static __always_inline void timekeeping_apply_adjustment(struct timekeeper *tk,
         *      xtime_nsec_2 = xtime_nsec_1 - offset
         * Which simplfies to:
         *      xtime_nsec -= offset
-        *
-        * XXX - TODO: Doc ntp_error calculation.
         */
        if ((mult_adj > 0) && (tk->tkr_mono.mult + mult_adj < mult_adj)) {
                /* NTP adjustment caused clocksource mult overflow */
@@ -1872,7 +1870,6 @@ static __always_inline void timekeeping_apply_adjustment(struct timekeeper *tk,
        tk->tkr_mono.mult += mult_adj;
        tk->xtime_interval += interval;
        tk->tkr_mono.xtime_nsec -= offset;
-       tk->ntp_error -= (interval - offset) << tk->ntp_error_shift;
 }
 
 /*