time: Fix sign bug in NTP mult overflow warning

In commit 6067dc5 ("time: Avoid possible NTP adjustment
mult overflow") a new check was added to watch for adjustments
that could cause a mult overflow.

Unfortunately the check compares a signed with unsigned value
and ignored the case where the adjustment was negative, which
causes spurious warn-ons on some systems (and seems like it
would result in problematic time adjustments there as well, due
to the early return).

Thus this patch adds a check to make sure the adjustment is
positive before we check for an overflow, and resovles the issue
in my testing.

Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Debugged-by: pang.xunlei <pang.xunlei@linaro.org>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/1416890145-30048-1-git-send-email-john.stultz@linaro.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>

kernel/time/timekeeping.c

@@ -1330,7 +1330,7 @@ static __always_inline void timekeeping_apply_adjustment(struct timekeeper *tk,
 	 *
 	 * XXX - TODO: Doc ntp_error calculation.
 	 */
-	if (tk->tkr.mult + mult_adj < mult_adj) {
+	if ((mult_adj > 0) && (tk->tkr.mult + mult_adj < mult_adj)) {
 		/* NTP adjustment caused clocksource mult overflow */
 		WARN_ON_ONCE(1);
 		return;